2016-10-22 12:46:35 +03:00
|
|
|
/*
|
|
|
|
* QEMU PowerPC PowerNV machine model
|
|
|
|
*
|
|
|
|
* Copyright (c) 2016, IBM Corporation.
|
|
|
|
*
|
|
|
|
* This library is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU Lesser General Public
|
|
|
|
* License as published by the Free Software Foundation; either
|
2020-10-16 17:53:46 +03:00
|
|
|
* version 2.1 of the License, or (at your option) any later version.
|
2016-10-22 12:46:35 +03:00
|
|
|
*
|
|
|
|
* This library is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
|
* Lesser General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU Lesser General Public
|
|
|
|
* License along with this library; if not, see <http://www.gnu.org/licenses/>.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include "qemu/osdep.h"
|
2020-10-28 14:36:57 +03:00
|
|
|
#include "qemu/datadir.h"
|
2018-06-25 15:41:58 +03:00
|
|
|
#include "qemu/units.h"
|
ppc/pnv: Set default RAM size to 1 GB
The memory layout of the PowerNV machine is defined as :
#define KERNEL_LOAD_BASE ((void *)0x20000000)
#define KERNEL_LOAD_SIZE 0x08000000
#define INITRAMFS_LOAD_BASE KERNEL_LOAD_BASE + KERNEL_LOAD_SIZE
#define INITRAMFS_LOAD_SIZE 0x08000000
#define SKIBOOT_BASE 0x30000000
#define SKIBOOT_SIZE 0x01c10000
#define CPU_STACKS_BASE (SKIBOOT_BASE + SKIBOOT_SIZE)
#define STACK_SHIFT 15
#define STACK_SIZE (1 << STACK_SHIFT)
The overall size of the CPU stacks is (max PIR + 1) * 32K and the
machine easily reaches 800MB of minimum required RAM.
Any value below will result in a skiboot crash :
[ 0.034949905,3] MEM: Partial overlap detected between regions:
[ 0.034959039,3] MEM: ibm,firmware-stacks [0x31c10000-0x3a450000] (new)
[ 0.034968576,3] MEM: ibm,firmware-allocs-memory@0 [0x31c10000-0x38400000]
[ 0.034980367,3] Out of memory adding skiboot reserved areas
[ 0.035074945,3] ***********************************************
[ 0.035093627,3] < assert failed at core/mem_region.c:1129 >
[ 0.035104247,3] .
[ 0.035108025,3] .
[ 0.035111651,3] .
[ 0.035115231,3] OO__)
[ 0.035119198,3] <"__/
[ 0.035122980,3] ^ ^
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20210129111719.790692-1-clg@kaod.org>
Reviewed-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2021-01-29 14:17:19 +03:00
|
|
|
#include "qemu/cutils.h"
|
2016-10-22 12:46:35 +03:00
|
|
|
#include "qapi/error.h"
|
ppc/pnv: Silence missing BMC warning with qtest
The device introspect test in qtest emits some warnings with the
the pnv machine types during the "nodefaults" phase:
TEST check-qtest-ppc64: tests/qtest/device-introspect-test
qemu-system-ppc64: warning: machine has no BMC device. Use '-device
ipmi-bmc-sim,id=bmc0 -device isa-ipmi-bt,bmc=bmc0,irq=10' to define
one
qemu-system-ppc64: warning: machine has no BMC device. Use '-device
ipmi-bmc-sim,id=bmc0 -device isa-ipmi-bt,bmc=bmc0,irq=10' to define
one
qemu-system-ppc64: warning: machine has no BMC device. Use '-device
ipmi-bmc-sim,id=bmc0 -device isa-ipmi-bt,bmc=bmc0,irq=10' to define
one
This is expected since the pnv machine doesn't create the internal
BMC simulator fallback when "-nodefaults" is passed on the command
line, but these warnings appear in ci logs and confuse people.
Not having a BMC isn't recommended but it is still a supported
configuration, so a straightforward fix is to just silent this
warning when qtest is enabled.
Fixes: 25f3170b0654 ("ppc/pnv: Create BMC devices only when defaults are enabled")
Reported-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Greg Kurz <groug@kaod.org>
Message-Id: <159280903824.485572.831378159272329707.stgit@bahia.lan>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2020-06-22 09:57:18 +03:00
|
|
|
#include "sysemu/qtest.h"
|
2016-10-22 12:46:35 +03:00
|
|
|
#include "sysemu/sysemu.h"
|
|
|
|
#include "sysemu/numa.h"
|
2019-08-12 08:23:38 +03:00
|
|
|
#include "sysemu/reset.h"
|
2019-08-12 08:23:59 +03:00
|
|
|
#include "sysemu/runstate.h"
|
2017-03-03 14:01:16 +03:00
|
|
|
#include "sysemu/cpus.h"
|
2019-06-06 20:47:32 +03:00
|
|
|
#include "sysemu/device_tree.h"
|
2020-03-25 17:41:44 +03:00
|
|
|
#include "sysemu/hw_accel.h"
|
2016-10-11 09:56:52 +03:00
|
|
|
#include "target/ppc/cpu.h"
|
2016-10-22 12:46:35 +03:00
|
|
|
#include "hw/ppc/fdt.h"
|
|
|
|
#include "hw/ppc/ppc.h"
|
|
|
|
#include "hw/ppc/pnv.h"
|
2016-10-22 12:46:39 +03:00
|
|
|
#include "hw/ppc/pnv_core.h"
|
2016-10-22 12:46:35 +03:00
|
|
|
#include "hw/loader.h"
|
2020-03-25 17:41:44 +03:00
|
|
|
#include "hw/nmi.h"
|
2016-10-22 12:46:36 +03:00
|
|
|
#include "qapi/visitor.h"
|
2017-04-03 10:46:02 +03:00
|
|
|
#include "monitor/monitor.h"
|
|
|
|
#include "hw/intc/intc.h"
|
2017-04-11 18:30:05 +03:00
|
|
|
#include "hw/ipmi/ipmi.h"
|
2018-03-23 06:11:07 +03:00
|
|
|
#include "target/ppc/mmu-hash64.h"
|
ppc/pnv: Add models for POWER9 PHB4 PCIe Host bridge
These changes introduces models for the PCIe Host Bridge (PHB4) of the
POWER9 processor. It includes the PowerBus logic interface (PBCQ),
IOMMU support, a single PCIe Gen.4 Root Complex, and support for MSI
and LSI interrupt sources as found on a POWER9 system using the XIVE
interrupt controller.
POWER9 processor comes with 3 PHB4 PEC (PCI Express Controller) and
each PEC can have several PHBs. By default,
* PEC0 provides 1 PHB (PHB0)
* PEC1 provides 2 PHBs (PHB1 and PHB2)
* PEC2 provides 3 PHBs (PHB3, PHB4 and PHB5)
Each PEC has a set "global" registers and some "per-stack" (per-PHB)
registers. Those are organized in two XSCOM ranges, the "Nest" range
and the "PCI" range, each range contains both some "PEC" registers and
some "per-stack" registers.
No default device layout is provided and PCI devices can be added on
any of the available PCIe Root Port (pcie.0 .. 2 of a Power9 chip)
with address 0x0 as the firwware (skiboot) only accepts a single
device per root port. To run a simple system with a network and a
storage adapters, use a command line options such as :
-device e1000e,netdev=net0,mac=C0:FF:EE:00:00:02,bus=pcie.0,addr=0x0
-netdev bridge,id=net0,helper=/usr/libexec/qemu-bridge-helper,br=virbr0,id=hostnet0
-device megasas,id=scsi0,bus=pcie.1,addr=0x0
-drive file=$disk,if=none,id=drive-scsi0-0-0-0,format=qcow2,cache=none
-device scsi-hd,bus=scsi0.0,channel=0,scsi-id=0,lun=0,drive=drive-scsi0-0-0-0,id=scsi0-0-0-0,bootindex=2
If more are needed, include a bridge.
Multi chip is supported, each chip adding its set of PHB4 controllers
and its PCI busses. The model doesn't emulate the EEH error handling.
This model is not ready for hotplug yet.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[ clg: - numerous cleanups
- commit log
- fix for broken LSI support
- PHB pic printinfo
- large QOM rework ]
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20200127144506.11132-2-clg@kaod.org>
[dwg: Use device_class_set_props()]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2020-01-27 17:45:05 +03:00
|
|
|
#include "hw/pci/msi.h"
|
2022-06-24 11:49:12 +03:00
|
|
|
#include "hw/pci-host/pnv_phb.h"
|
2016-10-22 12:46:35 +03:00
|
|
|
|
2017-04-03 10:46:01 +03:00
|
|
|
#include "hw/ppc/xics.h"
|
2019-08-12 08:23:51 +03:00
|
|
|
#include "hw/qdev-properties.h"
|
ppc/pnv: add XSCOM infrastructure
On a real POWER8 system, the Pervasive Interconnect Bus (PIB) serves
as a backbone to connect different units of the system. The host
firmware connects to the PIB through a bridge unit, the
Alter-Display-Unit (ADU), which gives him access to all the chiplets
on the PCB network (Pervasive Connect Bus), the PIB acting as the root
of this network.
XSCOM (serial communication) is the interface to the sideband bus
provided by the POWER8 pervasive unit to read and write to chiplets
resources. This is needed by the host firmware, OPAL and to a lesser
extent, Linux. This is among others how the PCI Host bridges get
configured at boot or how the LPC bus is accessed.
To represent the ADU of a real system, we introduce a specific
AddressSpace to dispatch XSCOM accesses to the targeted chiplets. The
translation of an XSCOM address into a PCB register address is
slightly different between the P9 and the P8. This is handled before
the dispatch using a 8byte alignment for all.
To customize the device tree, a QOM InterfaceClass, PnvXScomInterface,
is provided with a populate() handler. The chip populates the device
tree by simply looping on its children. Therefore, each model needing
custom nodes should not forget to declare itself as a child at
instantiation time.
Based on previous work done by :
Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
[dwg: Added cpu parameter to xscom_complete()]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2016-10-22 12:46:40 +03:00
|
|
|
#include "hw/ppc/pnv_xscom.h"
|
2019-10-21 16:12:11 +03:00
|
|
|
#include "hw/ppc/pnv_pnor.h"
|
ppc/pnv: add XSCOM infrastructure
On a real POWER8 system, the Pervasive Interconnect Bus (PIB) serves
as a backbone to connect different units of the system. The host
firmware connects to the PIB through a bridge unit, the
Alter-Display-Unit (ADU), which gives him access to all the chiplets
on the PCB network (Pervasive Connect Bus), the PIB acting as the root
of this network.
XSCOM (serial communication) is the interface to the sideband bus
provided by the POWER8 pervasive unit to read and write to chiplets
resources. This is needed by the host firmware, OPAL and to a lesser
extent, Linux. This is among others how the PCI Host bridges get
configured at boot or how the LPC bus is accessed.
To represent the ADU of a real system, we introduce a specific
AddressSpace to dispatch XSCOM accesses to the targeted chiplets. The
translation of an XSCOM address into a PCB register address is
slightly different between the P9 and the P8. This is handled before
the dispatch using a 8byte alignment for all.
To customize the device tree, a QOM InterfaceClass, PnvXScomInterface,
is provided with a populate() handler. The chip populates the device
tree by simply looping on its children. Therefore, each model needing
custom nodes should not forget to declare itself as a child at
instantiation time.
Based on previous work done by :
Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
[dwg: Added cpu parameter to xscom_complete()]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2016-10-22 12:46:40 +03:00
|
|
|
|
2016-10-22 12:46:43 +03:00
|
|
|
#include "hw/isa/isa.h"
|
|
|
|
#include "hw/char/serial.h"
|
2019-10-04 02:03:53 +03:00
|
|
|
#include "hw/rtc/mc146818rtc.h"
|
2016-10-22 12:46:43 +03:00
|
|
|
|
2016-10-22 12:46:35 +03:00
|
|
|
#include <libfdt.h>
|
|
|
|
|
2019-02-25 20:01:55 +03:00
|
|
|
#define FDT_MAX_SIZE (1 * MiB)
|
2016-10-22 12:46:35 +03:00
|
|
|
|
|
|
|
#define FW_FILE_NAME "skiboot.lid"
|
|
|
|
#define FW_LOAD_ADDR 0x0
|
2020-10-02 12:14:40 +03:00
|
|
|
#define FW_MAX_SIZE (16 * MiB)
|
2016-10-22 12:46:35 +03:00
|
|
|
|
|
|
|
#define KERNEL_LOAD_ADDR 0x20000000
|
2021-01-26 20:10:55 +03:00
|
|
|
#define KERNEL_MAX_SIZE (128 * MiB)
|
|
|
|
#define INITRD_LOAD_ADDR 0x28000000
|
|
|
|
#define INITRD_MAX_SIZE (128 * MiB)
|
2016-10-22 12:46:35 +03:00
|
|
|
|
2017-10-09 22:51:10 +03:00
|
|
|
static const char *pnv_chip_core_typename(const PnvChip *o)
|
|
|
|
{
|
|
|
|
const char *chip_type = object_class_get_name(object_get_class(OBJECT(o)));
|
|
|
|
int len = strlen(chip_type) - strlen(PNV_CHIP_TYPE_SUFFIX);
|
|
|
|
char *s = g_strdup_printf(PNV_CORE_TYPE_NAME("%.*s"), len, chip_type);
|
|
|
|
const char *core_type = object_class_get_name(object_class_by_name(s));
|
|
|
|
g_free(s);
|
|
|
|
return core_type;
|
|
|
|
}
|
|
|
|
|
2016-10-22 12:46:35 +03:00
|
|
|
/*
|
|
|
|
* On Power Systems E880 (POWER8), the max cpus (threads) should be :
|
|
|
|
* 4 * 4 sockets * 12 cores * 8 threads = 1536
|
|
|
|
* Let's make it 2^11
|
|
|
|
*/
|
|
|
|
#define MAX_CPUS 2048
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Memory nodes are created by hostboot, one for each range of memory
|
|
|
|
* that has a different "affinity". In practice, it means one range
|
|
|
|
* per chip.
|
|
|
|
*/
|
2017-12-15 16:56:01 +03:00
|
|
|
static void pnv_dt_memory(void *fdt, int chip_id, hwaddr start, hwaddr size)
|
2016-10-22 12:46:35 +03:00
|
|
|
{
|
|
|
|
char *mem_name;
|
|
|
|
uint64_t mem_reg_property[2];
|
|
|
|
int off;
|
|
|
|
|
|
|
|
mem_reg_property[0] = cpu_to_be64(start);
|
|
|
|
mem_reg_property[1] = cpu_to_be64(size);
|
|
|
|
|
|
|
|
mem_name = g_strdup_printf("memory@%"HWADDR_PRIx, start);
|
|
|
|
off = fdt_add_subnode(fdt, 0, mem_name);
|
|
|
|
g_free(mem_name);
|
|
|
|
|
|
|
|
_FDT((fdt_setprop_string(fdt, off, "device_type", "memory")));
|
|
|
|
_FDT((fdt_setprop(fdt, off, "reg", mem_reg_property,
|
|
|
|
sizeof(mem_reg_property))));
|
|
|
|
_FDT((fdt_setprop_cell(fdt, off, "ibm,chip-id", chip_id)));
|
|
|
|
}
|
|
|
|
|
2016-10-22 12:46:39 +03:00
|
|
|
static int get_cpus_node(void *fdt)
|
|
|
|
{
|
|
|
|
int cpus_offset = fdt_path_offset(fdt, "/cpus");
|
|
|
|
|
|
|
|
if (cpus_offset < 0) {
|
2017-10-03 17:13:11 +03:00
|
|
|
cpus_offset = fdt_add_subnode(fdt, 0, "cpus");
|
2016-10-22 12:46:39 +03:00
|
|
|
if (cpus_offset) {
|
|
|
|
_FDT((fdt_setprop_cell(fdt, cpus_offset, "#address-cells", 0x1)));
|
|
|
|
_FDT((fdt_setprop_cell(fdt, cpus_offset, "#size-cells", 0x0)));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
_FDT(cpus_offset);
|
|
|
|
return cpus_offset;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The PowerNV cores (and threads) need to use real HW ids and not an
|
|
|
|
* incremental index like it has been done on other platforms. This HW
|
|
|
|
* id is stored in the CPU PIR, it is used to create cpu nodes in the
|
|
|
|
* device tree, used in XSCOM to address cores and in interrupt
|
|
|
|
* servers.
|
|
|
|
*/
|
2017-12-15 16:56:01 +03:00
|
|
|
static void pnv_dt_core(PnvChip *chip, PnvCore *pc, void *fdt)
|
2016-10-22 12:46:39 +03:00
|
|
|
{
|
2018-06-13 04:57:37 +03:00
|
|
|
PowerPCCPU *cpu = pc->threads[0];
|
|
|
|
CPUState *cs = CPU(cpu);
|
2016-10-22 12:46:39 +03:00
|
|
|
DeviceClass *dc = DEVICE_GET_CLASS(cs);
|
2016-11-01 02:25:29 +03:00
|
|
|
int smt_threads = CPU_CORE(pc)->nr_threads;
|
2016-10-22 12:46:39 +03:00
|
|
|
CPUPPCState *env = &cpu->env;
|
|
|
|
PowerPCCPUClass *pcc = POWERPC_CPU_GET_CLASS(cs);
|
2022-08-19 18:39:26 +03:00
|
|
|
g_autofree uint32_t *servers_prop = g_new(uint32_t, smt_threads);
|
2016-10-22 12:46:39 +03:00
|
|
|
int i;
|
|
|
|
uint32_t segs[] = {cpu_to_be32(28), cpu_to_be32(40),
|
|
|
|
0xffffffff, 0xffffffff};
|
|
|
|
uint32_t tbfreq = PNV_TIMEBASE_FREQ;
|
|
|
|
uint32_t cpufreq = 1000000000;
|
|
|
|
uint32_t page_sizes_prop[64];
|
|
|
|
size_t page_sizes_prop_size;
|
|
|
|
const uint8_t pa_features[] = { 24, 0,
|
|
|
|
0xf6, 0x3f, 0xc7, 0xc0, 0x80, 0xf0,
|
|
|
|
0x80, 0x00, 0x00, 0x00, 0x00, 0x00,
|
|
|
|
0x00, 0x00, 0x00, 0x00, 0x80, 0x00,
|
|
|
|
0x80, 0x00, 0x80, 0x00, 0x80, 0x00 };
|
|
|
|
int offset;
|
|
|
|
char *nodename;
|
|
|
|
int cpus_offset = get_cpus_node(fdt);
|
|
|
|
|
|
|
|
nodename = g_strdup_printf("%s@%x", dc->fw_name, pc->pir);
|
|
|
|
offset = fdt_add_subnode(fdt, cpus_offset, nodename);
|
|
|
|
_FDT(offset);
|
|
|
|
g_free(nodename);
|
|
|
|
|
|
|
|
_FDT((fdt_setprop_cell(fdt, offset, "ibm,chip-id", chip->chip_id)));
|
|
|
|
|
|
|
|
_FDT((fdt_setprop_cell(fdt, offset, "reg", pc->pir)));
|
|
|
|
_FDT((fdt_setprop_cell(fdt, offset, "ibm,pir", pc->pir)));
|
|
|
|
_FDT((fdt_setprop_string(fdt, offset, "device_type", "cpu")));
|
|
|
|
|
|
|
|
_FDT((fdt_setprop_cell(fdt, offset, "cpu-version", env->spr[SPR_PVR])));
|
|
|
|
_FDT((fdt_setprop_cell(fdt, offset, "d-cache-block-size",
|
|
|
|
env->dcache_line_size)));
|
|
|
|
_FDT((fdt_setprop_cell(fdt, offset, "d-cache-line-size",
|
|
|
|
env->dcache_line_size)));
|
|
|
|
_FDT((fdt_setprop_cell(fdt, offset, "i-cache-block-size",
|
|
|
|
env->icache_line_size)));
|
|
|
|
_FDT((fdt_setprop_cell(fdt, offset, "i-cache-line-size",
|
|
|
|
env->icache_line_size)));
|
|
|
|
|
|
|
|
if (pcc->l1_dcache_size) {
|
|
|
|
_FDT((fdt_setprop_cell(fdt, offset, "d-cache-size",
|
|
|
|
pcc->l1_dcache_size)));
|
|
|
|
} else {
|
2017-07-12 16:57:41 +03:00
|
|
|
warn_report("Unknown L1 dcache size for cpu");
|
2016-10-22 12:46:39 +03:00
|
|
|
}
|
|
|
|
if (pcc->l1_icache_size) {
|
|
|
|
_FDT((fdt_setprop_cell(fdt, offset, "i-cache-size",
|
|
|
|
pcc->l1_icache_size)));
|
|
|
|
} else {
|
2017-07-12 16:57:41 +03:00
|
|
|
warn_report("Unknown L1 icache size for cpu");
|
2016-10-22 12:46:39 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
_FDT((fdt_setprop_cell(fdt, offset, "timebase-frequency", tbfreq)));
|
|
|
|
_FDT((fdt_setprop_cell(fdt, offset, "clock-frequency", cpufreq)));
|
2019-09-11 17:29:25 +03:00
|
|
|
_FDT((fdt_setprop_cell(fdt, offset, "ibm,slb-size",
|
|
|
|
cpu->hash64_opts->slb_size)));
|
2016-10-22 12:46:39 +03:00
|
|
|
_FDT((fdt_setprop_string(fdt, offset, "status", "okay")));
|
|
|
|
_FDT((fdt_setprop(fdt, offset, "64-bit", NULL, 0)));
|
|
|
|
|
2021-05-07 19:41:46 +03:00
|
|
|
if (ppc_has_spr(cpu, SPR_PURR)) {
|
2016-10-22 12:46:39 +03:00
|
|
|
_FDT((fdt_setprop(fdt, offset, "ibm,purr", NULL, 0)));
|
|
|
|
}
|
|
|
|
|
2018-03-23 06:11:07 +03:00
|
|
|
if (ppc_hash64_has(cpu, PPC_HASH64_1TSEG)) {
|
2016-10-22 12:46:39 +03:00
|
|
|
_FDT((fdt_setprop(fdt, offset, "ibm,processor-segment-sizes",
|
|
|
|
segs, sizeof(segs))));
|
|
|
|
}
|
|
|
|
|
2019-09-11 17:29:25 +03:00
|
|
|
/*
|
|
|
|
* Advertise VMX/VSX (vector extensions) if available
|
2016-10-22 12:46:39 +03:00
|
|
|
* 0 / no property == no vector extensions
|
|
|
|
* 1 == VMX / Altivec available
|
2019-09-11 17:29:25 +03:00
|
|
|
* 2 == VSX available
|
|
|
|
*/
|
2016-10-22 12:46:39 +03:00
|
|
|
if (env->insns_flags & PPC_ALTIVEC) {
|
|
|
|
uint32_t vmx = (env->insns_flags2 & PPC2_VSX) ? 2 : 1;
|
|
|
|
|
|
|
|
_FDT((fdt_setprop_cell(fdt, offset, "ibm,vmx", vmx)));
|
|
|
|
}
|
|
|
|
|
2019-09-11 17:29:25 +03:00
|
|
|
/*
|
|
|
|
* Advertise DFP (Decimal Floating Point) if available
|
2016-10-22 12:46:39 +03:00
|
|
|
* 0 / no property == no DFP
|
2019-09-11 17:29:25 +03:00
|
|
|
* 1 == DFP available
|
|
|
|
*/
|
2016-10-22 12:46:39 +03:00
|
|
|
if (env->insns_flags2 & PPC2_DFP) {
|
|
|
|
_FDT((fdt_setprop_cell(fdt, offset, "ibm,dfp", 1)));
|
|
|
|
}
|
|
|
|
|
2018-03-22 08:18:40 +03:00
|
|
|
page_sizes_prop_size = ppc_create_page_sizes_prop(cpu, page_sizes_prop,
|
|
|
|
sizeof(page_sizes_prop));
|
2016-10-22 12:46:39 +03:00
|
|
|
if (page_sizes_prop_size) {
|
|
|
|
_FDT((fdt_setprop(fdt, offset, "ibm,segment-page-sizes",
|
|
|
|
page_sizes_prop, page_sizes_prop_size)));
|
|
|
|
}
|
|
|
|
|
|
|
|
_FDT((fdt_setprop(fdt, offset, "ibm,pa-features",
|
|
|
|
pa_features, sizeof(pa_features))));
|
|
|
|
|
|
|
|
/* Build interrupt servers properties */
|
|
|
|
for (i = 0; i < smt_threads; i++) {
|
|
|
|
servers_prop[i] = cpu_to_be32(pc->pir + i);
|
|
|
|
}
|
|
|
|
_FDT((fdt_setprop(fdt, offset, "ibm,ppc-interrupt-server#s",
|
2022-08-19 18:39:26 +03:00
|
|
|
servers_prop, sizeof(*servers_prop) * smt_threads)));
|
2016-10-22 12:46:39 +03:00
|
|
|
}
|
|
|
|
|
2017-12-15 16:56:01 +03:00
|
|
|
static void pnv_dt_icp(PnvChip *chip, void *fdt, uint32_t pir,
|
|
|
|
uint32_t nr_threads)
|
2017-04-03 10:46:05 +03:00
|
|
|
{
|
|
|
|
uint64_t addr = PNV_ICP_BASE(chip) | (pir << 12);
|
|
|
|
char *name;
|
|
|
|
const char compat[] = "IBM,power8-icp\0IBM,ppc-xicp";
|
|
|
|
uint32_t irange[2], i, rsize;
|
|
|
|
uint64_t *reg;
|
|
|
|
int offset;
|
|
|
|
|
|
|
|
irange[0] = cpu_to_be32(pir);
|
|
|
|
irange[1] = cpu_to_be32(nr_threads);
|
|
|
|
|
|
|
|
rsize = sizeof(uint64_t) * 2 * nr_threads;
|
|
|
|
reg = g_malloc(rsize);
|
|
|
|
for (i = 0; i < nr_threads; i++) {
|
|
|
|
reg[i * 2] = cpu_to_be64(addr | ((pir + i) * 0x1000));
|
|
|
|
reg[i * 2 + 1] = cpu_to_be64(0x1000);
|
|
|
|
}
|
|
|
|
|
|
|
|
name = g_strdup_printf("interrupt-controller@%"PRIX64, addr);
|
|
|
|
offset = fdt_add_subnode(fdt, 0, name);
|
|
|
|
_FDT(offset);
|
|
|
|
g_free(name);
|
|
|
|
|
|
|
|
_FDT((fdt_setprop(fdt, offset, "compatible", compat, sizeof(compat))));
|
|
|
|
_FDT((fdt_setprop(fdt, offset, "reg", reg, rsize)));
|
|
|
|
_FDT((fdt_setprop_string(fdt, offset, "device_type",
|
|
|
|
"PowerPC-External-Interrupt-Presentation")));
|
|
|
|
_FDT((fdt_setprop(fdt, offset, "interrupt-controller", NULL, 0)));
|
|
|
|
_FDT((fdt_setprop(fdt, offset, "ibm,interrupt-server-ranges",
|
|
|
|
irange, sizeof(irange))));
|
|
|
|
_FDT((fdt_setprop_cell(fdt, offset, "#interrupt-cells", 1)));
|
|
|
|
_FDT((fdt_setprop_cell(fdt, offset, "#address-cells", 0)));
|
|
|
|
g_free(reg);
|
|
|
|
}
|
|
|
|
|
2022-08-11 19:39:46 +03:00
|
|
|
static PnvPhb4PecState *pnv_phb4_get_pec(PnvChip *chip, PnvPHB4 *phb,
|
|
|
|
Error **errp)
|
|
|
|
{
|
2022-08-11 19:39:48 +03:00
|
|
|
PnvPHB *phb_base = phb->phb_base;
|
|
|
|
PnvPhb4PecState *pecs = NULL;
|
2022-08-11 19:39:46 +03:00
|
|
|
int chip_id = phb->chip_id;
|
|
|
|
int index = phb->phb_id;
|
|
|
|
int i, j;
|
|
|
|
|
2022-08-11 19:39:48 +03:00
|
|
|
if (phb_base->version == 4) {
|
|
|
|
Pnv9Chip *chip9 = PNV9_CHIP(chip);
|
|
|
|
|
|
|
|
pecs = chip9->pecs;
|
|
|
|
} else if (phb_base->version == 5) {
|
|
|
|
Pnv10Chip *chip10 = PNV10_CHIP(chip);
|
|
|
|
|
|
|
|
pecs = chip10->pecs;
|
|
|
|
} else {
|
|
|
|
g_assert_not_reached();
|
|
|
|
}
|
|
|
|
|
2022-08-11 19:39:46 +03:00
|
|
|
for (i = 0; i < chip->num_pecs; i++) {
|
|
|
|
/*
|
|
|
|
* For each PEC, check the amount of phbs it supports
|
|
|
|
* and see if the given phb4 index matches an index.
|
|
|
|
*/
|
2022-08-11 19:39:48 +03:00
|
|
|
PnvPhb4PecState *pec = &pecs[i];
|
2022-08-11 19:39:46 +03:00
|
|
|
|
|
|
|
for (j = 0; j < pec->num_phbs; j++) {
|
|
|
|
if (index == pnv_phb4_pec_get_phb_id(pec, j)) {
|
|
|
|
return pec;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
error_setg(errp,
|
|
|
|
"pnv-phb4 chip-id %d index %d didn't match any existing PEC",
|
|
|
|
chip_id, index);
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
2022-08-11 19:39:43 +03:00
|
|
|
/*
|
|
|
|
* Adds a PnvPHB to the chip. Returns the parent obj of the
|
|
|
|
* PHB which varies with each version (phb version 3 is parented
|
|
|
|
* by the chip, version 4 and 5 are parented by the PEC
|
|
|
|
* device).
|
|
|
|
*
|
|
|
|
* TODO: for version 3 we're still parenting the PHB with the
|
|
|
|
* chip. We should parent with a (so far not implemented)
|
|
|
|
* PHB3 PEC device.
|
|
|
|
*/
|
|
|
|
Object *pnv_chip_add_phb(PnvChip *chip, PnvPHB *phb, Error **errp)
|
|
|
|
{
|
|
|
|
if (phb->version == 3) {
|
2022-08-11 19:39:44 +03:00
|
|
|
Pnv8Chip *chip8 = PNV8_CHIP(chip);
|
|
|
|
|
|
|
|
phb->chip = chip;
|
|
|
|
|
|
|
|
chip8->phbs[chip8->num_phbs] = phb;
|
|
|
|
chip8->num_phbs++;
|
|
|
|
|
2022-08-11 19:39:43 +03:00
|
|
|
return OBJECT(chip);
|
|
|
|
}
|
2022-08-11 19:39:46 +03:00
|
|
|
|
|
|
|
phb->pec = pnv_phb4_get_pec(chip, PNV_PHB4(phb->backend), errp);
|
|
|
|
|
|
|
|
return OBJECT(phb->pec);
|
2022-08-11 19:39:43 +03:00
|
|
|
}
|
|
|
|
|
2019-03-06 11:50:12 +03:00
|
|
|
static void pnv_chip_power8_dt_populate(PnvChip *chip, void *fdt)
|
2016-10-22 12:46:36 +03:00
|
|
|
{
|
2019-12-13 15:00:24 +03:00
|
|
|
static const char compat[] = "ibm,power8-xscom\0ibm,xscom";
|
2016-10-22 12:46:39 +03:00
|
|
|
int i;
|
|
|
|
|
2019-12-13 15:00:18 +03:00
|
|
|
pnv_dt_xscom(chip, fdt, 0,
|
|
|
|
cpu_to_be64(PNV_XSCOM_BASE(chip)),
|
2019-12-13 15:00:24 +03:00
|
|
|
cpu_to_be64(PNV_XSCOM_SIZE),
|
|
|
|
compat, sizeof(compat));
|
ppc/pnv: add XSCOM infrastructure
On a real POWER8 system, the Pervasive Interconnect Bus (PIB) serves
as a backbone to connect different units of the system. The host
firmware connects to the PIB through a bridge unit, the
Alter-Display-Unit (ADU), which gives him access to all the chiplets
on the PCB network (Pervasive Connect Bus), the PIB acting as the root
of this network.
XSCOM (serial communication) is the interface to the sideband bus
provided by the POWER8 pervasive unit to read and write to chiplets
resources. This is needed by the host firmware, OPAL and to a lesser
extent, Linux. This is among others how the PCI Host bridges get
configured at boot or how the LPC bus is accessed.
To represent the ADU of a real system, we introduce a specific
AddressSpace to dispatch XSCOM accesses to the targeted chiplets. The
translation of an XSCOM address into a PCB register address is
slightly different between the P9 and the P8. This is handled before
the dispatch using a 8byte alignment for all.
To customize the device tree, a QOM InterfaceClass, PnvXScomInterface,
is provided with a populate() handler. The chip populates the device
tree by simply looping on its children. Therefore, each model needing
custom nodes should not forget to declare itself as a child at
instantiation time.
Based on previous work done by :
Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
[dwg: Added cpu parameter to xscom_complete()]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2016-10-22 12:46:40 +03:00
|
|
|
|
2016-10-22 12:46:39 +03:00
|
|
|
for (i = 0; i < chip->nr_cores; i++) {
|
2019-11-25 09:58:03 +03:00
|
|
|
PnvCore *pnv_core = chip->cores[i];
|
2016-10-22 12:46:39 +03:00
|
|
|
|
2017-12-15 16:56:01 +03:00
|
|
|
pnv_dt_core(chip, pnv_core, fdt);
|
2017-04-03 10:46:05 +03:00
|
|
|
|
|
|
|
/* Interrupt Control Presenters (ICP). One per core. */
|
2017-12-15 16:56:01 +03:00
|
|
|
pnv_dt_icp(chip, fdt, pnv_core->pir, CPU_CORE(pnv_core)->nr_threads);
|
2016-10-22 12:46:39 +03:00
|
|
|
}
|
|
|
|
|
2016-10-22 12:46:36 +03:00
|
|
|
if (chip->ram_size) {
|
2017-12-15 16:56:01 +03:00
|
|
|
pnv_dt_memory(fdt, chip->chip_id, chip->ram_start, chip->ram_size);
|
2016-10-22 12:46:36 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2019-03-06 11:50:12 +03:00
|
|
|
static void pnv_chip_power9_dt_populate(PnvChip *chip, void *fdt)
|
|
|
|
{
|
2019-12-13 15:00:24 +03:00
|
|
|
static const char compat[] = "ibm,power9-xscom\0ibm,xscom";
|
2019-03-06 11:50:12 +03:00
|
|
|
int i;
|
|
|
|
|
2019-12-13 15:00:18 +03:00
|
|
|
pnv_dt_xscom(chip, fdt, 0,
|
|
|
|
cpu_to_be64(PNV9_XSCOM_BASE(chip)),
|
2019-12-13 15:00:24 +03:00
|
|
|
cpu_to_be64(PNV9_XSCOM_SIZE),
|
|
|
|
compat, sizeof(compat));
|
2019-03-06 11:50:12 +03:00
|
|
|
|
|
|
|
for (i = 0; i < chip->nr_cores; i++) {
|
2019-11-25 09:58:03 +03:00
|
|
|
PnvCore *pnv_core = chip->cores[i];
|
2019-03-06 11:50:12 +03:00
|
|
|
|
|
|
|
pnv_dt_core(chip, pnv_core, fdt);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (chip->ram_size) {
|
|
|
|
pnv_dt_memory(fdt, chip->chip_id, chip->ram_start, chip->ram_size);
|
|
|
|
}
|
2019-03-08 01:35:39 +03:00
|
|
|
|
2019-12-05 21:44:54 +03:00
|
|
|
pnv_dt_lpc(chip, fdt, 0, PNV9_LPCM_BASE(chip), PNV9_LPCM_SIZE);
|
2019-03-06 11:50:12 +03:00
|
|
|
}
|
|
|
|
|
2019-12-05 21:44:51 +03:00
|
|
|
static void pnv_chip_power10_dt_populate(PnvChip *chip, void *fdt)
|
|
|
|
{
|
2019-12-13 15:00:24 +03:00
|
|
|
static const char compat[] = "ibm,power10-xscom\0ibm,xscom";
|
2019-12-05 21:44:51 +03:00
|
|
|
int i;
|
|
|
|
|
2019-12-13 15:00:18 +03:00
|
|
|
pnv_dt_xscom(chip, fdt, 0,
|
|
|
|
cpu_to_be64(PNV10_XSCOM_BASE(chip)),
|
2019-12-13 15:00:24 +03:00
|
|
|
cpu_to_be64(PNV10_XSCOM_SIZE),
|
|
|
|
compat, sizeof(compat));
|
2019-12-05 21:44:51 +03:00
|
|
|
|
|
|
|
for (i = 0; i < chip->nr_cores; i++) {
|
|
|
|
PnvCore *pnv_core = chip->cores[i];
|
|
|
|
|
|
|
|
pnv_dt_core(chip, pnv_core, fdt);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (chip->ram_size) {
|
|
|
|
pnv_dt_memory(fdt, chip->chip_id, chip->ram_start, chip->ram_size);
|
|
|
|
}
|
2019-12-05 21:44:54 +03:00
|
|
|
|
|
|
|
pnv_dt_lpc(chip, fdt, 0, PNV10_LPCM_BASE(chip), PNV10_LPCM_SIZE);
|
2019-12-05 21:44:51 +03:00
|
|
|
}
|
|
|
|
|
2017-12-15 16:56:01 +03:00
|
|
|
static void pnv_dt_rtc(ISADevice *d, void *fdt, int lpc_off)
|
2017-04-11 18:30:02 +03:00
|
|
|
{
|
|
|
|
uint32_t io_base = d->ioport_id;
|
|
|
|
uint32_t io_regs[] = {
|
|
|
|
cpu_to_be32(1),
|
|
|
|
cpu_to_be32(io_base),
|
|
|
|
cpu_to_be32(2)
|
|
|
|
};
|
|
|
|
char *name;
|
|
|
|
int node;
|
|
|
|
|
|
|
|
name = g_strdup_printf("%s@i%x", qdev_fw_name(DEVICE(d)), io_base);
|
|
|
|
node = fdt_add_subnode(fdt, lpc_off, name);
|
|
|
|
_FDT(node);
|
|
|
|
g_free(name);
|
|
|
|
|
|
|
|
_FDT((fdt_setprop(fdt, node, "reg", io_regs, sizeof(io_regs))));
|
|
|
|
_FDT((fdt_setprop_string(fdt, node, "compatible", "pnpPNP,b00")));
|
|
|
|
}
|
|
|
|
|
2017-12-15 16:56:01 +03:00
|
|
|
static void pnv_dt_serial(ISADevice *d, void *fdt, int lpc_off)
|
2017-04-11 18:30:03 +03:00
|
|
|
{
|
|
|
|
const char compatible[] = "ns16550\0pnpPNP,501";
|
|
|
|
uint32_t io_base = d->ioport_id;
|
|
|
|
uint32_t io_regs[] = {
|
|
|
|
cpu_to_be32(1),
|
|
|
|
cpu_to_be32(io_base),
|
|
|
|
cpu_to_be32(8)
|
|
|
|
};
|
2022-03-02 08:51:36 +03:00
|
|
|
uint32_t irq;
|
2017-04-11 18:30:03 +03:00
|
|
|
char *name;
|
|
|
|
int node;
|
|
|
|
|
2022-03-02 08:51:36 +03:00
|
|
|
irq = object_property_get_uint(OBJECT(d), "irq", &error_fatal);
|
|
|
|
|
2017-04-11 18:30:03 +03:00
|
|
|
name = g_strdup_printf("%s@i%x", qdev_fw_name(DEVICE(d)), io_base);
|
|
|
|
node = fdt_add_subnode(fdt, lpc_off, name);
|
|
|
|
_FDT(node);
|
|
|
|
g_free(name);
|
|
|
|
|
|
|
|
_FDT((fdt_setprop(fdt, node, "reg", io_regs, sizeof(io_regs))));
|
|
|
|
_FDT((fdt_setprop(fdt, node, "compatible", compatible,
|
|
|
|
sizeof(compatible))));
|
|
|
|
|
|
|
|
_FDT((fdt_setprop_cell(fdt, node, "clock-frequency", 1843200)));
|
|
|
|
_FDT((fdt_setprop_cell(fdt, node, "current-speed", 115200)));
|
2022-03-02 08:51:36 +03:00
|
|
|
_FDT((fdt_setprop_cell(fdt, node, "interrupts", irq)));
|
2017-04-11 18:30:03 +03:00
|
|
|
_FDT((fdt_setprop_cell(fdt, node, "interrupt-parent",
|
|
|
|
fdt_get_phandle(fdt, lpc_off))));
|
|
|
|
|
|
|
|
/* This is needed by Linux */
|
|
|
|
_FDT((fdt_setprop_string(fdt, node, "device_type", "serial")));
|
|
|
|
}
|
|
|
|
|
2017-12-15 16:56:01 +03:00
|
|
|
static void pnv_dt_ipmi_bt(ISADevice *d, void *fdt, int lpc_off)
|
2017-04-11 18:30:04 +03:00
|
|
|
{
|
|
|
|
const char compatible[] = "bt\0ipmi-bt";
|
|
|
|
uint32_t io_base;
|
|
|
|
uint32_t io_regs[] = {
|
|
|
|
cpu_to_be32(1),
|
|
|
|
0, /* 'io_base' retrieved from the 'ioport' property of 'isa-ipmi-bt' */
|
|
|
|
cpu_to_be32(3)
|
|
|
|
};
|
|
|
|
uint32_t irq;
|
|
|
|
char *name;
|
|
|
|
int node;
|
|
|
|
|
|
|
|
io_base = object_property_get_int(OBJECT(d), "ioport", &error_fatal);
|
|
|
|
io_regs[1] = cpu_to_be32(io_base);
|
|
|
|
|
|
|
|
irq = object_property_get_int(OBJECT(d), "irq", &error_fatal);
|
|
|
|
|
|
|
|
name = g_strdup_printf("%s@i%x", qdev_fw_name(DEVICE(d)), io_base);
|
|
|
|
node = fdt_add_subnode(fdt, lpc_off, name);
|
|
|
|
_FDT(node);
|
|
|
|
g_free(name);
|
|
|
|
|
2017-06-05 18:44:21 +03:00
|
|
|
_FDT((fdt_setprop(fdt, node, "reg", io_regs, sizeof(io_regs))));
|
|
|
|
_FDT((fdt_setprop(fdt, node, "compatible", compatible,
|
|
|
|
sizeof(compatible))));
|
2017-04-11 18:30:04 +03:00
|
|
|
|
|
|
|
/* Mark it as reserved to avoid Linux trying to claim it */
|
|
|
|
_FDT((fdt_setprop_string(fdt, node, "status", "reserved")));
|
|
|
|
_FDT((fdt_setprop_cell(fdt, node, "interrupts", irq)));
|
|
|
|
_FDT((fdt_setprop_cell(fdt, node, "interrupt-parent",
|
|
|
|
fdt_get_phandle(fdt, lpc_off))));
|
|
|
|
}
|
|
|
|
|
2017-04-11 18:30:01 +03:00
|
|
|
typedef struct ForeachPopulateArgs {
|
|
|
|
void *fdt;
|
|
|
|
int offset;
|
|
|
|
} ForeachPopulateArgs;
|
|
|
|
|
2017-12-15 16:56:01 +03:00
|
|
|
static int pnv_dt_isa_device(DeviceState *dev, void *opaque)
|
2017-04-11 18:30:01 +03:00
|
|
|
{
|
2017-04-11 18:30:02 +03:00
|
|
|
ForeachPopulateArgs *args = opaque;
|
|
|
|
ISADevice *d = ISA_DEVICE(dev);
|
|
|
|
|
|
|
|
if (object_dynamic_cast(OBJECT(dev), TYPE_MC146818_RTC)) {
|
2017-12-15 16:56:01 +03:00
|
|
|
pnv_dt_rtc(d, args->fdt, args->offset);
|
2017-04-11 18:30:03 +03:00
|
|
|
} else if (object_dynamic_cast(OBJECT(dev), TYPE_ISA_SERIAL)) {
|
2017-12-15 16:56:01 +03:00
|
|
|
pnv_dt_serial(d, args->fdt, args->offset);
|
2017-04-11 18:30:04 +03:00
|
|
|
} else if (object_dynamic_cast(OBJECT(dev), "isa-ipmi-bt")) {
|
2017-12-15 16:56:01 +03:00
|
|
|
pnv_dt_ipmi_bt(d, args->fdt, args->offset);
|
2017-04-11 18:30:02 +03:00
|
|
|
} else {
|
|
|
|
error_report("unknown isa device %s@i%x", qdev_fw_name(dev),
|
|
|
|
d->ioport_id);
|
|
|
|
}
|
|
|
|
|
2017-04-11 18:30:01 +03:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2019-09-11 17:29:25 +03:00
|
|
|
/*
|
|
|
|
* The default LPC bus of a multichip system is on chip 0. It's
|
2018-06-18 20:05:40 +03:00
|
|
|
* recognized by the firmware (skiboot) using a "primary" property.
|
|
|
|
*/
|
|
|
|
static void pnv_dt_isa(PnvMachineState *pnv, void *fdt)
|
|
|
|
{
|
2019-03-08 01:35:38 +03:00
|
|
|
int isa_offset = fdt_path_offset(fdt, pnv->chips[0]->dt_isa_nodename);
|
2017-04-11 18:30:01 +03:00
|
|
|
ForeachPopulateArgs args = {
|
|
|
|
.fdt = fdt,
|
2018-06-18 20:05:40 +03:00
|
|
|
.offset = isa_offset,
|
2017-04-11 18:30:01 +03:00
|
|
|
};
|
2019-07-23 12:01:38 +03:00
|
|
|
uint32_t phandle;
|
2017-04-11 18:30:01 +03:00
|
|
|
|
2018-06-18 20:05:40 +03:00
|
|
|
_FDT((fdt_setprop(fdt, isa_offset, "primary", NULL, 0)));
|
|
|
|
|
2019-07-23 12:01:38 +03:00
|
|
|
phandle = qemu_fdt_alloc_phandle(fdt);
|
|
|
|
assert(phandle > 0);
|
|
|
|
_FDT((fdt_setprop_cell(fdt, isa_offset, "phandle", phandle)));
|
|
|
|
|
2019-09-11 17:29:25 +03:00
|
|
|
/*
|
|
|
|
* ISA devices are not necessarily parented to the ISA bus so we
|
|
|
|
* can not use object_child_foreach()
|
|
|
|
*/
|
2018-06-18 20:05:40 +03:00
|
|
|
qbus_walk_children(BUS(pnv->isa_bus), pnv_dt_isa_device, NULL, NULL, NULL,
|
|
|
|
&args);
|
2017-04-11 18:30:01 +03:00
|
|
|
}
|
|
|
|
|
2019-12-13 14:59:56 +03:00
|
|
|
static void pnv_dt_power_mgt(PnvMachineState *pnv, void *fdt)
|
2019-03-08 01:35:47 +03:00
|
|
|
{
|
|
|
|
int off;
|
|
|
|
|
|
|
|
off = fdt_add_subnode(fdt, 0, "ibm,opal");
|
|
|
|
off = fdt_add_subnode(fdt, off, "power-mgt");
|
|
|
|
|
|
|
|
_FDT(fdt_setprop_cell(fdt, off, "ibm,enabled-stop-levels", 0xc0000000));
|
|
|
|
}
|
|
|
|
|
2017-12-15 16:56:01 +03:00
|
|
|
static void *pnv_dt_create(MachineState *machine)
|
2016-10-22 12:46:35 +03:00
|
|
|
{
|
2019-12-13 14:59:50 +03:00
|
|
|
PnvMachineClass *pmc = PNV_MACHINE_GET_CLASS(machine);
|
2017-12-15 16:56:01 +03:00
|
|
|
PnvMachineState *pnv = PNV_MACHINE(machine);
|
2016-10-22 12:46:35 +03:00
|
|
|
void *fdt;
|
|
|
|
char *buf;
|
|
|
|
int off;
|
2016-10-22 12:46:36 +03:00
|
|
|
int i;
|
2016-10-22 12:46:35 +03:00
|
|
|
|
|
|
|
fdt = g_malloc0(FDT_MAX_SIZE);
|
|
|
|
_FDT((fdt_create_empty_tree(fdt, FDT_MAX_SIZE)));
|
|
|
|
|
2019-11-06 17:21:29 +03:00
|
|
|
/* /qemu node */
|
|
|
|
_FDT((fdt_add_subnode(fdt, 0, "qemu")));
|
|
|
|
|
2016-10-22 12:46:35 +03:00
|
|
|
/* Root node */
|
|
|
|
_FDT((fdt_setprop_cell(fdt, 0, "#address-cells", 0x2)));
|
|
|
|
_FDT((fdt_setprop_cell(fdt, 0, "#size-cells", 0x2)));
|
|
|
|
_FDT((fdt_setprop_string(fdt, 0, "model",
|
|
|
|
"IBM PowerNV (emulated by qemu)")));
|
2019-12-13 14:59:50 +03:00
|
|
|
_FDT((fdt_setprop(fdt, 0, "compatible", pmc->compat, pmc->compat_size)));
|
2016-10-22 12:46:35 +03:00
|
|
|
|
|
|
|
buf = qemu_uuid_unparse_strdup(&qemu_uuid);
|
|
|
|
_FDT((fdt_setprop_string(fdt, 0, "vm,uuid", buf)));
|
|
|
|
if (qemu_uuid_set) {
|
2021-12-17 19:57:13 +03:00
|
|
|
_FDT((fdt_setprop_string(fdt, 0, "system-id", buf)));
|
2016-10-22 12:46:35 +03:00
|
|
|
}
|
|
|
|
g_free(buf);
|
|
|
|
|
|
|
|
off = fdt_add_subnode(fdt, 0, "chosen");
|
|
|
|
if (machine->kernel_cmdline) {
|
|
|
|
_FDT((fdt_setprop_string(fdt, off, "bootargs",
|
|
|
|
machine->kernel_cmdline)));
|
|
|
|
}
|
|
|
|
|
|
|
|
if (pnv->initrd_size) {
|
|
|
|
uint32_t start_prop = cpu_to_be32(pnv->initrd_base);
|
|
|
|
uint32_t end_prop = cpu_to_be32(pnv->initrd_base + pnv->initrd_size);
|
|
|
|
|
|
|
|
_FDT((fdt_setprop(fdt, off, "linux,initrd-start",
|
|
|
|
&start_prop, sizeof(start_prop))));
|
|
|
|
_FDT((fdt_setprop(fdt, off, "linux,initrd-end",
|
|
|
|
&end_prop, sizeof(end_prop))));
|
|
|
|
}
|
|
|
|
|
2016-10-22 12:46:36 +03:00
|
|
|
/* Populate device tree for each chip */
|
|
|
|
for (i = 0; i < pnv->num_chips; i++) {
|
2019-03-06 11:50:12 +03:00
|
|
|
PNV_CHIP_GET_CLASS(pnv->chips[i])->dt_populate(pnv->chips[i], fdt);
|
2016-10-22 12:46:36 +03:00
|
|
|
}
|
2017-04-11 18:30:01 +03:00
|
|
|
|
|
|
|
/* Populate ISA devices on chip 0 */
|
2018-06-18 20:05:40 +03:00
|
|
|
pnv_dt_isa(pnv, fdt);
|
2017-04-11 18:30:05 +03:00
|
|
|
|
|
|
|
if (pnv->bmc) {
|
2017-12-15 16:56:01 +03:00
|
|
|
pnv_dt_bmc_sensors(pnv->bmc, fdt);
|
2017-04-11 18:30:05 +03:00
|
|
|
}
|
|
|
|
|
2019-12-13 14:59:56 +03:00
|
|
|
/* Create an extra node for power management on machines that support it */
|
|
|
|
if (pmc->dt_power_mgt) {
|
|
|
|
pmc->dt_power_mgt(pnv, fdt);
|
2019-03-08 01:35:47 +03:00
|
|
|
}
|
|
|
|
|
2016-10-22 12:46:35 +03:00
|
|
|
return fdt;
|
|
|
|
}
|
|
|
|
|
2017-04-11 18:30:06 +03:00
|
|
|
static void pnv_powerdown_notify(Notifier *n, void *opaque)
|
|
|
|
{
|
2019-12-19 21:11:43 +03:00
|
|
|
PnvMachineState *pnv = container_of(n, PnvMachineState, powerdown_notifier);
|
2017-04-11 18:30:06 +03:00
|
|
|
|
|
|
|
if (pnv->bmc) {
|
|
|
|
pnv_bmc_powerdown(pnv->bmc);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2019-05-18 23:54:20 +03:00
|
|
|
static void pnv_reset(MachineState *machine)
|
2016-10-22 12:46:35 +03:00
|
|
|
{
|
2020-04-04 18:36:55 +03:00
|
|
|
PnvMachineState *pnv = PNV_MACHINE(machine);
|
|
|
|
IPMIBmc *bmc;
|
2016-10-22 12:46:35 +03:00
|
|
|
void *fdt;
|
|
|
|
|
|
|
|
qemu_devices_reset();
|
|
|
|
|
2020-04-04 18:36:55 +03:00
|
|
|
/*
|
|
|
|
* The machine should provide by default an internal BMC simulator.
|
|
|
|
* If not, try to use the BMC device that was provided on the command
|
|
|
|
* line.
|
|
|
|
*/
|
|
|
|
bmc = pnv_bmc_find(&error_fatal);
|
|
|
|
if (!pnv->bmc) {
|
|
|
|
if (!bmc) {
|
ppc/pnv: Silence missing BMC warning with qtest
The device introspect test in qtest emits some warnings with the
the pnv machine types during the "nodefaults" phase:
TEST check-qtest-ppc64: tests/qtest/device-introspect-test
qemu-system-ppc64: warning: machine has no BMC device. Use '-device
ipmi-bmc-sim,id=bmc0 -device isa-ipmi-bt,bmc=bmc0,irq=10' to define
one
qemu-system-ppc64: warning: machine has no BMC device. Use '-device
ipmi-bmc-sim,id=bmc0 -device isa-ipmi-bt,bmc=bmc0,irq=10' to define
one
qemu-system-ppc64: warning: machine has no BMC device. Use '-device
ipmi-bmc-sim,id=bmc0 -device isa-ipmi-bt,bmc=bmc0,irq=10' to define
one
This is expected since the pnv machine doesn't create the internal
BMC simulator fallback when "-nodefaults" is passed on the command
line, but these warnings appear in ci logs and confuse people.
Not having a BMC isn't recommended but it is still a supported
configuration, so a straightforward fix is to just silent this
warning when qtest is enabled.
Fixes: 25f3170b0654 ("ppc/pnv: Create BMC devices only when defaults are enabled")
Reported-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Greg Kurz <groug@kaod.org>
Message-Id: <159280903824.485572.831378159272329707.stgit@bahia.lan>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2020-06-22 09:57:18 +03:00
|
|
|
if (!qtest_enabled()) {
|
|
|
|
warn_report("machine has no BMC device. Use '-device "
|
|
|
|
"ipmi-bmc-sim,id=bmc0 -device isa-ipmi-bt,bmc=bmc0,irq=10' "
|
|
|
|
"to define one");
|
|
|
|
}
|
2020-04-04 18:36:55 +03:00
|
|
|
} else {
|
|
|
|
pnv_bmc_set_pnor(bmc, pnv->pnor);
|
|
|
|
pnv->bmc = bmc;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2017-12-15 16:56:01 +03:00
|
|
|
fdt = pnv_dt_create(machine);
|
2016-10-22 12:46:35 +03:00
|
|
|
|
|
|
|
/* Pack resulting tree */
|
|
|
|
_FDT((fdt_pack(fdt)));
|
|
|
|
|
2019-06-06 20:47:32 +03:00
|
|
|
qemu_fdt_dumpdtb(fdt, fdt_totalsize(fdt));
|
2016-10-22 12:46:35 +03:00
|
|
|
cpu_physical_memory_write(PNV_FDT_ADDR, fdt, fdt_totalsize(fdt));
|
2020-02-14 06:32:06 +03:00
|
|
|
|
|
|
|
g_free(fdt);
|
2016-10-22 12:46:35 +03:00
|
|
|
}
|
|
|
|
|
2018-06-15 18:25:34 +03:00
|
|
|
static ISABus *pnv_chip_power8_isa_create(PnvChip *chip, Error **errp)
|
2016-10-22 12:46:43 +03:00
|
|
|
{
|
2018-06-18 20:05:39 +03:00
|
|
|
Pnv8Chip *chip8 = PNV8_CHIP(chip);
|
2022-03-23 10:28:43 +03:00
|
|
|
qemu_irq irq = qdev_get_gpio_in(DEVICE(&chip8->psi), PSIHB_IRQ_EXTERNAL);
|
|
|
|
|
|
|
|
qdev_connect_gpio_out(DEVICE(&chip8->lpc), 0, irq);
|
2018-06-18 20:05:39 +03:00
|
|
|
return pnv_lpc_isa_create(&chip8->lpc, true, errp);
|
2018-06-15 18:25:34 +03:00
|
|
|
}
|
2016-10-22 12:46:43 +03:00
|
|
|
|
2018-06-15 18:25:34 +03:00
|
|
|
static ISABus *pnv_chip_power8nvl_isa_create(PnvChip *chip, Error **errp)
|
|
|
|
{
|
2018-06-18 20:05:39 +03:00
|
|
|
Pnv8Chip *chip8 = PNV8_CHIP(chip);
|
2022-03-23 10:28:43 +03:00
|
|
|
qemu_irq irq = qdev_get_gpio_in(DEVICE(&chip8->psi), PSIHB_IRQ_LPC_I2C);
|
|
|
|
|
|
|
|
qdev_connect_gpio_out(DEVICE(&chip8->lpc), 0, irq);
|
2018-06-18 20:05:39 +03:00
|
|
|
return pnv_lpc_isa_create(&chip8->lpc, false, errp);
|
2018-06-15 18:25:34 +03:00
|
|
|
}
|
2016-10-22 12:46:43 +03:00
|
|
|
|
2018-06-15 18:25:34 +03:00
|
|
|
static ISABus *pnv_chip_power9_isa_create(PnvChip *chip, Error **errp)
|
|
|
|
{
|
2019-03-08 01:35:39 +03:00
|
|
|
Pnv9Chip *chip9 = PNV9_CHIP(chip);
|
2022-03-23 10:28:43 +03:00
|
|
|
qemu_irq irq = qdev_get_gpio_in(DEVICE(&chip9->psi), PSIHB9_IRQ_LPCHC);
|
|
|
|
|
|
|
|
qdev_connect_gpio_out(DEVICE(&chip9->lpc), 0, irq);
|
2019-03-08 01:35:39 +03:00
|
|
|
return pnv_lpc_isa_create(&chip9->lpc, false, errp);
|
2018-06-15 18:25:34 +03:00
|
|
|
}
|
2016-10-22 12:46:43 +03:00
|
|
|
|
2019-12-05 21:44:51 +03:00
|
|
|
static ISABus *pnv_chip_power10_isa_create(PnvChip *chip, Error **errp)
|
|
|
|
{
|
2019-12-05 21:44:54 +03:00
|
|
|
Pnv10Chip *chip10 = PNV10_CHIP(chip);
|
2022-03-23 10:28:43 +03:00
|
|
|
qemu_irq irq = qdev_get_gpio_in(DEVICE(&chip10->psi), PSIHB9_IRQ_LPCHC);
|
|
|
|
|
|
|
|
qdev_connect_gpio_out(DEVICE(&chip10->lpc), 0, irq);
|
2019-12-05 21:44:54 +03:00
|
|
|
return pnv_lpc_isa_create(&chip10->lpc, false, errp);
|
2019-12-05 21:44:51 +03:00
|
|
|
}
|
|
|
|
|
2018-06-15 18:25:34 +03:00
|
|
|
static ISABus *pnv_isa_create(PnvChip *chip, Error **errp)
|
|
|
|
{
|
|
|
|
return PNV_CHIP_GET_CLASS(chip)->isa_create(chip, errp);
|
2016-10-22 12:46:43 +03:00
|
|
|
}
|
|
|
|
|
2019-03-06 11:50:13 +03:00
|
|
|
static void pnv_chip_power8_pic_print_info(PnvChip *chip, Monitor *mon)
|
|
|
|
{
|
|
|
|
Pnv8Chip *chip8 = PNV8_CHIP(chip);
|
2022-06-21 20:34:34 +03:00
|
|
|
int i;
|
2019-03-06 11:50:13 +03:00
|
|
|
|
|
|
|
ics_pic_print_info(&chip8->psi.ics, mon);
|
2022-06-21 20:34:34 +03:00
|
|
|
|
|
|
|
for (i = 0; i < chip8->num_phbs; i++) {
|
2022-08-11 19:39:44 +03:00
|
|
|
PnvPHB *phb = chip8->phbs[i];
|
2022-06-24 11:49:12 +03:00
|
|
|
PnvPHB3 *phb3 = PNV_PHB3(phb->backend);
|
2022-06-21 20:34:34 +03:00
|
|
|
|
|
|
|
pnv_phb3_msi_pic_print_info(&phb3->msis, mon);
|
|
|
|
ics_pic_print_info(&phb3->lsis, mon);
|
|
|
|
}
|
2019-03-06 11:50:13 +03:00
|
|
|
}
|
|
|
|
|
2021-12-17 19:57:19 +03:00
|
|
|
static int pnv_chip_power9_pic_print_info_child(Object *child, void *opaque)
|
|
|
|
{
|
|
|
|
Monitor *mon = opaque;
|
2022-06-24 11:49:14 +03:00
|
|
|
PnvPHB *phb = (PnvPHB *) object_dynamic_cast(child, TYPE_PNV_PHB);
|
2021-12-17 19:57:19 +03:00
|
|
|
|
2022-06-24 11:49:14 +03:00
|
|
|
if (!phb) {
|
|
|
|
return 0;
|
2021-12-17 19:57:19 +03:00
|
|
|
}
|
2022-06-24 11:49:14 +03:00
|
|
|
|
|
|
|
pnv_phb4_pic_print_info(PNV_PHB4(phb->backend), mon);
|
|
|
|
|
2021-12-17 19:57:19 +03:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2019-03-06 11:50:13 +03:00
|
|
|
static void pnv_chip_power9_pic_print_info(PnvChip *chip, Monitor *mon)
|
|
|
|
{
|
|
|
|
Pnv9Chip *chip9 = PNV9_CHIP(chip);
|
|
|
|
|
|
|
|
pnv_xive_pic_print_info(&chip9->xive, mon);
|
2019-03-08 01:35:35 +03:00
|
|
|
pnv_psi_pic_print_info(&chip9->psi, mon);
|
ppc/pnv: Add models for POWER9 PHB4 PCIe Host bridge
These changes introduces models for the PCIe Host Bridge (PHB4) of the
POWER9 processor. It includes the PowerBus logic interface (PBCQ),
IOMMU support, a single PCIe Gen.4 Root Complex, and support for MSI
and LSI interrupt sources as found on a POWER9 system using the XIVE
interrupt controller.
POWER9 processor comes with 3 PHB4 PEC (PCI Express Controller) and
each PEC can have several PHBs. By default,
* PEC0 provides 1 PHB (PHB0)
* PEC1 provides 2 PHBs (PHB1 and PHB2)
* PEC2 provides 3 PHBs (PHB3, PHB4 and PHB5)
Each PEC has a set "global" registers and some "per-stack" (per-PHB)
registers. Those are organized in two XSCOM ranges, the "Nest" range
and the "PCI" range, each range contains both some "PEC" registers and
some "per-stack" registers.
No default device layout is provided and PCI devices can be added on
any of the available PCIe Root Port (pcie.0 .. 2 of a Power9 chip)
with address 0x0 as the firwware (skiboot) only accepts a single
device per root port. To run a simple system with a network and a
storage adapters, use a command line options such as :
-device e1000e,netdev=net0,mac=C0:FF:EE:00:00:02,bus=pcie.0,addr=0x0
-netdev bridge,id=net0,helper=/usr/libexec/qemu-bridge-helper,br=virbr0,id=hostnet0
-device megasas,id=scsi0,bus=pcie.1,addr=0x0
-drive file=$disk,if=none,id=drive-scsi0-0-0-0,format=qcow2,cache=none
-device scsi-hd,bus=scsi0.0,channel=0,scsi-id=0,lun=0,drive=drive-scsi0-0-0-0,id=scsi0-0-0-0,bootindex=2
If more are needed, include a bridge.
Multi chip is supported, each chip adding its set of PHB4 controllers
and its PCI busses. The model doesn't emulate the EEH error handling.
This model is not ready for hotplug yet.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[ clg: - numerous cleanups
- commit log
- fix for broken LSI support
- PHB pic printinfo
- large QOM rework ]
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20200127144506.11132-2-clg@kaod.org>
[dwg: Use device_class_set_props()]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2020-01-27 17:45:05 +03:00
|
|
|
|
2021-12-17 19:57:19 +03:00
|
|
|
object_child_foreach_recursive(OBJECT(chip),
|
|
|
|
pnv_chip_power9_pic_print_info_child, mon);
|
2019-03-06 11:50:13 +03:00
|
|
|
}
|
|
|
|
|
2019-12-13 15:00:13 +03:00
|
|
|
static uint64_t pnv_chip_power8_xscom_core_base(PnvChip *chip,
|
|
|
|
uint32_t core_id)
|
|
|
|
{
|
|
|
|
return PNV_XSCOM_EX_BASE(core_id);
|
|
|
|
}
|
|
|
|
|
|
|
|
static uint64_t pnv_chip_power9_xscom_core_base(PnvChip *chip,
|
|
|
|
uint32_t core_id)
|
|
|
|
{
|
|
|
|
return PNV9_XSCOM_EC_BASE(core_id);
|
|
|
|
}
|
|
|
|
|
|
|
|
static uint64_t pnv_chip_power10_xscom_core_base(PnvChip *chip,
|
|
|
|
uint32_t core_id)
|
|
|
|
{
|
|
|
|
return PNV10_XSCOM_EC_BASE(core_id);
|
|
|
|
}
|
|
|
|
|
2019-07-31 17:12:16 +03:00
|
|
|
static bool pnv_match_cpu(const char *default_type, const char *cpu_type)
|
|
|
|
{
|
|
|
|
PowerPCCPUClass *ppc_default =
|
|
|
|
POWERPC_CPU_CLASS(object_class_by_name(default_type));
|
|
|
|
PowerPCCPUClass *ppc =
|
|
|
|
POWERPC_CPU_CLASS(object_class_by_name(cpu_type));
|
|
|
|
|
2022-07-31 04:33:58 +03:00
|
|
|
return ppc_default->pvr_match(ppc_default, ppc->pvr, false);
|
2019-07-31 17:12:16 +03:00
|
|
|
}
|
|
|
|
|
ppc/pnv: Create BMC devices at machine init
The BMC of the OpenPOWER systems monitors the machine state using
sensors, controls the power and controls the access to the PNOR flash
device containing the firmware image required to boot the host.
QEMU models the power cycle process, access to the sensors and access
to the PNOR device. But, for these features to be available, the QEMU
PowerNV machine needs two extras devices on the command line, an IPMI
BT device for communication and a BMC backend device:
-device ipmi-bmc-sim,id=bmc0 -device isa-ipmi-bt,bmc=bmc0,irq=10
The BMC properties are then defined accordingly in the device tree and
OPAL self adapts. If a BMC device and an IPMI BT device are not
available, OPAL does not try to communicate with the BMC in any
manner. This is not how real systems behave.
To be closer to the default behavior, create an IPMI BMC simulator
device and an IPMI BT device at machine initialization time. We loose
the ability to define an external BMC device but there are benefits:
- a better match with real systems,
- a better test coverage of the OPAL code,
- system powerdown and reset commands that work,
- a QEMU device tree compliant with the specifications (*).
(*) Still needs a MBOX device.
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20191121162340.11049-1-clg@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2019-11-21 19:23:40 +03:00
|
|
|
static void pnv_ipmi_bt_init(ISABus *bus, IPMIBmc *bmc, uint32_t irq)
|
|
|
|
{
|
2020-06-10 08:32:09 +03:00
|
|
|
ISADevice *dev = isa_new("isa-ipmi-bt");
|
ppc/pnv: Create BMC devices at machine init
The BMC of the OpenPOWER systems monitors the machine state using
sensors, controls the power and controls the access to the PNOR flash
device containing the firmware image required to boot the host.
QEMU models the power cycle process, access to the sensors and access
to the PNOR device. But, for these features to be available, the QEMU
PowerNV machine needs two extras devices on the command line, an IPMI
BT device for communication and a BMC backend device:
-device ipmi-bmc-sim,id=bmc0 -device isa-ipmi-bt,bmc=bmc0,irq=10
The BMC properties are then defined accordingly in the device tree and
OPAL self adapts. If a BMC device and an IPMI BT device are not
available, OPAL does not try to communicate with the BMC in any
manner. This is not how real systems behave.
To be closer to the default behavior, create an IPMI BMC simulator
device and an IPMI BT device at machine initialization time. We loose
the ability to define an external BMC device but there are benefits:
- a better match with real systems,
- a better test coverage of the OPAL code,
- system powerdown and reset commands that work,
- a QEMU device tree compliant with the specifications (*).
(*) Still needs a MBOX device.
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20191121162340.11049-1-clg@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2019-11-21 19:23:40 +03:00
|
|
|
|
qom: Put name parameter before value / visitor parameter
The object_property_set_FOO() setters take property name and value in
an unusual order:
void object_property_set_FOO(Object *obj, FOO_TYPE value,
const char *name, Error **errp)
Having to pass value before name feels grating. Swap them.
Same for object_property_set(), object_property_get(), and
object_property_parse().
Convert callers with this Coccinelle script:
@@
identifier fun = {
object_property_get, object_property_parse, object_property_set_str,
object_property_set_link, object_property_set_bool,
object_property_set_int, object_property_set_uint, object_property_set,
object_property_set_qobject
};
expression obj, v, name, errp;
@@
- fun(obj, v, name, errp)
+ fun(obj, name, v, errp)
Chokes on hw/arm/musicpal.c's lcd_refresh() with the unhelpful error
message "no position information". Convert that one manually.
Fails to convert hw/arm/armsse.c, because Coccinelle gets confused by
ARMSSE being used both as typedef and function-like macro there.
Convert manually.
Fails to convert hw/rx/rx-gdbsim.c, because Coccinelle gets confused
by RXCPU being used both as typedef and function-like macro there.
Convert manually. The other files using RXCPU that way don't need
conversion.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20200707160613.848843-27-armbru@redhat.com>
[Straightforwad conflict with commit 2336172d9b "audio: set default
value for pcspk.iobase property" resolved]
2020-07-07 19:05:54 +03:00
|
|
|
object_property_set_link(OBJECT(dev), "bmc", OBJECT(bmc), &error_fatal);
|
|
|
|
object_property_set_int(OBJECT(dev), "irq", irq, &error_fatal);
|
2020-06-10 08:32:09 +03:00
|
|
|
isa_realize_and_unref(dev, bus, &error_fatal);
|
ppc/pnv: Create BMC devices at machine init
The BMC of the OpenPOWER systems monitors the machine state using
sensors, controls the power and controls the access to the PNOR flash
device containing the firmware image required to boot the host.
QEMU models the power cycle process, access to the sensors and access
to the PNOR device. But, for these features to be available, the QEMU
PowerNV machine needs two extras devices on the command line, an IPMI
BT device for communication and a BMC backend device:
-device ipmi-bmc-sim,id=bmc0 -device isa-ipmi-bt,bmc=bmc0,irq=10
The BMC properties are then defined accordingly in the device tree and
OPAL self adapts. If a BMC device and an IPMI BT device are not
available, OPAL does not try to communicate with the BMC in any
manner. This is not how real systems behave.
To be closer to the default behavior, create an IPMI BMC simulator
device and an IPMI BT device at machine initialization time. We loose
the ability to define an external BMC device but there are benefits:
- a better match with real systems,
- a better test coverage of the OPAL code,
- system powerdown and reset commands that work,
- a QEMU device tree compliant with the specifications (*).
(*) Still needs a MBOX device.
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20191121162340.11049-1-clg@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2019-11-21 19:23:40 +03:00
|
|
|
}
|
|
|
|
|
2019-12-05 21:44:51 +03:00
|
|
|
static void pnv_chip_power10_pic_print_info(PnvChip *chip, Monitor *mon)
|
|
|
|
{
|
2019-12-05 21:44:53 +03:00
|
|
|
Pnv10Chip *chip10 = PNV10_CHIP(chip);
|
|
|
|
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 08:51:38 +03:00
|
|
|
pnv_xive2_pic_print_info(&chip10->xive, mon);
|
2019-12-05 21:44:53 +03:00
|
|
|
pnv_psi_pic_print_info(&chip10->psi, mon);
|
2022-03-02 08:51:39 +03:00
|
|
|
|
|
|
|
object_child_foreach_recursive(OBJECT(chip),
|
|
|
|
pnv_chip_power9_pic_print_info_child, mon);
|
2019-12-05 21:44:51 +03:00
|
|
|
}
|
|
|
|
|
2021-08-09 16:45:26 +03:00
|
|
|
/* Always give the first 1GB to chip 0 else we won't boot */
|
|
|
|
static uint64_t pnv_chip_get_ram_size(PnvMachineState *pnv, int chip_id)
|
|
|
|
{
|
|
|
|
MachineState *machine = MACHINE(pnv);
|
|
|
|
uint64_t ram_per_chip;
|
|
|
|
|
|
|
|
assert(machine->ram_size >= 1 * GiB);
|
|
|
|
|
|
|
|
ram_per_chip = machine->ram_size / pnv->num_chips;
|
|
|
|
if (ram_per_chip >= 1 * GiB) {
|
|
|
|
return QEMU_ALIGN_DOWN(ram_per_chip, 1 * MiB);
|
|
|
|
}
|
|
|
|
|
2021-09-02 16:09:10 +03:00
|
|
|
assert(pnv->num_chips > 1);
|
|
|
|
|
2021-08-09 16:45:26 +03:00
|
|
|
ram_per_chip = (machine->ram_size - 1 * GiB) / (pnv->num_chips - 1);
|
|
|
|
return chip_id == 0 ? 1 * GiB : QEMU_ALIGN_DOWN(ram_per_chip, 1 * MiB);
|
|
|
|
}
|
|
|
|
|
2017-12-15 16:56:01 +03:00
|
|
|
static void pnv_init(MachineState *machine)
|
2016-10-22 12:46:35 +03:00
|
|
|
{
|
2020-10-26 17:30:23 +03:00
|
|
|
const char *bios_name = machine->firmware ?: FW_FILE_NAME;
|
2017-12-15 16:56:01 +03:00
|
|
|
PnvMachineState *pnv = PNV_MACHINE(machine);
|
2019-07-31 17:12:16 +03:00
|
|
|
MachineClass *mc = MACHINE_GET_CLASS(machine);
|
2016-10-22 12:46:35 +03:00
|
|
|
char *fw_filename;
|
|
|
|
long fw_size;
|
2021-08-09 16:45:26 +03:00
|
|
|
uint64_t chip_ram_start = 0;
|
2016-10-22 12:46:36 +03:00
|
|
|
int i;
|
|
|
|
char *chip_typename;
|
2019-10-21 16:12:11 +03:00
|
|
|
DriveInfo *pnor = drive_get(IF_MTD, 0, 0);
|
|
|
|
DeviceState *dev;
|
2016-10-22 12:46:35 +03:00
|
|
|
|
2021-12-17 19:57:13 +03:00
|
|
|
if (kvm_enabled()) {
|
|
|
|
error_report("The powernv machine does not work with KVM acceleration");
|
|
|
|
exit(EXIT_FAILURE);
|
|
|
|
}
|
|
|
|
|
2016-10-22 12:46:35 +03:00
|
|
|
/* allocate RAM */
|
ppc/pnv: Set default RAM size to 1 GB
The memory layout of the PowerNV machine is defined as :
#define KERNEL_LOAD_BASE ((void *)0x20000000)
#define KERNEL_LOAD_SIZE 0x08000000
#define INITRAMFS_LOAD_BASE KERNEL_LOAD_BASE + KERNEL_LOAD_SIZE
#define INITRAMFS_LOAD_SIZE 0x08000000
#define SKIBOOT_BASE 0x30000000
#define SKIBOOT_SIZE 0x01c10000
#define CPU_STACKS_BASE (SKIBOOT_BASE + SKIBOOT_SIZE)
#define STACK_SHIFT 15
#define STACK_SIZE (1 << STACK_SHIFT)
The overall size of the CPU stacks is (max PIR + 1) * 32K and the
machine easily reaches 800MB of minimum required RAM.
Any value below will result in a skiboot crash :
[ 0.034949905,3] MEM: Partial overlap detected between regions:
[ 0.034959039,3] MEM: ibm,firmware-stacks [0x31c10000-0x3a450000] (new)
[ 0.034968576,3] MEM: ibm,firmware-allocs-memory@0 [0x31c10000-0x38400000]
[ 0.034980367,3] Out of memory adding skiboot reserved areas
[ 0.035074945,3] ***********************************************
[ 0.035093627,3] < assert failed at core/mem_region.c:1129 >
[ 0.035104247,3] .
[ 0.035108025,3] .
[ 0.035111651,3] .
[ 0.035115231,3] OO__)
[ 0.035119198,3] <"__/
[ 0.035122980,3] ^ ^
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20210129111719.790692-1-clg@kaod.org>
Reviewed-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2021-01-29 14:17:19 +03:00
|
|
|
if (machine->ram_size < mc->default_ram_size) {
|
|
|
|
char *sz = size_to_str(mc->default_ram_size);
|
|
|
|
error_report("Invalid RAM size, should be bigger than %s", sz);
|
|
|
|
g_free(sz);
|
|
|
|
exit(EXIT_FAILURE);
|
2016-10-22 12:46:35 +03:00
|
|
|
}
|
2020-02-19 19:09:36 +03:00
|
|
|
memory_region_add_subregion(get_system_memory(), 0, machine->ram);
|
2016-10-22 12:46:35 +03:00
|
|
|
|
2019-10-21 16:12:11 +03:00
|
|
|
/*
|
|
|
|
* Create our simple PNOR device
|
|
|
|
*/
|
qdev: Convert uses of qdev_create() with Coccinelle
This is the transformation explained in the commit before previous.
Takes care of just one pattern that needs conversion. More to come in
this series.
Coccinelle script:
@ depends on !(file in "hw/arm/highbank.c")@
expression bus, type_name, dev, expr;
@@
- dev = qdev_create(bus, type_name);
+ dev = qdev_new(type_name);
... when != dev = expr
- qdev_init_nofail(dev);
+ qdev_realize_and_unref(dev, bus, &error_fatal);
@@
expression bus, type_name, dev, expr;
identifier DOWN;
@@
- dev = DOWN(qdev_create(bus, type_name));
+ dev = DOWN(qdev_new(type_name));
... when != dev = expr
- qdev_init_nofail(DEVICE(dev));
+ qdev_realize_and_unref(DEVICE(dev), bus, &error_fatal);
@@
expression bus, type_name, expr;
identifier dev;
@@
- DeviceState *dev = qdev_create(bus, type_name);
+ DeviceState *dev = qdev_new(type_name);
... when != dev = expr
- qdev_init_nofail(dev);
+ qdev_realize_and_unref(dev, bus, &error_fatal);
@@
expression bus, type_name, dev, expr, errp;
symbol true;
@@
- dev = qdev_create(bus, type_name);
+ dev = qdev_new(type_name);
... when != dev = expr
- object_property_set_bool(OBJECT(dev), true, "realized", errp);
+ qdev_realize_and_unref(dev, bus, errp);
@@
expression bus, type_name, expr, errp;
identifier dev;
symbol true;
@@
- DeviceState *dev = qdev_create(bus, type_name);
+ DeviceState *dev = qdev_new(type_name);
... when != dev = expr
- object_property_set_bool(OBJECT(dev), true, "realized", errp);
+ qdev_realize_and_unref(dev, bus, errp);
The first rule exempts hw/arm/highbank.c, because it matches along two
control flow paths there, with different @type_name. Covered by the
next commit's manual conversions.
Missing #include "qapi/error.h" added manually.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200610053247.1583243-10-armbru@redhat.com>
[Conflicts in hw/misc/empty_slot.c and hw/sparc/leon3.c resolved]
2020-06-10 08:31:58 +03:00
|
|
|
dev = qdev_new(TYPE_PNV_PNOR);
|
2019-10-21 16:12:11 +03:00
|
|
|
if (pnor) {
|
qdev: Make qdev_prop_set_drive() match the other helpers
qdev_prop_set_drive() can fail. None of the other qdev_prop_set_FOO()
can; they abort on error.
To clean up this inconsistency, rename qdev_prop_set_drive() to
qdev_prop_set_drive_err(), and create a qdev_prop_set_drive() that
aborts on error.
Coccinelle script to update callers:
@ depends on !(file in "hw/core/qdev-properties-system.c")@
expression dev, name, value;
symbol error_abort;
@@
- qdev_prop_set_drive(dev, name, value, &error_abort);
+ qdev_prop_set_drive(dev, name, value);
@@
expression dev, name, value, errp;
@@
- qdev_prop_set_drive(dev, name, value, errp);
+ qdev_prop_set_drive_err(dev, name, value, errp);
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20200622094227.1271650-14-armbru@redhat.com>
2020-06-22 12:42:24 +03:00
|
|
|
qdev_prop_set_drive(dev, "drive", blk_by_legacy_dinfo(pnor));
|
2019-10-21 16:12:11 +03:00
|
|
|
}
|
sysbus: Convert to sysbus_realize() etc. with Coccinelle
Convert from qdev_realize(), qdev_realize_and_unref() with null @bus
argument to sysbus_realize(), sysbus_realize_and_unref().
Coccinelle script:
@@
expression dev, errp;
@@
- qdev_realize(DEVICE(dev), NULL, errp);
+ sysbus_realize(SYS_BUS_DEVICE(dev), errp);
@@
expression sysbus_dev, dev, errp;
@@
+ sysbus_dev = SYS_BUS_DEVICE(dev);
- qdev_realize_and_unref(dev, NULL, errp);
+ sysbus_realize_and_unref(sysbus_dev, errp);
- sysbus_dev = SYS_BUS_DEVICE(dev);
@@
expression sysbus_dev, dev, errp;
expression expr;
@@
sysbus_dev = SYS_BUS_DEVICE(dev);
... when != dev = expr;
- qdev_realize_and_unref(dev, NULL, errp);
+ sysbus_realize_and_unref(sysbus_dev, errp);
@@
expression dev, errp;
@@
- qdev_realize_and_unref(DEVICE(dev), NULL, errp);
+ sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), errp);
@@
expression dev, errp;
@@
- qdev_realize_and_unref(dev, NULL, errp);
+ sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), errp);
Whitespace changes minimized manually.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200610053247.1583243-46-armbru@redhat.com>
[Conflicts in hw/misc/empty_slot.c and hw/sparc/leon3.c resolved]
2020-06-10 08:32:34 +03:00
|
|
|
sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), &error_fatal);
|
2019-10-21 16:12:11 +03:00
|
|
|
pnv->pnor = PNV_PNOR(dev);
|
|
|
|
|
2016-10-22 12:46:35 +03:00
|
|
|
/* load skiboot firmware */
|
|
|
|
fw_filename = qemu_find_file(QEMU_FILE_TYPE_BIOS, bios_name);
|
2017-09-25 11:58:25 +03:00
|
|
|
if (!fw_filename) {
|
|
|
|
error_report("Could not find OPAL firmware '%s'", bios_name);
|
|
|
|
exit(1);
|
|
|
|
}
|
2016-10-22 12:46:35 +03:00
|
|
|
|
2020-01-27 17:41:54 +03:00
|
|
|
fw_size = load_image_targphys(fw_filename, pnv->fw_load_addr, FW_MAX_SIZE);
|
2016-10-22 12:46:35 +03:00
|
|
|
if (fw_size < 0) {
|
2017-09-25 11:58:25 +03:00
|
|
|
error_report("Could not load OPAL firmware '%s'", fw_filename);
|
2016-10-22 12:46:35 +03:00
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
g_free(fw_filename);
|
|
|
|
|
|
|
|
/* load kernel */
|
|
|
|
if (machine->kernel_filename) {
|
|
|
|
long kernel_size;
|
|
|
|
|
|
|
|
kernel_size = load_image_targphys(machine->kernel_filename,
|
2019-02-25 20:01:53 +03:00
|
|
|
KERNEL_LOAD_ADDR, KERNEL_MAX_SIZE);
|
2016-10-22 12:46:35 +03:00
|
|
|
if (kernel_size < 0) {
|
2017-02-08 21:31:57 +03:00
|
|
|
error_report("Could not load kernel '%s'",
|
2017-02-01 14:03:10 +03:00
|
|
|
machine->kernel_filename);
|
2016-10-22 12:46:35 +03:00
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* load initrd */
|
|
|
|
if (machine->initrd_filename) {
|
|
|
|
pnv->initrd_base = INITRD_LOAD_ADDR;
|
|
|
|
pnv->initrd_size = load_image_targphys(machine->initrd_filename,
|
2019-02-25 20:01:54 +03:00
|
|
|
pnv->initrd_base, INITRD_MAX_SIZE);
|
2016-10-22 12:46:35 +03:00
|
|
|
if (pnv->initrd_size < 0) {
|
2017-02-08 21:31:57 +03:00
|
|
|
error_report("Could not load initial ram disk '%s'",
|
2016-10-22 12:46:35 +03:00
|
|
|
machine->initrd_filename);
|
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
}
|
2016-10-22 12:46:36 +03:00
|
|
|
|
ppc/pnv: Add models for POWER9 PHB4 PCIe Host bridge
These changes introduces models for the PCIe Host Bridge (PHB4) of the
POWER9 processor. It includes the PowerBus logic interface (PBCQ),
IOMMU support, a single PCIe Gen.4 Root Complex, and support for MSI
and LSI interrupt sources as found on a POWER9 system using the XIVE
interrupt controller.
POWER9 processor comes with 3 PHB4 PEC (PCI Express Controller) and
each PEC can have several PHBs. By default,
* PEC0 provides 1 PHB (PHB0)
* PEC1 provides 2 PHBs (PHB1 and PHB2)
* PEC2 provides 3 PHBs (PHB3, PHB4 and PHB5)
Each PEC has a set "global" registers and some "per-stack" (per-PHB)
registers. Those are organized in two XSCOM ranges, the "Nest" range
and the "PCI" range, each range contains both some "PEC" registers and
some "per-stack" registers.
No default device layout is provided and PCI devices can be added on
any of the available PCIe Root Port (pcie.0 .. 2 of a Power9 chip)
with address 0x0 as the firwware (skiboot) only accepts a single
device per root port. To run a simple system with a network and a
storage adapters, use a command line options such as :
-device e1000e,netdev=net0,mac=C0:FF:EE:00:00:02,bus=pcie.0,addr=0x0
-netdev bridge,id=net0,helper=/usr/libexec/qemu-bridge-helper,br=virbr0,id=hostnet0
-device megasas,id=scsi0,bus=pcie.1,addr=0x0
-drive file=$disk,if=none,id=drive-scsi0-0-0-0,format=qcow2,cache=none
-device scsi-hd,bus=scsi0.0,channel=0,scsi-id=0,lun=0,drive=drive-scsi0-0-0-0,id=scsi0-0-0-0,bootindex=2
If more are needed, include a bridge.
Multi chip is supported, each chip adding its set of PHB4 controllers
and its PCI busses. The model doesn't emulate the EEH error handling.
This model is not ready for hotplug yet.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[ clg: - numerous cleanups
- commit log
- fix for broken LSI support
- PHB pic printinfo
- large QOM rework ]
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20200127144506.11132-2-clg@kaod.org>
[dwg: Use device_class_set_props()]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2020-01-27 17:45:05 +03:00
|
|
|
/* MSIs are supported on this platform */
|
|
|
|
msi_nonbroken = true;
|
|
|
|
|
2019-07-31 17:12:16 +03:00
|
|
|
/*
|
|
|
|
* Check compatibility of the specified CPU with the machine
|
|
|
|
* default.
|
|
|
|
*/
|
|
|
|
if (!pnv_match_cpu(mc->default_cpu_type, machine->cpu_type)) {
|
|
|
|
error_report("invalid CPU model '%s' for %s machine",
|
|
|
|
machine->cpu_type, mc->name);
|
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
|
2016-10-22 12:46:36 +03:00
|
|
|
/* Create the processor chips */
|
2017-10-09 22:51:06 +03:00
|
|
|
i = strlen(machine->cpu_type) - strlen(POWERPC_CPU_TYPE_SUFFIX);
|
2017-10-09 22:51:07 +03:00
|
|
|
chip_typename = g_strdup_printf(PNV_CHIP_TYPE_NAME("%.*s"),
|
2017-10-09 22:51:06 +03:00
|
|
|
i, machine->cpu_type);
|
2016-10-22 12:46:36 +03:00
|
|
|
if (!object_class_by_name(chip_typename)) {
|
2019-07-31 17:12:16 +03:00
|
|
|
error_report("invalid chip model '%.*s' for %s machine",
|
|
|
|
i, machine->cpu_type, mc->name);
|
2016-10-22 12:46:36 +03:00
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
|
2020-01-06 13:29:42 +03:00
|
|
|
pnv->num_chips =
|
|
|
|
machine->smp.max_cpus / (machine->smp.cores * machine->smp.threads);
|
|
|
|
/*
|
|
|
|
* TODO: should we decide on how many chips we can create based
|
|
|
|
* on #cores and Venice vs. Murano vs. Naples chip type etc...,
|
|
|
|
*/
|
2021-08-09 16:45:25 +03:00
|
|
|
if (!is_power_of_2(pnv->num_chips) || pnv->num_chips > 16) {
|
2020-01-06 13:29:42 +03:00
|
|
|
error_report("invalid number of chips: '%d'", pnv->num_chips);
|
2021-08-09 16:45:25 +03:00
|
|
|
error_printf(
|
|
|
|
"Try '-smp sockets=N'. Valid values are : 1, 2, 4, 8 and 16.\n");
|
2020-01-06 13:29:42 +03:00
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
|
2016-10-22 12:46:36 +03:00
|
|
|
pnv->chips = g_new0(PnvChip *, pnv->num_chips);
|
|
|
|
for (i = 0; i < pnv->num_chips; i++) {
|
|
|
|
char chip_name[32];
|
2020-06-10 08:31:59 +03:00
|
|
|
Object *chip = OBJECT(qdev_new(chip_typename));
|
2021-09-01 12:41:49 +03:00
|
|
|
uint64_t chip_ram_size = pnv_chip_get_ram_size(pnv, i);
|
2016-10-22 12:46:36 +03:00
|
|
|
|
|
|
|
pnv->chips[i] = PNV_CHIP(chip);
|
|
|
|
|
2021-08-09 16:45:26 +03:00
|
|
|
/* Distribute RAM among the chips */
|
|
|
|
object_property_set_int(chip, "ram-start", chip_ram_start,
|
|
|
|
&error_fatal);
|
|
|
|
object_property_set_int(chip, "ram-size", chip_ram_size,
|
|
|
|
&error_fatal);
|
|
|
|
chip_ram_start += chip_ram_size;
|
2016-10-22 12:46:36 +03:00
|
|
|
|
2021-09-01 12:41:49 +03:00
|
|
|
snprintf(chip_name, sizeof(chip_name), "chip[%d]", i);
|
qom: Drop parameter @errp of object_property_add() & friends
The only way object_property_add() can fail is when a property with
the same name already exists. Since our property names are all
hardcoded, failure is a programming error, and the appropriate way to
handle it is passing &error_abort.
Same for its variants, except for object_property_add_child(), which
additionally fails when the child already has a parent. Parentage is
also under program control, so this is a programming error, too.
We have a bit over 500 callers. Almost half of them pass
&error_abort, slightly fewer ignore errors, one test case handles
errors, and the remaining few callers pass them to their own callers.
The previous few commits demonstrated once again that ignoring
programming errors is a bad idea.
Of the few ones that pass on errors, several violate the Error API.
The Error ** argument must be NULL, &error_abort, &error_fatal, or a
pointer to a variable containing NULL. Passing an argument of the
latter kind twice without clearing it in between is wrong: if the
first call sets an error, it no longer points to NULL for the second
call. ich9_pm_add_properties(), sparc32_ledma_realize(),
sparc32_dma_realize(), xilinx_axidma_realize(), xilinx_enet_realize()
are wrong that way.
When the one appropriate choice of argument is &error_abort, letting
users pick the argument is a bad idea.
Drop parameter @errp and assert the preconditions instead.
There's one exception to "duplicate property name is a programming
error": the way object_property_add() implements the magic (and
undocumented) "automatic arrayification". Don't drop @errp there.
Instead, rename object_property_add() to object_property_try_add(),
and add the obvious wrapper object_property_add().
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200505152926.18877-15-armbru@redhat.com>
[Two semantic rebase conflicts resolved]
2020-05-05 18:29:22 +03:00
|
|
|
object_property_add_child(OBJECT(pnv), chip_name, chip);
|
2021-09-01 12:41:49 +03:00
|
|
|
object_property_set_int(chip, "chip-id", i, &error_fatal);
|
qom: Put name parameter before value / visitor parameter
The object_property_set_FOO() setters take property name and value in
an unusual order:
void object_property_set_FOO(Object *obj, FOO_TYPE value,
const char *name, Error **errp)
Having to pass value before name feels grating. Swap them.
Same for object_property_set(), object_property_get(), and
object_property_parse().
Convert callers with this Coccinelle script:
@@
identifier fun = {
object_property_get, object_property_parse, object_property_set_str,
object_property_set_link, object_property_set_bool,
object_property_set_int, object_property_set_uint, object_property_set,
object_property_set_qobject
};
expression obj, v, name, errp;
@@
- fun(obj, v, name, errp)
+ fun(obj, name, v, errp)
Chokes on hw/arm/musicpal.c's lcd_refresh() with the unhelpful error
message "no position information". Convert that one manually.
Fails to convert hw/arm/armsse.c, because Coccinelle gets confused by
ARMSSE being used both as typedef and function-like macro there.
Convert manually.
Fails to convert hw/rx/rx-gdbsim.c, because Coccinelle gets confused
by RXCPU being used both as typedef and function-like macro there.
Convert manually. The other files using RXCPU that way don't need
conversion.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20200707160613.848843-27-armbru@redhat.com>
[Straightforwad conflict with commit 2336172d9b "audio: set default
value for pcspk.iobase property" resolved]
2020-07-07 19:05:54 +03:00
|
|
|
object_property_set_int(chip, "nr-cores", machine->smp.cores,
|
|
|
|
&error_fatal);
|
|
|
|
object_property_set_int(chip, "nr-threads", machine->smp.threads,
|
2016-10-22 12:46:36 +03:00
|
|
|
&error_fatal);
|
2020-01-06 17:56:35 +03:00
|
|
|
/*
|
|
|
|
* The POWER8 machine use the XICS interrupt interface.
|
|
|
|
* Propagate the XICS fabric to the chip and its controllers.
|
|
|
|
*/
|
|
|
|
if (object_dynamic_cast(OBJECT(pnv), TYPE_XICS_FABRIC)) {
|
qom: Put name parameter before value / visitor parameter
The object_property_set_FOO() setters take property name and value in
an unusual order:
void object_property_set_FOO(Object *obj, FOO_TYPE value,
const char *name, Error **errp)
Having to pass value before name feels grating. Swap them.
Same for object_property_set(), object_property_get(), and
object_property_parse().
Convert callers with this Coccinelle script:
@@
identifier fun = {
object_property_get, object_property_parse, object_property_set_str,
object_property_set_link, object_property_set_bool,
object_property_set_int, object_property_set_uint, object_property_set,
object_property_set_qobject
};
expression obj, v, name, errp;
@@
- fun(obj, v, name, errp)
+ fun(obj, name, v, errp)
Chokes on hw/arm/musicpal.c's lcd_refresh() with the unhelpful error
message "no position information". Convert that one manually.
Fails to convert hw/arm/armsse.c, because Coccinelle gets confused by
ARMSSE being used both as typedef and function-like macro there.
Convert manually.
Fails to convert hw/rx/rx-gdbsim.c, because Coccinelle gets confused
by RXCPU being used both as typedef and function-like macro there.
Convert manually. The other files using RXCPU that way don't need
conversion.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20200707160613.848843-27-armbru@redhat.com>
[Straightforwad conflict with commit 2336172d9b "audio: set default
value for pcspk.iobase property" resolved]
2020-07-07 19:05:54 +03:00
|
|
|
object_property_set_link(chip, "xics", OBJECT(pnv), &error_abort);
|
2020-01-06 17:56:35 +03:00
|
|
|
}
|
2020-01-06 17:56:37 +03:00
|
|
|
if (object_dynamic_cast(OBJECT(pnv), TYPE_XIVE_FABRIC)) {
|
qom: Put name parameter before value / visitor parameter
The object_property_set_FOO() setters take property name and value in
an unusual order:
void object_property_set_FOO(Object *obj, FOO_TYPE value,
const char *name, Error **errp)
Having to pass value before name feels grating. Swap them.
Same for object_property_set(), object_property_get(), and
object_property_parse().
Convert callers with this Coccinelle script:
@@
identifier fun = {
object_property_get, object_property_parse, object_property_set_str,
object_property_set_link, object_property_set_bool,
object_property_set_int, object_property_set_uint, object_property_set,
object_property_set_qobject
};
expression obj, v, name, errp;
@@
- fun(obj, v, name, errp)
+ fun(obj, name, v, errp)
Chokes on hw/arm/musicpal.c's lcd_refresh() with the unhelpful error
message "no position information". Convert that one manually.
Fails to convert hw/arm/armsse.c, because Coccinelle gets confused by
ARMSSE being used both as typedef and function-like macro there.
Convert manually.
Fails to convert hw/rx/rx-gdbsim.c, because Coccinelle gets confused
by RXCPU being used both as typedef and function-like macro there.
Convert manually. The other files using RXCPU that way don't need
conversion.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20200707160613.848843-27-armbru@redhat.com>
[Straightforwad conflict with commit 2336172d9b "audio: set default
value for pcspk.iobase property" resolved]
2020-07-07 19:05:54 +03:00
|
|
|
object_property_set_link(chip, "xive-fabric", OBJECT(pnv),
|
2020-01-06 17:56:37 +03:00
|
|
|
&error_abort);
|
|
|
|
}
|
sysbus: Convert to sysbus_realize() etc. with Coccinelle
Convert from qdev_realize(), qdev_realize_and_unref() with null @bus
argument to sysbus_realize(), sysbus_realize_and_unref().
Coccinelle script:
@@
expression dev, errp;
@@
- qdev_realize(DEVICE(dev), NULL, errp);
+ sysbus_realize(SYS_BUS_DEVICE(dev), errp);
@@
expression sysbus_dev, dev, errp;
@@
+ sysbus_dev = SYS_BUS_DEVICE(dev);
- qdev_realize_and_unref(dev, NULL, errp);
+ sysbus_realize_and_unref(sysbus_dev, errp);
- sysbus_dev = SYS_BUS_DEVICE(dev);
@@
expression sysbus_dev, dev, errp;
expression expr;
@@
sysbus_dev = SYS_BUS_DEVICE(dev);
... when != dev = expr;
- qdev_realize_and_unref(dev, NULL, errp);
+ sysbus_realize_and_unref(sysbus_dev, errp);
@@
expression dev, errp;
@@
- qdev_realize_and_unref(DEVICE(dev), NULL, errp);
+ sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), errp);
@@
expression dev, errp;
@@
- qdev_realize_and_unref(dev, NULL, errp);
+ sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), errp);
Whitespace changes minimized manually.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200610053247.1583243-46-armbru@redhat.com>
[Conflicts in hw/misc/empty_slot.c and hw/sparc/leon3.c resolved]
2020-06-10 08:32:34 +03:00
|
|
|
sysbus_realize_and_unref(SYS_BUS_DEVICE(chip), &error_fatal);
|
2016-10-22 12:46:36 +03:00
|
|
|
}
|
|
|
|
g_free(chip_typename);
|
2016-10-22 12:46:43 +03:00
|
|
|
|
|
|
|
/* Instantiate ISA bus on chip 0 */
|
2018-06-15 18:25:34 +03:00
|
|
|
pnv->isa_bus = pnv_isa_create(pnv->chips[0], &error_fatal);
|
2016-10-22 12:46:43 +03:00
|
|
|
|
|
|
|
/* Create serial port */
|
2018-04-20 17:52:46 +03:00
|
|
|
serial_hds_isa_init(pnv->isa_bus, 0, MAX_ISA_SERIAL_PORTS);
|
2016-10-22 12:46:43 +03:00
|
|
|
|
|
|
|
/* Create an RTC ISA device too */
|
2017-10-17 19:44:16 +03:00
|
|
|
mc146818_rtc_init(pnv->isa_bus, 2000, NULL);
|
2017-04-11 18:30:06 +03:00
|
|
|
|
2020-04-04 18:36:55 +03:00
|
|
|
/*
|
|
|
|
* Create the machine BMC simulator and the IPMI BT device for
|
|
|
|
* communication with the BMC
|
|
|
|
*/
|
|
|
|
if (defaults_enabled()) {
|
|
|
|
pnv->bmc = pnv_bmc_create(pnv->pnor);
|
|
|
|
pnv_ipmi_bt_init(pnv->isa_bus, pnv->bmc, 10);
|
|
|
|
}
|
ppc/pnv: Create BMC devices at machine init
The BMC of the OpenPOWER systems monitors the machine state using
sensors, controls the power and controls the access to the PNOR flash
device containing the firmware image required to boot the host.
QEMU models the power cycle process, access to the sensors and access
to the PNOR device. But, for these features to be available, the QEMU
PowerNV machine needs two extras devices on the command line, an IPMI
BT device for communication and a BMC backend device:
-device ipmi-bmc-sim,id=bmc0 -device isa-ipmi-bt,bmc=bmc0,irq=10
The BMC properties are then defined accordingly in the device tree and
OPAL self adapts. If a BMC device and an IPMI BT device are not
available, OPAL does not try to communicate with the BMC in any
manner. This is not how real systems behave.
To be closer to the default behavior, create an IPMI BMC simulator
device and an IPMI BT device at machine initialization time. We loose
the ability to define an external BMC device but there are benefits:
- a better match with real systems,
- a better test coverage of the OPAL code,
- system powerdown and reset commands that work,
- a QEMU device tree compliant with the specifications (*).
(*) Still needs a MBOX device.
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20191121162340.11049-1-clg@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2019-11-21 19:23:40 +03:00
|
|
|
|
2021-01-26 20:10:59 +03:00
|
|
|
/*
|
|
|
|
* The PNOR is mapped on the LPC FW address space by the BMC.
|
|
|
|
* Since we can not reach the remote BMC machine with LPC memops,
|
|
|
|
* map it always for now.
|
|
|
|
*/
|
|
|
|
memory_region_add_subregion(pnv->chips[0]->fw_mr, PNOR_SPI_OFFSET,
|
|
|
|
&pnv->pnor->mmio);
|
|
|
|
|
2019-09-11 17:29:25 +03:00
|
|
|
/*
|
|
|
|
* OpenPOWER systems use a IPMI SEL Event message to notify the
|
|
|
|
* host to powerdown
|
|
|
|
*/
|
2017-04-11 18:30:06 +03:00
|
|
|
pnv->powerdown_notifier.notify = pnv_powerdown_notify;
|
|
|
|
qemu_register_powerdown_notifier(&pnv->powerdown_notifier);
|
2016-10-22 12:46:36 +03:00
|
|
|
}
|
|
|
|
|
2016-10-22 12:46:38 +03:00
|
|
|
/*
|
|
|
|
* 0:21 Reserved - Read as zeros
|
|
|
|
* 22:24 Chip ID
|
|
|
|
* 25:28 Core number
|
|
|
|
* 29:31 Thread ID
|
|
|
|
*/
|
|
|
|
static uint32_t pnv_chip_core_pir_p8(PnvChip *chip, uint32_t core_id)
|
|
|
|
{
|
|
|
|
return (chip->chip_id << 7) | (core_id << 3);
|
|
|
|
}
|
|
|
|
|
2019-01-02 08:57:34 +03:00
|
|
|
static void pnv_chip_power8_intc_create(PnvChip *chip, PowerPCCPU *cpu,
|
|
|
|
Error **errp)
|
2018-06-15 18:25:33 +03:00
|
|
|
{
|
2020-01-06 17:56:35 +03:00
|
|
|
Pnv8Chip *chip8 = PNV8_CHIP(chip);
|
2019-01-02 08:57:34 +03:00
|
|
|
Error *local_err = NULL;
|
|
|
|
Object *obj;
|
2019-01-17 10:53:25 +03:00
|
|
|
PnvCPUState *pnv_cpu = pnv_cpu_state(cpu);
|
2019-01-02 08:57:34 +03:00
|
|
|
|
2020-01-06 17:56:35 +03:00
|
|
|
obj = icp_create(OBJECT(cpu), TYPE_PNV_ICP, chip8->xics, &local_err);
|
2019-01-02 08:57:34 +03:00
|
|
|
if (local_err) {
|
|
|
|
error_propagate(errp, local_err);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2019-03-06 11:50:10 +03:00
|
|
|
pnv_cpu->intc = obj;
|
2018-06-15 18:25:33 +03:00
|
|
|
}
|
|
|
|
|
2019-10-24 17:27:22 +03:00
|
|
|
|
2019-10-22 19:38:10 +03:00
|
|
|
static void pnv_chip_power8_intc_reset(PnvChip *chip, PowerPCCPU *cpu)
|
|
|
|
{
|
|
|
|
PnvCPUState *pnv_cpu = pnv_cpu_state(cpu);
|
|
|
|
|
|
|
|
icp_reset(ICP(pnv_cpu->intc));
|
|
|
|
}
|
|
|
|
|
2019-10-24 17:27:22 +03:00
|
|
|
static void pnv_chip_power8_intc_destroy(PnvChip *chip, PowerPCCPU *cpu)
|
|
|
|
{
|
|
|
|
PnvCPUState *pnv_cpu = pnv_cpu_state(cpu);
|
|
|
|
|
|
|
|
icp_destroy(ICP(pnv_cpu->intc));
|
|
|
|
pnv_cpu->intc = NULL;
|
|
|
|
}
|
|
|
|
|
2019-12-13 15:00:07 +03:00
|
|
|
static void pnv_chip_power8_intc_print_info(PnvChip *chip, PowerPCCPU *cpu,
|
|
|
|
Monitor *mon)
|
|
|
|
{
|
|
|
|
icp_pic_print_info(ICP(pnv_cpu_state(cpu)->intc), mon);
|
|
|
|
}
|
|
|
|
|
2016-10-22 12:46:38 +03:00
|
|
|
/*
|
|
|
|
* 0:48 Reserved - Read as zeroes
|
|
|
|
* 49:52 Node ID
|
|
|
|
* 53:55 Chip ID
|
|
|
|
* 56 Reserved - Read as zero
|
|
|
|
* 57:61 Core number
|
|
|
|
* 62:63 Thread ID
|
|
|
|
*
|
|
|
|
* We only care about the lower bits. uint32_t is fine for the moment.
|
|
|
|
*/
|
|
|
|
static uint32_t pnv_chip_core_pir_p9(PnvChip *chip, uint32_t core_id)
|
|
|
|
{
|
|
|
|
return (chip->chip_id << 8) | (core_id << 2);
|
|
|
|
}
|
|
|
|
|
2019-12-05 21:44:51 +03:00
|
|
|
static uint32_t pnv_chip_core_pir_p10(PnvChip *chip, uint32_t core_id)
|
|
|
|
{
|
|
|
|
return (chip->chip_id << 8) | (core_id << 2);
|
|
|
|
}
|
|
|
|
|
2019-01-02 08:57:34 +03:00
|
|
|
static void pnv_chip_power9_intc_create(PnvChip *chip, PowerPCCPU *cpu,
|
|
|
|
Error **errp)
|
2018-06-15 18:25:33 +03:00
|
|
|
{
|
2019-03-06 11:50:11 +03:00
|
|
|
Pnv9Chip *chip9 = PNV9_CHIP(chip);
|
|
|
|
Error *local_err = NULL;
|
|
|
|
Object *obj;
|
|
|
|
PnvCPUState *pnv_cpu = pnv_cpu_state(cpu);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The core creates its interrupt presenter but the XIVE interrupt
|
|
|
|
* controller object is initialized afterwards. Hopefully, it's
|
|
|
|
* only used at runtime.
|
|
|
|
*/
|
2020-01-06 17:56:41 +03:00
|
|
|
obj = xive_tctx_create(OBJECT(cpu), XIVE_PRESENTER(&chip9->xive),
|
|
|
|
&local_err);
|
2019-03-06 11:50:11 +03:00
|
|
|
if (local_err) {
|
|
|
|
error_propagate(errp, local_err);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
pnv_cpu->intc = obj;
|
2018-06-15 18:25:33 +03:00
|
|
|
}
|
|
|
|
|
2019-10-22 19:38:10 +03:00
|
|
|
static void pnv_chip_power9_intc_reset(PnvChip *chip, PowerPCCPU *cpu)
|
|
|
|
{
|
|
|
|
PnvCPUState *pnv_cpu = pnv_cpu_state(cpu);
|
|
|
|
|
|
|
|
xive_tctx_reset(XIVE_TCTX(pnv_cpu->intc));
|
|
|
|
}
|
|
|
|
|
2019-10-24 17:27:22 +03:00
|
|
|
static void pnv_chip_power9_intc_destroy(PnvChip *chip, PowerPCCPU *cpu)
|
|
|
|
{
|
|
|
|
PnvCPUState *pnv_cpu = pnv_cpu_state(cpu);
|
|
|
|
|
|
|
|
xive_tctx_destroy(XIVE_TCTX(pnv_cpu->intc));
|
|
|
|
pnv_cpu->intc = NULL;
|
|
|
|
}
|
|
|
|
|
2019-12-13 15:00:07 +03:00
|
|
|
static void pnv_chip_power9_intc_print_info(PnvChip *chip, PowerPCCPU *cpu,
|
|
|
|
Monitor *mon)
|
|
|
|
{
|
|
|
|
xive_tctx_pic_print_info(XIVE_TCTX(pnv_cpu_state(cpu)->intc), mon);
|
|
|
|
}
|
|
|
|
|
2019-12-05 21:44:51 +03:00
|
|
|
static void pnv_chip_power10_intc_create(PnvChip *chip, PowerPCCPU *cpu,
|
|
|
|
Error **errp)
|
|
|
|
{
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 08:51:38 +03:00
|
|
|
Pnv10Chip *chip10 = PNV10_CHIP(chip);
|
|
|
|
Error *local_err = NULL;
|
|
|
|
Object *obj;
|
2019-12-05 21:44:51 +03:00
|
|
|
PnvCPUState *pnv_cpu = pnv_cpu_state(cpu);
|
|
|
|
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 08:51:38 +03:00
|
|
|
/*
|
|
|
|
* The core creates its interrupt presenter but the XIVE2 interrupt
|
|
|
|
* controller object is initialized afterwards. Hopefully, it's
|
|
|
|
* only used at runtime.
|
|
|
|
*/
|
|
|
|
obj = xive_tctx_create(OBJECT(cpu), XIVE_PRESENTER(&chip10->xive),
|
|
|
|
&local_err);
|
|
|
|
if (local_err) {
|
|
|
|
error_propagate(errp, local_err);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
pnv_cpu->intc = obj;
|
2019-12-05 21:44:51 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_chip_power10_intc_reset(PnvChip *chip, PowerPCCPU *cpu)
|
|
|
|
{
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 08:51:38 +03:00
|
|
|
PnvCPUState *pnv_cpu = pnv_cpu_state(cpu);
|
|
|
|
|
|
|
|
xive_tctx_reset(XIVE_TCTX(pnv_cpu->intc));
|
2019-12-05 21:44:51 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_chip_power10_intc_destroy(PnvChip *chip, PowerPCCPU *cpu)
|
|
|
|
{
|
|
|
|
PnvCPUState *pnv_cpu = pnv_cpu_state(cpu);
|
|
|
|
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 08:51:38 +03:00
|
|
|
xive_tctx_destroy(XIVE_TCTX(pnv_cpu->intc));
|
2019-12-05 21:44:51 +03:00
|
|
|
pnv_cpu->intc = NULL;
|
|
|
|
}
|
|
|
|
|
2019-12-13 15:00:07 +03:00
|
|
|
static void pnv_chip_power10_intc_print_info(PnvChip *chip, PowerPCCPU *cpu,
|
|
|
|
Monitor *mon)
|
|
|
|
{
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 08:51:38 +03:00
|
|
|
xive_tctx_pic_print_info(XIVE_TCTX(pnv_cpu_state(cpu)->intc), mon);
|
2019-12-13 15:00:07 +03:00
|
|
|
}
|
|
|
|
|
2019-09-11 17:29:25 +03:00
|
|
|
/*
|
|
|
|
* Allowed core identifiers on a POWER8 Processor Chip :
|
2016-10-22 12:46:37 +03:00
|
|
|
*
|
|
|
|
* <EX0 reserved>
|
|
|
|
* EX1 - Venice only
|
|
|
|
* EX2 - Venice only
|
|
|
|
* EX3 - Venice only
|
|
|
|
* EX4
|
|
|
|
* EX5
|
|
|
|
* EX6
|
|
|
|
* <EX7,8 reserved> <reserved>
|
|
|
|
* EX9 - Venice only
|
|
|
|
* EX10 - Venice only
|
|
|
|
* EX11 - Venice only
|
|
|
|
* EX12
|
|
|
|
* EX13
|
|
|
|
* EX14
|
|
|
|
* <EX15 reserved>
|
|
|
|
*/
|
|
|
|
#define POWER8E_CORE_MASK (0x7070ull)
|
|
|
|
#define POWER8_CORE_MASK (0x7e7eull)
|
|
|
|
|
|
|
|
/*
|
2018-01-15 21:04:02 +03:00
|
|
|
* POWER9 has 24 cores, ids starting at 0x0
|
2016-10-22 12:46:37 +03:00
|
|
|
*/
|
2018-01-15 21:04:02 +03:00
|
|
|
#define POWER9_CORE_MASK (0xffffffffffffffull)
|
2016-10-22 12:46:37 +03:00
|
|
|
|
2019-12-05 21:44:51 +03:00
|
|
|
|
|
|
|
#define POWER10_CORE_MASK (0xffffffffffffffull)
|
|
|
|
|
2018-06-18 20:05:39 +03:00
|
|
|
static void pnv_chip_power8_instance_init(Object *obj)
|
|
|
|
{
|
|
|
|
Pnv8Chip *chip8 = PNV8_CHIP(obj);
|
2020-01-27 17:45:06 +03:00
|
|
|
PnvChipClass *pcc = PNV_CHIP_GET_CLASS(obj);
|
|
|
|
int i;
|
2018-06-18 20:05:39 +03:00
|
|
|
|
2020-01-06 17:56:35 +03:00
|
|
|
object_property_add_link(obj, "xics", TYPE_XICS_FABRIC,
|
|
|
|
(Object **)&chip8->xics,
|
|
|
|
object_property_allow_set_link,
|
qom: Drop parameter @errp of object_property_add() & friends
The only way object_property_add() can fail is when a property with
the same name already exists. Since our property names are all
hardcoded, failure is a programming error, and the appropriate way to
handle it is passing &error_abort.
Same for its variants, except for object_property_add_child(), which
additionally fails when the child already has a parent. Parentage is
also under program control, so this is a programming error, too.
We have a bit over 500 callers. Almost half of them pass
&error_abort, slightly fewer ignore errors, one test case handles
errors, and the remaining few callers pass them to their own callers.
The previous few commits demonstrated once again that ignoring
programming errors is a bad idea.
Of the few ones that pass on errors, several violate the Error API.
The Error ** argument must be NULL, &error_abort, &error_fatal, or a
pointer to a variable containing NULL. Passing an argument of the
latter kind twice without clearing it in between is wrong: if the
first call sets an error, it no longer points to NULL for the second
call. ich9_pm_add_properties(), sparc32_ledma_realize(),
sparc32_dma_realize(), xilinx_axidma_realize(), xilinx_enet_realize()
are wrong that way.
When the one appropriate choice of argument is &error_abort, letting
users pick the argument is a bad idea.
Drop parameter @errp and assert the preconditions instead.
There's one exception to "duplicate property name is a programming
error": the way object_property_add() implements the magic (and
undocumented) "automatic arrayification". Don't drop @errp there.
Instead, rename object_property_add() to object_property_try_add(),
and add the obvious wrapper object_property_add().
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200505152926.18877-15-armbru@redhat.com>
[Two semantic rebase conflicts resolved]
2020-05-05 18:29:22 +03:00
|
|
|
OBJ_PROP_LINK_STRONG);
|
2020-01-06 17:56:35 +03:00
|
|
|
|
qom: Less verbose object_initialize_child()
All users of object_initialize_child() pass the obvious child size
argument. Almost all pass &error_abort and no properties. Tiresome.
Rename object_initialize_child() to
object_initialize_child_with_props() to free the name. New
convenience wrapper object_initialize_child() automates the size
argument, and passes &error_abort and no properties.
Rename object_initialize_childv() to
object_initialize_child_with_propsv() for consistency.
Convert callers with this Coccinelle script:
@@
expression parent, propname, type;
expression child, size;
symbol error_abort;
@@
- object_initialize_child(parent, propname, OBJECT(child), size, type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, size, type, &error_abort, NULL)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, child, sizeof(*child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, type)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, &child, sizeof(child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, &child, type)
@@
expression parent, propname, type;
expression child, size, err;
expression list props;
@@
- object_initialize_child(parent, propname, child, size, type, err, props)
+ object_initialize_child_with_props(parent, propname, child, size, type, err, props)
Note that Coccinelle chokes on ARMSSE typedef vs. macro in
hw/arm/armsse.c. Worked around by temporarily renaming the macro for
the spatch run.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
[Rebased: machine opentitan is new (commit fe0fe4735e7)]
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200610053247.1583243-37-armbru@redhat.com>
2020-06-10 08:32:25 +03:00
|
|
|
object_initialize_child(obj, "psi", &chip8->psi, TYPE_PNV8_PSI);
|
2018-06-18 20:05:39 +03:00
|
|
|
|
qom: Less verbose object_initialize_child()
All users of object_initialize_child() pass the obvious child size
argument. Almost all pass &error_abort and no properties. Tiresome.
Rename object_initialize_child() to
object_initialize_child_with_props() to free the name. New
convenience wrapper object_initialize_child() automates the size
argument, and passes &error_abort and no properties.
Rename object_initialize_childv() to
object_initialize_child_with_propsv() for consistency.
Convert callers with this Coccinelle script:
@@
expression parent, propname, type;
expression child, size;
symbol error_abort;
@@
- object_initialize_child(parent, propname, OBJECT(child), size, type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, size, type, &error_abort, NULL)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, child, sizeof(*child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, type)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, &child, sizeof(child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, &child, type)
@@
expression parent, propname, type;
expression child, size, err;
expression list props;
@@
- object_initialize_child(parent, propname, child, size, type, err, props)
+ object_initialize_child_with_props(parent, propname, child, size, type, err, props)
Note that Coccinelle chokes on ARMSSE typedef vs. macro in
hw/arm/armsse.c. Worked around by temporarily renaming the macro for
the spatch run.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
[Rebased: machine opentitan is new (commit fe0fe4735e7)]
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200610053247.1583243-37-armbru@redhat.com>
2020-06-10 08:32:25 +03:00
|
|
|
object_initialize_child(obj, "lpc", &chip8->lpc, TYPE_PNV8_LPC);
|
2018-06-18 20:05:39 +03:00
|
|
|
|
qom: Less verbose object_initialize_child()
All users of object_initialize_child() pass the obvious child size
argument. Almost all pass &error_abort and no properties. Tiresome.
Rename object_initialize_child() to
object_initialize_child_with_props() to free the name. New
convenience wrapper object_initialize_child() automates the size
argument, and passes &error_abort and no properties.
Rename object_initialize_childv() to
object_initialize_child_with_propsv() for consistency.
Convert callers with this Coccinelle script:
@@
expression parent, propname, type;
expression child, size;
symbol error_abort;
@@
- object_initialize_child(parent, propname, OBJECT(child), size, type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, size, type, &error_abort, NULL)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, child, sizeof(*child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, type)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, &child, sizeof(child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, &child, type)
@@
expression parent, propname, type;
expression child, size, err;
expression list props;
@@
- object_initialize_child(parent, propname, child, size, type, err, props)
+ object_initialize_child_with_props(parent, propname, child, size, type, err, props)
Note that Coccinelle chokes on ARMSSE typedef vs. macro in
hw/arm/armsse.c. Worked around by temporarily renaming the macro for
the spatch run.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
[Rebased: machine opentitan is new (commit fe0fe4735e7)]
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200610053247.1583243-37-armbru@redhat.com>
2020-06-10 08:32:25 +03:00
|
|
|
object_initialize_child(obj, "occ", &chip8->occ, TYPE_PNV8_OCC);
|
2019-09-12 12:30:54 +03:00
|
|
|
|
qom: Less verbose object_initialize_child()
All users of object_initialize_child() pass the obvious child size
argument. Almost all pass &error_abort and no properties. Tiresome.
Rename object_initialize_child() to
object_initialize_child_with_props() to free the name. New
convenience wrapper object_initialize_child() automates the size
argument, and passes &error_abort and no properties.
Rename object_initialize_childv() to
object_initialize_child_with_propsv() for consistency.
Convert callers with this Coccinelle script:
@@
expression parent, propname, type;
expression child, size;
symbol error_abort;
@@
- object_initialize_child(parent, propname, OBJECT(child), size, type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, size, type, &error_abort, NULL)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, child, sizeof(*child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, type)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, &child, sizeof(child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, &child, type)
@@
expression parent, propname, type;
expression child, size, err;
expression list props;
@@
- object_initialize_child(parent, propname, child, size, type, err, props)
+ object_initialize_child_with_props(parent, propname, child, size, type, err, props)
Note that Coccinelle chokes on ARMSSE typedef vs. macro in
hw/arm/armsse.c. Worked around by temporarily renaming the macro for
the spatch run.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
[Rebased: machine opentitan is new (commit fe0fe4735e7)]
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200610053247.1583243-37-armbru@redhat.com>
2020-06-10 08:32:25 +03:00
|
|
|
object_initialize_child(obj, "homer", &chip8->homer, TYPE_PNV8_HOMER);
|
2020-01-27 17:45:06 +03:00
|
|
|
|
2022-08-11 19:39:45 +03:00
|
|
|
if (defaults_enabled()) {
|
|
|
|
chip8->num_phbs = pcc->num_phbs;
|
|
|
|
|
|
|
|
for (i = 0; i < chip8->num_phbs; i++) {
|
|
|
|
Object *phb = object_new(TYPE_PNV_PHB);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We need the chip to parent the PHB to allow the DT
|
|
|
|
* to build correctly (via pnv_xscom_dt()).
|
|
|
|
*
|
|
|
|
* TODO: the PHB should be parented by a PEC device that, at
|
|
|
|
* this moment, is not modelled powernv8/phb3.
|
|
|
|
*/
|
|
|
|
object_property_add_child(obj, "phb[*]", phb);
|
|
|
|
chip8->phbs[i] = PNV_PHB(phb);
|
|
|
|
}
|
2020-01-27 17:45:06 +03:00
|
|
|
}
|
|
|
|
|
2018-06-18 20:05:39 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_chip_icp_realize(Pnv8Chip *chip8, Error **errp)
|
|
|
|
{
|
|
|
|
PnvChip *chip = PNV_CHIP(chip8);
|
|
|
|
PnvChipClass *pcc = PNV_CHIP_GET_CLASS(chip);
|
|
|
|
int i, j;
|
|
|
|
char *name;
|
|
|
|
|
|
|
|
name = g_strdup_printf("icp-%x", chip->chip_id);
|
|
|
|
memory_region_init(&chip8->icp_mmio, OBJECT(chip), name, PNV_ICP_SIZE);
|
|
|
|
sysbus_init_mmio(SYS_BUS_DEVICE(chip), &chip8->icp_mmio);
|
|
|
|
g_free(name);
|
|
|
|
|
|
|
|
sysbus_mmio_map(SYS_BUS_DEVICE(chip), 1, PNV_ICP_BASE(chip));
|
|
|
|
|
|
|
|
/* Map the ICP registers for each thread */
|
|
|
|
for (i = 0; i < chip->nr_cores; i++) {
|
2019-11-25 09:58:03 +03:00
|
|
|
PnvCore *pnv_core = chip->cores[i];
|
2018-06-18 20:05:39 +03:00
|
|
|
int core_hwid = CPU_CORE(pnv_core)->core_id;
|
|
|
|
|
|
|
|
for (j = 0; j < CPU_CORE(pnv_core)->nr_threads; j++) {
|
|
|
|
uint32_t pir = pcc->core_pir(chip, core_hwid) + j;
|
2020-01-06 17:56:35 +03:00
|
|
|
PnvICPState *icp = PNV_ICP(xics_icp_get(chip8->xics, pir));
|
2018-06-18 20:05:39 +03:00
|
|
|
|
|
|
|
memory_region_add_subregion(&chip8->icp_mmio, pir << 12,
|
|
|
|
&icp->mmio);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_chip_power8_realize(DeviceState *dev, Error **errp)
|
|
|
|
{
|
|
|
|
PnvChipClass *pcc = PNV_CHIP_GET_CLASS(dev);
|
|
|
|
PnvChip *chip = PNV_CHIP(dev);
|
|
|
|
Pnv8Chip *chip8 = PNV8_CHIP(dev);
|
2019-03-08 01:35:34 +03:00
|
|
|
Pnv8Psi *psi8 = &chip8->psi;
|
2018-06-18 20:05:39 +03:00
|
|
|
Error *local_err = NULL;
|
2020-01-27 17:45:06 +03:00
|
|
|
int i;
|
2018-06-18 20:05:39 +03:00
|
|
|
|
2020-01-06 17:56:35 +03:00
|
|
|
assert(chip8->xics);
|
|
|
|
|
2019-06-12 20:43:44 +03:00
|
|
|
/* XSCOM bridge is first */
|
|
|
|
pnv_xscom_realize(chip, PNV_XSCOM_SIZE, &local_err);
|
|
|
|
if (local_err) {
|
|
|
|
error_propagate(errp, local_err);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
sysbus_mmio_map(SYS_BUS_DEVICE(chip), 0, PNV_XSCOM_BASE(chip));
|
|
|
|
|
2018-06-18 20:05:39 +03:00
|
|
|
pcc->parent_realize(dev, &local_err);
|
|
|
|
if (local_err) {
|
|
|
|
error_propagate(errp, local_err);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Processor Service Interface (PSI) Host Bridge */
|
qom: Put name parameter before value / visitor parameter
The object_property_set_FOO() setters take property name and value in
an unusual order:
void object_property_set_FOO(Object *obj, FOO_TYPE value,
const char *name, Error **errp)
Having to pass value before name feels grating. Swap them.
Same for object_property_set(), object_property_get(), and
object_property_parse().
Convert callers with this Coccinelle script:
@@
identifier fun = {
object_property_get, object_property_parse, object_property_set_str,
object_property_set_link, object_property_set_bool,
object_property_set_int, object_property_set_uint, object_property_set,
object_property_set_qobject
};
expression obj, v, name, errp;
@@
- fun(obj, v, name, errp)
+ fun(obj, name, v, errp)
Chokes on hw/arm/musicpal.c's lcd_refresh() with the unhelpful error
message "no position information". Convert that one manually.
Fails to convert hw/arm/armsse.c, because Coccinelle gets confused by
ARMSSE being used both as typedef and function-like macro there.
Convert manually.
Fails to convert hw/rx/rx-gdbsim.c, because Coccinelle gets confused
by RXCPU being used both as typedef and function-like macro there.
Convert manually. The other files using RXCPU that way don't need
conversion.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20200707160613.848843-27-armbru@redhat.com>
[Straightforwad conflict with commit 2336172d9b "audio: set default
value for pcspk.iobase property" resolved]
2020-07-07 19:05:54 +03:00
|
|
|
object_property_set_int(OBJECT(&chip8->psi), "bar", PNV_PSIHB_BASE(chip),
|
|
|
|
&error_fatal);
|
|
|
|
object_property_set_link(OBJECT(&chip8->psi), ICS_PROP_XICS,
|
|
|
|
OBJECT(chip8->xics), &error_abort);
|
error: Eliminate error_propagate() with Coccinelle, part 1
When all we do with an Error we receive into a local variable is
propagating to somewhere else, we can just as well receive it there
right away. Convert
if (!foo(..., &err)) {
...
error_propagate(errp, err);
...
return ...
}
to
if (!foo(..., errp)) {
...
...
return ...
}
where nothing else needs @err. Coccinelle script:
@rule1 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
binary operator op;
constant c1, c2;
symbol false;
@@
if (
(
- fun(args, &err, args2)
+ fun(args, errp, args2)
|
- !fun(args, &err, args2)
+ !fun(args, errp, args2)
|
- fun(args, &err, args2) op c1
+ fun(args, errp, args2) op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
)
}
@rule2 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
expression var;
binary operator op;
constant c1, c2;
symbol false;
@@
- var = fun(args, &err, args2);
+ var = fun(args, errp, args2);
... when != err
if (
(
var
|
!var
|
var op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
|
return var;
)
}
@depends on rule1 || rule2@
identifier err;
@@
- Error *err = NULL;
... when != err
Not exactly elegant, I'm afraid.
The "when != lbl:" is necessary to avoid transforming
if (fun(args, &err)) {
goto out
}
...
out:
error_propagate(errp, err);
even though other paths to label out still need the error_propagate().
For an actual example, see sclp_realize().
Without the "when strict", Coccinelle transforms vfio_msix_setup(),
incorrectly. I don't know what exactly "when strict" does, only that
it helps here.
The match of return is narrower than what I want, but I can't figure
out how to express "return where the operand doesn't use @err". For
an example where it's too narrow, see vfio_intx_enable().
Silently fails to convert hw/arm/armsse.c, because Coccinelle gets
confused by ARMSSE being used both as typedef and function-like macro
there. Converted manually.
Line breaks tidied up manually. One nested declaration of @local_err
deleted manually. Preexisting unwanted blank line dropped in
hw/riscv/sifive_e.c.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200707160613.848843-35-armbru@redhat.com>
2020-07-07 19:06:02 +03:00
|
|
|
if (!qdev_realize(DEVICE(&chip8->psi), NULL, errp)) {
|
2018-06-18 20:05:39 +03:00
|
|
|
return;
|
|
|
|
}
|
2019-03-08 01:35:34 +03:00
|
|
|
pnv_xscom_add_subregion(chip, PNV_XSCOM_PSIHB_BASE,
|
|
|
|
&PNV_PSI(psi8)->xscom_regs);
|
2018-06-18 20:05:39 +03:00
|
|
|
|
|
|
|
/* Create LPC controller */
|
qdev: Convert bus-less devices to qdev_realize() with Coccinelle
All remaining conversions to qdev_realize() are for bus-less devices.
Coccinelle script:
// only correct for bus-less @dev!
@@
expression errp;
expression dev;
@@
- qdev_init_nofail(dev);
+ qdev_realize(dev, NULL, &error_fatal);
@ depends on !(file in "hw/core/qdev.c") && !(file in "hw/core/bus.c")@
expression errp;
expression dev;
symbol true;
@@
- object_property_set_bool(OBJECT(dev), true, "realized", errp);
+ qdev_realize(DEVICE(dev), NULL, errp);
@ depends on !(file in "hw/core/qdev.c") && !(file in "hw/core/bus.c")@
expression errp;
expression dev;
symbol true;
@@
- object_property_set_bool(dev, true, "realized", errp);
+ qdev_realize(DEVICE(dev), NULL, errp);
Note that Coccinelle chokes on ARMSSE typedef vs. macro in
hw/arm/armsse.c. Worked around by temporarily renaming the macro for
the spatch run.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200610053247.1583243-57-armbru@redhat.com>
2020-06-10 08:32:45 +03:00
|
|
|
qdev_realize(DEVICE(&chip8->lpc), NULL, &error_fatal);
|
2018-06-18 20:05:39 +03:00
|
|
|
pnv_xscom_add_subregion(chip, PNV_XSCOM_LPC_BASE, &chip8->lpc.xscom_regs);
|
|
|
|
|
2021-01-26 20:10:59 +03:00
|
|
|
chip->fw_mr = &chip8->lpc.isa_fw;
|
2019-03-08 01:35:38 +03:00
|
|
|
chip->dt_isa_nodename = g_strdup_printf("/xscom@%" PRIx64 "/isa@%x",
|
|
|
|
(uint64_t) PNV_XSCOM_BASE(chip),
|
|
|
|
PNV_XSCOM_LPC_BASE);
|
|
|
|
|
2019-09-11 17:29:25 +03:00
|
|
|
/*
|
|
|
|
* Interrupt Management Area. This is the memory region holding
|
|
|
|
* all the Interrupt Control Presenter (ICP) registers
|
|
|
|
*/
|
2018-06-18 20:05:39 +03:00
|
|
|
pnv_chip_icp_realize(chip8, &local_err);
|
|
|
|
if (local_err) {
|
|
|
|
error_propagate(errp, local_err);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Create the simplified OCC model */
|
error: Eliminate error_propagate() with Coccinelle, part 1
When all we do with an Error we receive into a local variable is
propagating to somewhere else, we can just as well receive it there
right away. Convert
if (!foo(..., &err)) {
...
error_propagate(errp, err);
...
return ...
}
to
if (!foo(..., errp)) {
...
...
return ...
}
where nothing else needs @err. Coccinelle script:
@rule1 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
binary operator op;
constant c1, c2;
symbol false;
@@
if (
(
- fun(args, &err, args2)
+ fun(args, errp, args2)
|
- !fun(args, &err, args2)
+ !fun(args, errp, args2)
|
- fun(args, &err, args2) op c1
+ fun(args, errp, args2) op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
)
}
@rule2 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
expression var;
binary operator op;
constant c1, c2;
symbol false;
@@
- var = fun(args, &err, args2);
+ var = fun(args, errp, args2);
... when != err
if (
(
var
|
!var
|
var op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
|
return var;
)
}
@depends on rule1 || rule2@
identifier err;
@@
- Error *err = NULL;
... when != err
Not exactly elegant, I'm afraid.
The "when != lbl:" is necessary to avoid transforming
if (fun(args, &err)) {
goto out
}
...
out:
error_propagate(errp, err);
even though other paths to label out still need the error_propagate().
For an actual example, see sclp_realize().
Without the "when strict", Coccinelle transforms vfio_msix_setup(),
incorrectly. I don't know what exactly "when strict" does, only that
it helps here.
The match of return is narrower than what I want, but I can't figure
out how to express "return where the operand doesn't use @err". For
an example where it's too narrow, see vfio_intx_enable().
Silently fails to convert hw/arm/armsse.c, because Coccinelle gets
confused by ARMSSE being used both as typedef and function-like macro
there. Converted manually.
Line breaks tidied up manually. One nested declaration of @local_err
deleted manually. Preexisting unwanted blank line dropped in
hw/riscv/sifive_e.c.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200707160613.848843-35-armbru@redhat.com>
2020-07-07 19:06:02 +03:00
|
|
|
if (!qdev_realize(DEVICE(&chip8->occ), NULL, errp)) {
|
2018-06-18 20:05:39 +03:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
pnv_xscom_add_subregion(chip, PNV_XSCOM_OCC_BASE, &chip8->occ.xscom_regs);
|
2022-03-23 10:28:44 +03:00
|
|
|
qdev_connect_gpio_out(DEVICE(&chip8->occ), 0,
|
|
|
|
qdev_get_gpio_in(DEVICE(&chip8->psi), PSIHB_IRQ_OCC));
|
2019-09-12 12:30:53 +03:00
|
|
|
|
|
|
|
/* OCC SRAM model */
|
2019-12-11 11:29:12 +03:00
|
|
|
memory_region_add_subregion(get_system_memory(), PNV_OCC_SENSOR_BASE(chip),
|
2019-09-12 12:30:53 +03:00
|
|
|
&chip8->occ.sram_regs);
|
2019-09-12 12:30:54 +03:00
|
|
|
|
|
|
|
/* HOMER */
|
qom: Put name parameter before value / visitor parameter
The object_property_set_FOO() setters take property name and value in
an unusual order:
void object_property_set_FOO(Object *obj, FOO_TYPE value,
const char *name, Error **errp)
Having to pass value before name feels grating. Swap them.
Same for object_property_set(), object_property_get(), and
object_property_parse().
Convert callers with this Coccinelle script:
@@
identifier fun = {
object_property_get, object_property_parse, object_property_set_str,
object_property_set_link, object_property_set_bool,
object_property_set_int, object_property_set_uint, object_property_set,
object_property_set_qobject
};
expression obj, v, name, errp;
@@
- fun(obj, v, name, errp)
+ fun(obj, name, v, errp)
Chokes on hw/arm/musicpal.c's lcd_refresh() with the unhelpful error
message "no position information". Convert that one manually.
Fails to convert hw/arm/armsse.c, because Coccinelle gets confused by
ARMSSE being used both as typedef and function-like macro there.
Convert manually.
Fails to convert hw/rx/rx-gdbsim.c, because Coccinelle gets confused
by RXCPU being used both as typedef and function-like macro there.
Convert manually. The other files using RXCPU that way don't need
conversion.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20200707160613.848843-27-armbru@redhat.com>
[Straightforwad conflict with commit 2336172d9b "audio: set default
value for pcspk.iobase property" resolved]
2020-07-07 19:05:54 +03:00
|
|
|
object_property_set_link(OBJECT(&chip8->homer), "chip", OBJECT(chip),
|
2019-11-15 18:55:54 +03:00
|
|
|
&error_abort);
|
error: Eliminate error_propagate() with Coccinelle, part 1
When all we do with an Error we receive into a local variable is
propagating to somewhere else, we can just as well receive it there
right away. Convert
if (!foo(..., &err)) {
...
error_propagate(errp, err);
...
return ...
}
to
if (!foo(..., errp)) {
...
...
return ...
}
where nothing else needs @err. Coccinelle script:
@rule1 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
binary operator op;
constant c1, c2;
symbol false;
@@
if (
(
- fun(args, &err, args2)
+ fun(args, errp, args2)
|
- !fun(args, &err, args2)
+ !fun(args, errp, args2)
|
- fun(args, &err, args2) op c1
+ fun(args, errp, args2) op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
)
}
@rule2 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
expression var;
binary operator op;
constant c1, c2;
symbol false;
@@
- var = fun(args, &err, args2);
+ var = fun(args, errp, args2);
... when != err
if (
(
var
|
!var
|
var op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
|
return var;
)
}
@depends on rule1 || rule2@
identifier err;
@@
- Error *err = NULL;
... when != err
Not exactly elegant, I'm afraid.
The "when != lbl:" is necessary to avoid transforming
if (fun(args, &err)) {
goto out
}
...
out:
error_propagate(errp, err);
even though other paths to label out still need the error_propagate().
For an actual example, see sclp_realize().
Without the "when strict", Coccinelle transforms vfio_msix_setup(),
incorrectly. I don't know what exactly "when strict" does, only that
it helps here.
The match of return is narrower than what I want, but I can't figure
out how to express "return where the operand doesn't use @err". For
an example where it's too narrow, see vfio_intx_enable().
Silently fails to convert hw/arm/armsse.c, because Coccinelle gets
confused by ARMSSE being used both as typedef and function-like macro
there. Converted manually.
Line breaks tidied up manually. One nested declaration of @local_err
deleted manually. Preexisting unwanted blank line dropped in
hw/riscv/sifive_e.c.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200707160613.848843-35-armbru@redhat.com>
2020-07-07 19:06:02 +03:00
|
|
|
if (!qdev_realize(DEVICE(&chip8->homer), NULL, errp)) {
|
2019-09-12 12:30:54 +03:00
|
|
|
return;
|
|
|
|
}
|
2019-12-11 11:29:11 +03:00
|
|
|
/* Homer Xscom region */
|
|
|
|
pnv_xscom_add_subregion(chip, PNV_XSCOM_PBA_BASE, &chip8->homer.pba_regs);
|
|
|
|
|
|
|
|
/* Homer mmio region */
|
2019-09-12 12:30:54 +03:00
|
|
|
memory_region_add_subregion(get_system_memory(), PNV_HOMER_BASE(chip),
|
|
|
|
&chip8->homer.regs);
|
2020-01-27 17:45:06 +03:00
|
|
|
|
2022-06-24 11:49:12 +03:00
|
|
|
/* PHB controllers */
|
2022-01-12 13:28:27 +03:00
|
|
|
for (i = 0; i < chip8->num_phbs; i++) {
|
2022-08-11 19:39:44 +03:00
|
|
|
PnvPHB *phb = chip8->phbs[i];
|
2020-01-27 17:45:06 +03:00
|
|
|
|
qom: Put name parameter before value / visitor parameter
The object_property_set_FOO() setters take property name and value in
an unusual order:
void object_property_set_FOO(Object *obj, FOO_TYPE value,
const char *name, Error **errp)
Having to pass value before name feels grating. Swap them.
Same for object_property_set(), object_property_get(), and
object_property_parse().
Convert callers with this Coccinelle script:
@@
identifier fun = {
object_property_get, object_property_parse, object_property_set_str,
object_property_set_link, object_property_set_bool,
object_property_set_int, object_property_set_uint, object_property_set,
object_property_set_qobject
};
expression obj, v, name, errp;
@@
- fun(obj, v, name, errp)
+ fun(obj, name, v, errp)
Chokes on hw/arm/musicpal.c's lcd_refresh() with the unhelpful error
message "no position information". Convert that one manually.
Fails to convert hw/arm/armsse.c, because Coccinelle gets confused by
ARMSSE being used both as typedef and function-like macro there.
Convert manually.
Fails to convert hw/rx/rx-gdbsim.c, because Coccinelle gets confused
by RXCPU being used both as typedef and function-like macro there.
Convert manually. The other files using RXCPU that way don't need
conversion.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20200707160613.848843-27-armbru@redhat.com>
[Straightforwad conflict with commit 2336172d9b "audio: set default
value for pcspk.iobase property" resolved]
2020-07-07 19:05:54 +03:00
|
|
|
object_property_set_int(OBJECT(phb), "index", i, &error_fatal);
|
|
|
|
object_property_set_int(OBJECT(phb), "chip-id", chip->chip_id,
|
2020-01-27 17:45:06 +03:00
|
|
|
&error_fatal);
|
2021-12-17 19:57:19 +03:00
|
|
|
object_property_set_link(OBJECT(phb), "chip", OBJECT(chip),
|
|
|
|
&error_fatal);
|
error: Eliminate error_propagate() with Coccinelle, part 1
When all we do with an Error we receive into a local variable is
propagating to somewhere else, we can just as well receive it there
right away. Convert
if (!foo(..., &err)) {
...
error_propagate(errp, err);
...
return ...
}
to
if (!foo(..., errp)) {
...
...
return ...
}
where nothing else needs @err. Coccinelle script:
@rule1 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
binary operator op;
constant c1, c2;
symbol false;
@@
if (
(
- fun(args, &err, args2)
+ fun(args, errp, args2)
|
- !fun(args, &err, args2)
+ !fun(args, errp, args2)
|
- fun(args, &err, args2) op c1
+ fun(args, errp, args2) op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
)
}
@rule2 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
expression var;
binary operator op;
constant c1, c2;
symbol false;
@@
- var = fun(args, &err, args2);
+ var = fun(args, errp, args2);
... when != err
if (
(
var
|
!var
|
var op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
|
return var;
)
}
@depends on rule1 || rule2@
identifier err;
@@
- Error *err = NULL;
... when != err
Not exactly elegant, I'm afraid.
The "when != lbl:" is necessary to avoid transforming
if (fun(args, &err)) {
goto out
}
...
out:
error_propagate(errp, err);
even though other paths to label out still need the error_propagate().
For an actual example, see sclp_realize().
Without the "when strict", Coccinelle transforms vfio_msix_setup(),
incorrectly. I don't know what exactly "when strict" does, only that
it helps here.
The match of return is narrower than what I want, but I can't figure
out how to express "return where the operand doesn't use @err". For
an example where it's too narrow, see vfio_intx_enable().
Silently fails to convert hw/arm/armsse.c, because Coccinelle gets
confused by ARMSSE being used both as typedef and function-like macro
there. Converted manually.
Line breaks tidied up manually. One nested declaration of @local_err
deleted manually. Preexisting unwanted blank line dropped in
hw/riscv/sifive_e.c.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200707160613.848843-35-armbru@redhat.com>
2020-07-07 19:06:02 +03:00
|
|
|
if (!sysbus_realize(SYS_BUS_DEVICE(phb), errp)) {
|
2020-01-27 17:45:06 +03:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
}
|
2018-06-18 20:05:39 +03:00
|
|
|
}
|
|
|
|
|
2019-12-13 15:00:35 +03:00
|
|
|
static uint32_t pnv_chip_power8_xscom_pcba(PnvChip *chip, uint64_t addr)
|
|
|
|
{
|
|
|
|
addr &= (PNV_XSCOM_SIZE - 1);
|
|
|
|
return ((addr >> 4) & ~0xfull) | ((addr >> 3) & 0xf);
|
|
|
|
}
|
|
|
|
|
2016-10-22 12:46:36 +03:00
|
|
|
static void pnv_chip_power8e_class_init(ObjectClass *klass, void *data)
|
|
|
|
{
|
|
|
|
DeviceClass *dc = DEVICE_CLASS(klass);
|
|
|
|
PnvChipClass *k = PNV_CHIP_CLASS(klass);
|
|
|
|
|
|
|
|
k->chip_cfam_id = 0x221ef04980000000ull; /* P8 Murano DD2.1 */
|
2016-10-22 12:46:37 +03:00
|
|
|
k->cores_mask = POWER8E_CORE_MASK;
|
2020-01-27 17:45:06 +03:00
|
|
|
k->num_phbs = 3;
|
2016-10-22 12:46:38 +03:00
|
|
|
k->core_pir = pnv_chip_core_pir_p8;
|
2018-06-15 18:25:33 +03:00
|
|
|
k->intc_create = pnv_chip_power8_intc_create;
|
2019-10-22 19:38:10 +03:00
|
|
|
k->intc_reset = pnv_chip_power8_intc_reset;
|
2019-10-24 17:27:22 +03:00
|
|
|
k->intc_destroy = pnv_chip_power8_intc_destroy;
|
2019-12-13 15:00:07 +03:00
|
|
|
k->intc_print_info = pnv_chip_power8_intc_print_info;
|
2018-06-15 18:25:34 +03:00
|
|
|
k->isa_create = pnv_chip_power8_isa_create;
|
2019-03-06 11:50:12 +03:00
|
|
|
k->dt_populate = pnv_chip_power8_dt_populate;
|
2019-03-06 11:50:13 +03:00
|
|
|
k->pic_print_info = pnv_chip_power8_pic_print_info;
|
2019-12-13 15:00:13 +03:00
|
|
|
k->xscom_core_base = pnv_chip_power8_xscom_core_base;
|
2019-12-13 15:00:35 +03:00
|
|
|
k->xscom_pcba = pnv_chip_power8_xscom_pcba;
|
2016-10-22 12:46:36 +03:00
|
|
|
dc->desc = "PowerNV Chip POWER8E";
|
2018-06-18 20:05:39 +03:00
|
|
|
|
|
|
|
device_class_set_parent_realize(dc, pnv_chip_power8_realize,
|
|
|
|
&k->parent_realize);
|
2016-10-22 12:46:36 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_chip_power8_class_init(ObjectClass *klass, void *data)
|
|
|
|
{
|
|
|
|
DeviceClass *dc = DEVICE_CLASS(klass);
|
|
|
|
PnvChipClass *k = PNV_CHIP_CLASS(klass);
|
|
|
|
|
|
|
|
k->chip_cfam_id = 0x220ea04980000000ull; /* P8 Venice DD2.0 */
|
2016-10-22 12:46:37 +03:00
|
|
|
k->cores_mask = POWER8_CORE_MASK;
|
2020-01-27 17:45:06 +03:00
|
|
|
k->num_phbs = 3;
|
2016-10-22 12:46:38 +03:00
|
|
|
k->core_pir = pnv_chip_core_pir_p8;
|
2018-06-15 18:25:33 +03:00
|
|
|
k->intc_create = pnv_chip_power8_intc_create;
|
2019-10-22 19:38:10 +03:00
|
|
|
k->intc_reset = pnv_chip_power8_intc_reset;
|
2019-10-24 17:27:22 +03:00
|
|
|
k->intc_destroy = pnv_chip_power8_intc_destroy;
|
2019-12-13 15:00:07 +03:00
|
|
|
k->intc_print_info = pnv_chip_power8_intc_print_info;
|
2018-06-15 18:25:34 +03:00
|
|
|
k->isa_create = pnv_chip_power8_isa_create;
|
2019-03-06 11:50:12 +03:00
|
|
|
k->dt_populate = pnv_chip_power8_dt_populate;
|
2019-03-06 11:50:13 +03:00
|
|
|
k->pic_print_info = pnv_chip_power8_pic_print_info;
|
2019-12-13 15:00:13 +03:00
|
|
|
k->xscom_core_base = pnv_chip_power8_xscom_core_base;
|
2019-12-13 15:00:35 +03:00
|
|
|
k->xscom_pcba = pnv_chip_power8_xscom_pcba;
|
2016-10-22 12:46:36 +03:00
|
|
|
dc->desc = "PowerNV Chip POWER8";
|
2018-06-18 20:05:39 +03:00
|
|
|
|
|
|
|
device_class_set_parent_realize(dc, pnv_chip_power8_realize,
|
|
|
|
&k->parent_realize);
|
2016-10-22 12:46:36 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_chip_power8nvl_class_init(ObjectClass *klass, void *data)
|
|
|
|
{
|
|
|
|
DeviceClass *dc = DEVICE_CLASS(klass);
|
|
|
|
PnvChipClass *k = PNV_CHIP_CLASS(klass);
|
|
|
|
|
|
|
|
k->chip_cfam_id = 0x120d304980000000ull; /* P8 Naples DD1.0 */
|
2016-10-22 12:46:37 +03:00
|
|
|
k->cores_mask = POWER8_CORE_MASK;
|
2022-01-04 09:55:33 +03:00
|
|
|
k->num_phbs = 4;
|
2016-10-22 12:46:38 +03:00
|
|
|
k->core_pir = pnv_chip_core_pir_p8;
|
2018-06-15 18:25:33 +03:00
|
|
|
k->intc_create = pnv_chip_power8_intc_create;
|
2019-10-22 19:38:10 +03:00
|
|
|
k->intc_reset = pnv_chip_power8_intc_reset;
|
2019-10-24 17:27:22 +03:00
|
|
|
k->intc_destroy = pnv_chip_power8_intc_destroy;
|
2019-12-13 15:00:07 +03:00
|
|
|
k->intc_print_info = pnv_chip_power8_intc_print_info;
|
2018-06-15 18:25:34 +03:00
|
|
|
k->isa_create = pnv_chip_power8nvl_isa_create;
|
2019-03-06 11:50:12 +03:00
|
|
|
k->dt_populate = pnv_chip_power8_dt_populate;
|
2019-03-06 11:50:13 +03:00
|
|
|
k->pic_print_info = pnv_chip_power8_pic_print_info;
|
2019-12-13 15:00:13 +03:00
|
|
|
k->xscom_core_base = pnv_chip_power8_xscom_core_base;
|
2019-12-13 15:00:35 +03:00
|
|
|
k->xscom_pcba = pnv_chip_power8_xscom_pcba;
|
2016-10-22 12:46:36 +03:00
|
|
|
dc->desc = "PowerNV Chip POWER8NVL";
|
2018-06-18 20:05:39 +03:00
|
|
|
|
|
|
|
device_class_set_parent_realize(dc, pnv_chip_power8_realize,
|
|
|
|
&k->parent_realize);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_chip_power9_instance_init(Object *obj)
|
|
|
|
{
|
ppc/pnv: Add models for POWER9 PHB4 PCIe Host bridge
These changes introduces models for the PCIe Host Bridge (PHB4) of the
POWER9 processor. It includes the PowerBus logic interface (PBCQ),
IOMMU support, a single PCIe Gen.4 Root Complex, and support for MSI
and LSI interrupt sources as found on a POWER9 system using the XIVE
interrupt controller.
POWER9 processor comes with 3 PHB4 PEC (PCI Express Controller) and
each PEC can have several PHBs. By default,
* PEC0 provides 1 PHB (PHB0)
* PEC1 provides 2 PHBs (PHB1 and PHB2)
* PEC2 provides 3 PHBs (PHB3, PHB4 and PHB5)
Each PEC has a set "global" registers and some "per-stack" (per-PHB)
registers. Those are organized in two XSCOM ranges, the "Nest" range
and the "PCI" range, each range contains both some "PEC" registers and
some "per-stack" registers.
No default device layout is provided and PCI devices can be added on
any of the available PCIe Root Port (pcie.0 .. 2 of a Power9 chip)
with address 0x0 as the firwware (skiboot) only accepts a single
device per root port. To run a simple system with a network and a
storage adapters, use a command line options such as :
-device e1000e,netdev=net0,mac=C0:FF:EE:00:00:02,bus=pcie.0,addr=0x0
-netdev bridge,id=net0,helper=/usr/libexec/qemu-bridge-helper,br=virbr0,id=hostnet0
-device megasas,id=scsi0,bus=pcie.1,addr=0x0
-drive file=$disk,if=none,id=drive-scsi0-0-0-0,format=qcow2,cache=none
-device scsi-hd,bus=scsi0.0,channel=0,scsi-id=0,lun=0,drive=drive-scsi0-0-0-0,id=scsi0-0-0-0,bootindex=2
If more are needed, include a bridge.
Multi chip is supported, each chip adding its set of PHB4 controllers
and its PCI busses. The model doesn't emulate the EEH error handling.
This model is not ready for hotplug yet.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[ clg: - numerous cleanups
- commit log
- fix for broken LSI support
- PHB pic printinfo
- large QOM rework ]
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20200127144506.11132-2-clg@kaod.org>
[dwg: Use device_class_set_props()]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2020-01-27 17:45:05 +03:00
|
|
|
PnvChip *chip = PNV_CHIP(obj);
|
2019-03-06 11:50:11 +03:00
|
|
|
Pnv9Chip *chip9 = PNV9_CHIP(obj);
|
ppc/pnv: Add models for POWER9 PHB4 PCIe Host bridge
These changes introduces models for the PCIe Host Bridge (PHB4) of the
POWER9 processor. It includes the PowerBus logic interface (PBCQ),
IOMMU support, a single PCIe Gen.4 Root Complex, and support for MSI
and LSI interrupt sources as found on a POWER9 system using the XIVE
interrupt controller.
POWER9 processor comes with 3 PHB4 PEC (PCI Express Controller) and
each PEC can have several PHBs. By default,
* PEC0 provides 1 PHB (PHB0)
* PEC1 provides 2 PHBs (PHB1 and PHB2)
* PEC2 provides 3 PHBs (PHB3, PHB4 and PHB5)
Each PEC has a set "global" registers and some "per-stack" (per-PHB)
registers. Those are organized in two XSCOM ranges, the "Nest" range
and the "PCI" range, each range contains both some "PEC" registers and
some "per-stack" registers.
No default device layout is provided and PCI devices can be added on
any of the available PCIe Root Port (pcie.0 .. 2 of a Power9 chip)
with address 0x0 as the firwware (skiboot) only accepts a single
device per root port. To run a simple system with a network and a
storage adapters, use a command line options such as :
-device e1000e,netdev=net0,mac=C0:FF:EE:00:00:02,bus=pcie.0,addr=0x0
-netdev bridge,id=net0,helper=/usr/libexec/qemu-bridge-helper,br=virbr0,id=hostnet0
-device megasas,id=scsi0,bus=pcie.1,addr=0x0
-drive file=$disk,if=none,id=drive-scsi0-0-0-0,format=qcow2,cache=none
-device scsi-hd,bus=scsi0.0,channel=0,scsi-id=0,lun=0,drive=drive-scsi0-0-0-0,id=scsi0-0-0-0,bootindex=2
If more are needed, include a bridge.
Multi chip is supported, each chip adding its set of PHB4 controllers
and its PCI busses. The model doesn't emulate the EEH error handling.
This model is not ready for hotplug yet.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[ clg: - numerous cleanups
- commit log
- fix for broken LSI support
- PHB pic printinfo
- large QOM rework ]
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20200127144506.11132-2-clg@kaod.org>
[dwg: Use device_class_set_props()]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2020-01-27 17:45:05 +03:00
|
|
|
PnvChipClass *pcc = PNV_CHIP_GET_CLASS(obj);
|
|
|
|
int i;
|
2019-03-06 11:50:11 +03:00
|
|
|
|
sysbus: Convert qdev_set_parent_bus() use with Coccinelle, part 2
This is the same transformation as in the previous commit, except
sysbus_init_child_obj() and realize are too separated for the commit's
Coccinelle script to handle, typically because sysbus_init_child_obj()
is in a device's instance_init() method, and the matching realize is
in its realize() method.
Perhaps a Coccinelle wizard could make it transform that pattern, but
I'm just a bungler, and the best I can do is transforming the two
separate parts separately:
@@
expression errp;
expression child;
symbol true;
@@
- object_property_set_bool(OBJECT(child), true, "realized", errp);
+ sysbus_realize(SYS_BUS_DEVICE(child), errp);
// only correct with a matching sysbus_init_child_obj() transformation!
@@
expression errp;
expression child;
symbol true;
@@
- object_property_set_bool(child, true, "realized", errp);
+ sysbus_realize(SYS_BUS_DEVICE(child), errp);
// only correct with a matching sysbus_init_child_obj() transformation!
@@
expression child;
@@
- qdev_init_nofail(DEVICE(child));
+ sysbus_realize(SYS_BUS_DEVICE(child), &error_fatal);
// only correct with a matching sysbus_init_child_obj() transformation!
@@
expression child;
expression dev;
@@
dev = DEVICE(child);
...
- qdev_init_nofail(dev);
+ sysbus_realize(SYS_BUS_DEVICE(dev), &error_fatal);
// only correct with a matching sysbus_init_child_obj() transformation!
@@
expression child;
identifier dev;
@@
DeviceState *dev = DEVICE(child);
...
- qdev_init_nofail(dev);
+ sysbus_realize(SYS_BUS_DEVICE(dev), &error_fatal);
// only correct with a matching sysbus_init_child_obj() transformation!
@@
expression parent, name, size, type;
expression child;
symbol true;
@@
- sysbus_init_child_obj(parent, name, child, size, type);
+ sysbus_init_child_XXX(parent, name, child, size, type);
@@
expression parent, propname, type;
expression child;
@@
- sysbus_init_child_XXX(parent, propname, child, sizeof(*child), type)
+ object_initialize_child(parent, propname, child, type)
@@
expression parent, propname, type;
expression child;
@@
- sysbus_init_child_XXX(parent, propname, &child, sizeof(child), type)
+ object_initialize_child(parent, propname, &child, type)
This script is *unsound*: we need to manually verify init and realize
conversions are properly paired.
This commit has only the pairs where object_initialize_child()'s
@child and sysbus_realize()'s @dev argument text match exactly within
the same source file.
Note that Coccinelle chokes on ARMSSE typedef vs. macro in
hw/arm/armsse.c. Worked around by temporarily renaming the macro for
the spatch run.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200610053247.1583243-49-armbru@redhat.com>
2020-06-10 08:32:37 +03:00
|
|
|
object_initialize_child(obj, "xive", &chip9->xive, TYPE_PNV_XIVE);
|
2020-01-06 17:56:37 +03:00
|
|
|
object_property_add_alias(obj, "xive-fabric", OBJECT(&chip9->xive),
|
qom: Drop parameter @errp of object_property_add() & friends
The only way object_property_add() can fail is when a property with
the same name already exists. Since our property names are all
hardcoded, failure is a programming error, and the appropriate way to
handle it is passing &error_abort.
Same for its variants, except for object_property_add_child(), which
additionally fails when the child already has a parent. Parentage is
also under program control, so this is a programming error, too.
We have a bit over 500 callers. Almost half of them pass
&error_abort, slightly fewer ignore errors, one test case handles
errors, and the remaining few callers pass them to their own callers.
The previous few commits demonstrated once again that ignoring
programming errors is a bad idea.
Of the few ones that pass on errors, several violate the Error API.
The Error ** argument must be NULL, &error_abort, &error_fatal, or a
pointer to a variable containing NULL. Passing an argument of the
latter kind twice without clearing it in between is wrong: if the
first call sets an error, it no longer points to NULL for the second
call. ich9_pm_add_properties(), sparc32_ledma_realize(),
sparc32_dma_realize(), xilinx_axidma_realize(), xilinx_enet_realize()
are wrong that way.
When the one appropriate choice of argument is &error_abort, letting
users pick the argument is a bad idea.
Drop parameter @errp and assert the preconditions instead.
There's one exception to "duplicate property name is a programming
error": the way object_property_add() implements the magic (and
undocumented) "automatic arrayification". Don't drop @errp there.
Instead, rename object_property_add() to object_property_try_add(),
and add the obvious wrapper object_property_add().
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200505152926.18877-15-armbru@redhat.com>
[Two semantic rebase conflicts resolved]
2020-05-05 18:29:22 +03:00
|
|
|
"xive-fabric");
|
2019-03-08 01:35:35 +03:00
|
|
|
|
qom: Less verbose object_initialize_child()
All users of object_initialize_child() pass the obvious child size
argument. Almost all pass &error_abort and no properties. Tiresome.
Rename object_initialize_child() to
object_initialize_child_with_props() to free the name. New
convenience wrapper object_initialize_child() automates the size
argument, and passes &error_abort and no properties.
Rename object_initialize_childv() to
object_initialize_child_with_propsv() for consistency.
Convert callers with this Coccinelle script:
@@
expression parent, propname, type;
expression child, size;
symbol error_abort;
@@
- object_initialize_child(parent, propname, OBJECT(child), size, type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, size, type, &error_abort, NULL)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, child, sizeof(*child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, type)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, &child, sizeof(child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, &child, type)
@@
expression parent, propname, type;
expression child, size, err;
expression list props;
@@
- object_initialize_child(parent, propname, child, size, type, err, props)
+ object_initialize_child_with_props(parent, propname, child, size, type, err, props)
Note that Coccinelle chokes on ARMSSE typedef vs. macro in
hw/arm/armsse.c. Worked around by temporarily renaming the macro for
the spatch run.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
[Rebased: machine opentitan is new (commit fe0fe4735e7)]
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200610053247.1583243-37-armbru@redhat.com>
2020-06-10 08:32:25 +03:00
|
|
|
object_initialize_child(obj, "psi", &chip9->psi, TYPE_PNV9_PSI);
|
2019-03-08 01:35:39 +03:00
|
|
|
|
qom: Less verbose object_initialize_child()
All users of object_initialize_child() pass the obvious child size
argument. Almost all pass &error_abort and no properties. Tiresome.
Rename object_initialize_child() to
object_initialize_child_with_props() to free the name. New
convenience wrapper object_initialize_child() automates the size
argument, and passes &error_abort and no properties.
Rename object_initialize_childv() to
object_initialize_child_with_propsv() for consistency.
Convert callers with this Coccinelle script:
@@
expression parent, propname, type;
expression child, size;
symbol error_abort;
@@
- object_initialize_child(parent, propname, OBJECT(child), size, type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, size, type, &error_abort, NULL)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, child, sizeof(*child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, type)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, &child, sizeof(child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, &child, type)
@@
expression parent, propname, type;
expression child, size, err;
expression list props;
@@
- object_initialize_child(parent, propname, child, size, type, err, props)
+ object_initialize_child_with_props(parent, propname, child, size, type, err, props)
Note that Coccinelle chokes on ARMSSE typedef vs. macro in
hw/arm/armsse.c. Worked around by temporarily renaming the macro for
the spatch run.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
[Rebased: machine opentitan is new (commit fe0fe4735e7)]
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200610053247.1583243-37-armbru@redhat.com>
2020-06-10 08:32:25 +03:00
|
|
|
object_initialize_child(obj, "lpc", &chip9->lpc, TYPE_PNV9_LPC);
|
2019-03-08 01:35:42 +03:00
|
|
|
|
qom: Less verbose object_initialize_child()
All users of object_initialize_child() pass the obvious child size
argument. Almost all pass &error_abort and no properties. Tiresome.
Rename object_initialize_child() to
object_initialize_child_with_props() to free the name. New
convenience wrapper object_initialize_child() automates the size
argument, and passes &error_abort and no properties.
Rename object_initialize_childv() to
object_initialize_child_with_propsv() for consistency.
Convert callers with this Coccinelle script:
@@
expression parent, propname, type;
expression child, size;
symbol error_abort;
@@
- object_initialize_child(parent, propname, OBJECT(child), size, type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, size, type, &error_abort, NULL)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, child, sizeof(*child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, type)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, &child, sizeof(child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, &child, type)
@@
expression parent, propname, type;
expression child, size, err;
expression list props;
@@
- object_initialize_child(parent, propname, child, size, type, err, props)
+ object_initialize_child_with_props(parent, propname, child, size, type, err, props)
Note that Coccinelle chokes on ARMSSE typedef vs. macro in
hw/arm/armsse.c. Worked around by temporarily renaming the macro for
the spatch run.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
[Rebased: machine opentitan is new (commit fe0fe4735e7)]
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200610053247.1583243-37-armbru@redhat.com>
2020-06-10 08:32:25 +03:00
|
|
|
object_initialize_child(obj, "occ", &chip9->occ, TYPE_PNV9_OCC);
|
2019-09-12 12:30:54 +03:00
|
|
|
|
2022-08-11 12:37:26 +03:00
|
|
|
object_initialize_child(obj, "sbe", &chip9->sbe, TYPE_PNV9_SBE);
|
|
|
|
|
qom: Less verbose object_initialize_child()
All users of object_initialize_child() pass the obvious child size
argument. Almost all pass &error_abort and no properties. Tiresome.
Rename object_initialize_child() to
object_initialize_child_with_props() to free the name. New
convenience wrapper object_initialize_child() automates the size
argument, and passes &error_abort and no properties.
Rename object_initialize_childv() to
object_initialize_child_with_propsv() for consistency.
Convert callers with this Coccinelle script:
@@
expression parent, propname, type;
expression child, size;
symbol error_abort;
@@
- object_initialize_child(parent, propname, OBJECT(child), size, type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, size, type, &error_abort, NULL)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, child, sizeof(*child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, type)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, &child, sizeof(child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, &child, type)
@@
expression parent, propname, type;
expression child, size, err;
expression list props;
@@
- object_initialize_child(parent, propname, child, size, type, err, props)
+ object_initialize_child_with_props(parent, propname, child, size, type, err, props)
Note that Coccinelle chokes on ARMSSE typedef vs. macro in
hw/arm/armsse.c. Worked around by temporarily renaming the macro for
the spatch run.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
[Rebased: machine opentitan is new (commit fe0fe4735e7)]
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200610053247.1583243-37-armbru@redhat.com>
2020-06-10 08:32:25 +03:00
|
|
|
object_initialize_child(obj, "homer", &chip9->homer, TYPE_PNV9_HOMER);
|
ppc/pnv: Add models for POWER9 PHB4 PCIe Host bridge
These changes introduces models for the PCIe Host Bridge (PHB4) of the
POWER9 processor. It includes the PowerBus logic interface (PBCQ),
IOMMU support, a single PCIe Gen.4 Root Complex, and support for MSI
and LSI interrupt sources as found on a POWER9 system using the XIVE
interrupt controller.
POWER9 processor comes with 3 PHB4 PEC (PCI Express Controller) and
each PEC can have several PHBs. By default,
* PEC0 provides 1 PHB (PHB0)
* PEC1 provides 2 PHBs (PHB1 and PHB2)
* PEC2 provides 3 PHBs (PHB3, PHB4 and PHB5)
Each PEC has a set "global" registers and some "per-stack" (per-PHB)
registers. Those are organized in two XSCOM ranges, the "Nest" range
and the "PCI" range, each range contains both some "PEC" registers and
some "per-stack" registers.
No default device layout is provided and PCI devices can be added on
any of the available PCIe Root Port (pcie.0 .. 2 of a Power9 chip)
with address 0x0 as the firwware (skiboot) only accepts a single
device per root port. To run a simple system with a network and a
storage adapters, use a command line options such as :
-device e1000e,netdev=net0,mac=C0:FF:EE:00:00:02,bus=pcie.0,addr=0x0
-netdev bridge,id=net0,helper=/usr/libexec/qemu-bridge-helper,br=virbr0,id=hostnet0
-device megasas,id=scsi0,bus=pcie.1,addr=0x0
-drive file=$disk,if=none,id=drive-scsi0-0-0-0,format=qcow2,cache=none
-device scsi-hd,bus=scsi0.0,channel=0,scsi-id=0,lun=0,drive=drive-scsi0-0-0-0,id=scsi0-0-0-0,bootindex=2
If more are needed, include a bridge.
Multi chip is supported, each chip adding its set of PHB4 controllers
and its PCI busses. The model doesn't emulate the EEH error handling.
This model is not ready for hotplug yet.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[ clg: - numerous cleanups
- commit log
- fix for broken LSI support
- PHB pic printinfo
- large QOM rework ]
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20200127144506.11132-2-clg@kaod.org>
[dwg: Use device_class_set_props()]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2020-01-27 17:45:05 +03:00
|
|
|
|
2021-12-17 19:57:19 +03:00
|
|
|
/* Number of PECs is the chip default */
|
|
|
|
chip->num_pecs = pcc->num_pecs;
|
|
|
|
|
|
|
|
for (i = 0; i < chip->num_pecs; i++) {
|
ppc/pnv: Add models for POWER9 PHB4 PCIe Host bridge
These changes introduces models for the PCIe Host Bridge (PHB4) of the
POWER9 processor. It includes the PowerBus logic interface (PBCQ),
IOMMU support, a single PCIe Gen.4 Root Complex, and support for MSI
and LSI interrupt sources as found on a POWER9 system using the XIVE
interrupt controller.
POWER9 processor comes with 3 PHB4 PEC (PCI Express Controller) and
each PEC can have several PHBs. By default,
* PEC0 provides 1 PHB (PHB0)
* PEC1 provides 2 PHBs (PHB1 and PHB2)
* PEC2 provides 3 PHBs (PHB3, PHB4 and PHB5)
Each PEC has a set "global" registers and some "per-stack" (per-PHB)
registers. Those are organized in two XSCOM ranges, the "Nest" range
and the "PCI" range, each range contains both some "PEC" registers and
some "per-stack" registers.
No default device layout is provided and PCI devices can be added on
any of the available PCIe Root Port (pcie.0 .. 2 of a Power9 chip)
with address 0x0 as the firwware (skiboot) only accepts a single
device per root port. To run a simple system with a network and a
storage adapters, use a command line options such as :
-device e1000e,netdev=net0,mac=C0:FF:EE:00:00:02,bus=pcie.0,addr=0x0
-netdev bridge,id=net0,helper=/usr/libexec/qemu-bridge-helper,br=virbr0,id=hostnet0
-device megasas,id=scsi0,bus=pcie.1,addr=0x0
-drive file=$disk,if=none,id=drive-scsi0-0-0-0,format=qcow2,cache=none
-device scsi-hd,bus=scsi0.0,channel=0,scsi-id=0,lun=0,drive=drive-scsi0-0-0-0,id=scsi0-0-0-0,bootindex=2
If more are needed, include a bridge.
Multi chip is supported, each chip adding its set of PHB4 controllers
and its PCI busses. The model doesn't emulate the EEH error handling.
This model is not ready for hotplug yet.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[ clg: - numerous cleanups
- commit log
- fix for broken LSI support
- PHB pic printinfo
- large QOM rework ]
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20200127144506.11132-2-clg@kaod.org>
[dwg: Use device_class_set_props()]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2020-01-27 17:45:05 +03:00
|
|
|
object_initialize_child(obj, "pec[*]", &chip9->pecs[i],
|
qom: Less verbose object_initialize_child()
All users of object_initialize_child() pass the obvious child size
argument. Almost all pass &error_abort and no properties. Tiresome.
Rename object_initialize_child() to
object_initialize_child_with_props() to free the name. New
convenience wrapper object_initialize_child() automates the size
argument, and passes &error_abort and no properties.
Rename object_initialize_childv() to
object_initialize_child_with_propsv() for consistency.
Convert callers with this Coccinelle script:
@@
expression parent, propname, type;
expression child, size;
symbol error_abort;
@@
- object_initialize_child(parent, propname, OBJECT(child), size, type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, size, type, &error_abort, NULL)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, child, sizeof(*child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, type)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, &child, sizeof(child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, &child, type)
@@
expression parent, propname, type;
expression child, size, err;
expression list props;
@@
- object_initialize_child(parent, propname, child, size, type, err, props)
+ object_initialize_child_with_props(parent, propname, child, size, type, err, props)
Note that Coccinelle chokes on ARMSSE typedef vs. macro in
hw/arm/armsse.c. Worked around by temporarily renaming the macro for
the spatch run.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
[Rebased: machine opentitan is new (commit fe0fe4735e7)]
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200610053247.1583243-37-armbru@redhat.com>
2020-06-10 08:32:25 +03:00
|
|
|
TYPE_PNV_PHB4_PEC);
|
ppc/pnv: Add models for POWER9 PHB4 PCIe Host bridge
These changes introduces models for the PCIe Host Bridge (PHB4) of the
POWER9 processor. It includes the PowerBus logic interface (PBCQ),
IOMMU support, a single PCIe Gen.4 Root Complex, and support for MSI
and LSI interrupt sources as found on a POWER9 system using the XIVE
interrupt controller.
POWER9 processor comes with 3 PHB4 PEC (PCI Express Controller) and
each PEC can have several PHBs. By default,
* PEC0 provides 1 PHB (PHB0)
* PEC1 provides 2 PHBs (PHB1 and PHB2)
* PEC2 provides 3 PHBs (PHB3, PHB4 and PHB5)
Each PEC has a set "global" registers and some "per-stack" (per-PHB)
registers. Those are organized in two XSCOM ranges, the "Nest" range
and the "PCI" range, each range contains both some "PEC" registers and
some "per-stack" registers.
No default device layout is provided and PCI devices can be added on
any of the available PCIe Root Port (pcie.0 .. 2 of a Power9 chip)
with address 0x0 as the firwware (skiboot) only accepts a single
device per root port. To run a simple system with a network and a
storage adapters, use a command line options such as :
-device e1000e,netdev=net0,mac=C0:FF:EE:00:00:02,bus=pcie.0,addr=0x0
-netdev bridge,id=net0,helper=/usr/libexec/qemu-bridge-helper,br=virbr0,id=hostnet0
-device megasas,id=scsi0,bus=pcie.1,addr=0x0
-drive file=$disk,if=none,id=drive-scsi0-0-0-0,format=qcow2,cache=none
-device scsi-hd,bus=scsi0.0,channel=0,scsi-id=0,lun=0,drive=drive-scsi0-0-0-0,id=scsi0-0-0-0,bootindex=2
If more are needed, include a bridge.
Multi chip is supported, each chip adding its set of PHB4 controllers
and its PCI busses. The model doesn't emulate the EEH error handling.
This model is not ready for hotplug yet.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[ clg: - numerous cleanups
- commit log
- fix for broken LSI support
- PHB pic printinfo
- large QOM rework ]
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20200127144506.11132-2-clg@kaod.org>
[dwg: Use device_class_set_props()]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2020-01-27 17:45:05 +03:00
|
|
|
}
|
2018-06-18 20:05:39 +03:00
|
|
|
}
|
|
|
|
|
2022-03-02 08:51:39 +03:00
|
|
|
static void pnv_chip_quad_realize_one(PnvChip *chip, PnvQuad *eq,
|
|
|
|
PnvCore *pnv_core)
|
|
|
|
{
|
|
|
|
char eq_name[32];
|
|
|
|
int core_id = CPU_CORE(pnv_core)->core_id;
|
|
|
|
|
|
|
|
snprintf(eq_name, sizeof(eq_name), "eq[%d]", core_id);
|
|
|
|
object_initialize_child_with_props(OBJECT(chip), eq_name, eq,
|
|
|
|
sizeof(*eq), TYPE_PNV_QUAD,
|
|
|
|
&error_fatal, NULL);
|
|
|
|
|
|
|
|
object_property_set_int(OBJECT(eq), "quad-id", core_id, &error_fatal);
|
|
|
|
qdev_realize(DEVICE(eq), NULL, &error_fatal);
|
|
|
|
}
|
|
|
|
|
2019-03-08 01:35:44 +03:00
|
|
|
static void pnv_chip_quad_realize(Pnv9Chip *chip9, Error **errp)
|
|
|
|
{
|
|
|
|
PnvChip *chip = PNV_CHIP(chip9);
|
|
|
|
int i;
|
|
|
|
|
|
|
|
chip9->nr_quads = DIV_ROUND_UP(chip->nr_cores, 4);
|
|
|
|
chip9->quads = g_new0(PnvQuad, chip9->nr_quads);
|
|
|
|
|
|
|
|
for (i = 0; i < chip9->nr_quads; i++) {
|
|
|
|
PnvQuad *eq = &chip9->quads[i];
|
|
|
|
|
2022-03-02 08:51:39 +03:00
|
|
|
pnv_chip_quad_realize_one(chip, eq, chip->cores[i * 4]);
|
2019-03-08 01:35:44 +03:00
|
|
|
|
2021-09-01 12:41:53 +03:00
|
|
|
pnv_xscom_add_subregion(chip, PNV9_XSCOM_EQ_BASE(eq->quad_id),
|
2019-03-08 01:35:44 +03:00
|
|
|
&eq->xscom_regs);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2021-12-17 19:57:19 +03:00
|
|
|
static void pnv_chip_power9_pec_realize(PnvChip *chip, Error **errp)
|
ppc/pnv: Add models for POWER9 PHB4 PCIe Host bridge
These changes introduces models for the PCIe Host Bridge (PHB4) of the
POWER9 processor. It includes the PowerBus logic interface (PBCQ),
IOMMU support, a single PCIe Gen.4 Root Complex, and support for MSI
and LSI interrupt sources as found on a POWER9 system using the XIVE
interrupt controller.
POWER9 processor comes with 3 PHB4 PEC (PCI Express Controller) and
each PEC can have several PHBs. By default,
* PEC0 provides 1 PHB (PHB0)
* PEC1 provides 2 PHBs (PHB1 and PHB2)
* PEC2 provides 3 PHBs (PHB3, PHB4 and PHB5)
Each PEC has a set "global" registers and some "per-stack" (per-PHB)
registers. Those are organized in two XSCOM ranges, the "Nest" range
and the "PCI" range, each range contains both some "PEC" registers and
some "per-stack" registers.
No default device layout is provided and PCI devices can be added on
any of the available PCIe Root Port (pcie.0 .. 2 of a Power9 chip)
with address 0x0 as the firwware (skiboot) only accepts a single
device per root port. To run a simple system with a network and a
storage adapters, use a command line options such as :
-device e1000e,netdev=net0,mac=C0:FF:EE:00:00:02,bus=pcie.0,addr=0x0
-netdev bridge,id=net0,helper=/usr/libexec/qemu-bridge-helper,br=virbr0,id=hostnet0
-device megasas,id=scsi0,bus=pcie.1,addr=0x0
-drive file=$disk,if=none,id=drive-scsi0-0-0-0,format=qcow2,cache=none
-device scsi-hd,bus=scsi0.0,channel=0,scsi-id=0,lun=0,drive=drive-scsi0-0-0-0,id=scsi0-0-0-0,bootindex=2
If more are needed, include a bridge.
Multi chip is supported, each chip adding its set of PHB4 controllers
and its PCI busses. The model doesn't emulate the EEH error handling.
This model is not ready for hotplug yet.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[ clg: - numerous cleanups
- commit log
- fix for broken LSI support
- PHB pic printinfo
- large QOM rework ]
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20200127144506.11132-2-clg@kaod.org>
[dwg: Use device_class_set_props()]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2020-01-27 17:45:05 +03:00
|
|
|
{
|
|
|
|
Pnv9Chip *chip9 = PNV9_CHIP(chip);
|
2021-12-17 19:57:19 +03:00
|
|
|
int i;
|
ppc/pnv: Add models for POWER9 PHB4 PCIe Host bridge
These changes introduces models for the PCIe Host Bridge (PHB4) of the
POWER9 processor. It includes the PowerBus logic interface (PBCQ),
IOMMU support, a single PCIe Gen.4 Root Complex, and support for MSI
and LSI interrupt sources as found on a POWER9 system using the XIVE
interrupt controller.
POWER9 processor comes with 3 PHB4 PEC (PCI Express Controller) and
each PEC can have several PHBs. By default,
* PEC0 provides 1 PHB (PHB0)
* PEC1 provides 2 PHBs (PHB1 and PHB2)
* PEC2 provides 3 PHBs (PHB3, PHB4 and PHB5)
Each PEC has a set "global" registers and some "per-stack" (per-PHB)
registers. Those are organized in two XSCOM ranges, the "Nest" range
and the "PCI" range, each range contains both some "PEC" registers and
some "per-stack" registers.
No default device layout is provided and PCI devices can be added on
any of the available PCIe Root Port (pcie.0 .. 2 of a Power9 chip)
with address 0x0 as the firwware (skiboot) only accepts a single
device per root port. To run a simple system with a network and a
storage adapters, use a command line options such as :
-device e1000e,netdev=net0,mac=C0:FF:EE:00:00:02,bus=pcie.0,addr=0x0
-netdev bridge,id=net0,helper=/usr/libexec/qemu-bridge-helper,br=virbr0,id=hostnet0
-device megasas,id=scsi0,bus=pcie.1,addr=0x0
-drive file=$disk,if=none,id=drive-scsi0-0-0-0,format=qcow2,cache=none
-device scsi-hd,bus=scsi0.0,channel=0,scsi-id=0,lun=0,drive=drive-scsi0-0-0-0,id=scsi0-0-0-0,bootindex=2
If more are needed, include a bridge.
Multi chip is supported, each chip adding its set of PHB4 controllers
and its PCI busses. The model doesn't emulate the EEH error handling.
This model is not ready for hotplug yet.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[ clg: - numerous cleanups
- commit log
- fix for broken LSI support
- PHB pic printinfo
- large QOM rework ]
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20200127144506.11132-2-clg@kaod.org>
[dwg: Use device_class_set_props()]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2020-01-27 17:45:05 +03:00
|
|
|
|
2021-12-17 19:57:19 +03:00
|
|
|
for (i = 0; i < chip->num_pecs; i++) {
|
ppc/pnv: Add models for POWER9 PHB4 PCIe Host bridge
These changes introduces models for the PCIe Host Bridge (PHB4) of the
POWER9 processor. It includes the PowerBus logic interface (PBCQ),
IOMMU support, a single PCIe Gen.4 Root Complex, and support for MSI
and LSI interrupt sources as found on a POWER9 system using the XIVE
interrupt controller.
POWER9 processor comes with 3 PHB4 PEC (PCI Express Controller) and
each PEC can have several PHBs. By default,
* PEC0 provides 1 PHB (PHB0)
* PEC1 provides 2 PHBs (PHB1 and PHB2)
* PEC2 provides 3 PHBs (PHB3, PHB4 and PHB5)
Each PEC has a set "global" registers and some "per-stack" (per-PHB)
registers. Those are organized in two XSCOM ranges, the "Nest" range
and the "PCI" range, each range contains both some "PEC" registers and
some "per-stack" registers.
No default device layout is provided and PCI devices can be added on
any of the available PCIe Root Port (pcie.0 .. 2 of a Power9 chip)
with address 0x0 as the firwware (skiboot) only accepts a single
device per root port. To run a simple system with a network and a
storage adapters, use a command line options such as :
-device e1000e,netdev=net0,mac=C0:FF:EE:00:00:02,bus=pcie.0,addr=0x0
-netdev bridge,id=net0,helper=/usr/libexec/qemu-bridge-helper,br=virbr0,id=hostnet0
-device megasas,id=scsi0,bus=pcie.1,addr=0x0
-drive file=$disk,if=none,id=drive-scsi0-0-0-0,format=qcow2,cache=none
-device scsi-hd,bus=scsi0.0,channel=0,scsi-id=0,lun=0,drive=drive-scsi0-0-0-0,id=scsi0-0-0-0,bootindex=2
If more are needed, include a bridge.
Multi chip is supported, each chip adding its set of PHB4 controllers
and its PCI busses. The model doesn't emulate the EEH error handling.
This model is not ready for hotplug yet.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[ clg: - numerous cleanups
- commit log
- fix for broken LSI support
- PHB pic printinfo
- large QOM rework ]
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20200127144506.11132-2-clg@kaod.org>
[dwg: Use device_class_set_props()]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2020-01-27 17:45:05 +03:00
|
|
|
PnvPhb4PecState *pec = &chip9->pecs[i];
|
|
|
|
PnvPhb4PecClass *pecc = PNV_PHB4_PEC_GET_CLASS(pec);
|
|
|
|
uint32_t pec_nest_base;
|
|
|
|
uint32_t pec_pci_base;
|
|
|
|
|
qom: Put name parameter before value / visitor parameter
The object_property_set_FOO() setters take property name and value in
an unusual order:
void object_property_set_FOO(Object *obj, FOO_TYPE value,
const char *name, Error **errp)
Having to pass value before name feels grating. Swap them.
Same for object_property_set(), object_property_get(), and
object_property_parse().
Convert callers with this Coccinelle script:
@@
identifier fun = {
object_property_get, object_property_parse, object_property_set_str,
object_property_set_link, object_property_set_bool,
object_property_set_int, object_property_set_uint, object_property_set,
object_property_set_qobject
};
expression obj, v, name, errp;
@@
- fun(obj, v, name, errp)
+ fun(obj, name, v, errp)
Chokes on hw/arm/musicpal.c's lcd_refresh() with the unhelpful error
message "no position information". Convert that one manually.
Fails to convert hw/arm/armsse.c, because Coccinelle gets confused by
ARMSSE being used both as typedef and function-like macro there.
Convert manually.
Fails to convert hw/rx/rx-gdbsim.c, because Coccinelle gets confused
by RXCPU being used both as typedef and function-like macro there.
Convert manually. The other files using RXCPU that way don't need
conversion.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20200707160613.848843-27-armbru@redhat.com>
[Straightforwad conflict with commit 2336172d9b "audio: set default
value for pcspk.iobase property" resolved]
2020-07-07 19:05:54 +03:00
|
|
|
object_property_set_int(OBJECT(pec), "index", i, &error_fatal);
|
|
|
|
object_property_set_int(OBJECT(pec), "chip-id", chip->chip_id,
|
ppc/pnv: Add models for POWER9 PHB4 PCIe Host bridge
These changes introduces models for the PCIe Host Bridge (PHB4) of the
POWER9 processor. It includes the PowerBus logic interface (PBCQ),
IOMMU support, a single PCIe Gen.4 Root Complex, and support for MSI
and LSI interrupt sources as found on a POWER9 system using the XIVE
interrupt controller.
POWER9 processor comes with 3 PHB4 PEC (PCI Express Controller) and
each PEC can have several PHBs. By default,
* PEC0 provides 1 PHB (PHB0)
* PEC1 provides 2 PHBs (PHB1 and PHB2)
* PEC2 provides 3 PHBs (PHB3, PHB4 and PHB5)
Each PEC has a set "global" registers and some "per-stack" (per-PHB)
registers. Those are organized in two XSCOM ranges, the "Nest" range
and the "PCI" range, each range contains both some "PEC" registers and
some "per-stack" registers.
No default device layout is provided and PCI devices can be added on
any of the available PCIe Root Port (pcie.0 .. 2 of a Power9 chip)
with address 0x0 as the firwware (skiboot) only accepts a single
device per root port. To run a simple system with a network and a
storage adapters, use a command line options such as :
-device e1000e,netdev=net0,mac=C0:FF:EE:00:00:02,bus=pcie.0,addr=0x0
-netdev bridge,id=net0,helper=/usr/libexec/qemu-bridge-helper,br=virbr0,id=hostnet0
-device megasas,id=scsi0,bus=pcie.1,addr=0x0
-drive file=$disk,if=none,id=drive-scsi0-0-0-0,format=qcow2,cache=none
-device scsi-hd,bus=scsi0.0,channel=0,scsi-id=0,lun=0,drive=drive-scsi0-0-0-0,id=scsi0-0-0-0,bootindex=2
If more are needed, include a bridge.
Multi chip is supported, each chip adding its set of PHB4 controllers
and its PCI busses. The model doesn't emulate the EEH error handling.
This model is not ready for hotplug yet.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[ clg: - numerous cleanups
- commit log
- fix for broken LSI support
- PHB pic printinfo
- large QOM rework ]
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20200127144506.11132-2-clg@kaod.org>
[dwg: Use device_class_set_props()]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2020-01-27 17:45:05 +03:00
|
|
|
&error_fatal);
|
2021-12-17 19:57:19 +03:00
|
|
|
object_property_set_link(OBJECT(pec), "chip", OBJECT(chip),
|
|
|
|
&error_fatal);
|
error: Eliminate error_propagate() with Coccinelle, part 1
When all we do with an Error we receive into a local variable is
propagating to somewhere else, we can just as well receive it there
right away. Convert
if (!foo(..., &err)) {
...
error_propagate(errp, err);
...
return ...
}
to
if (!foo(..., errp)) {
...
...
return ...
}
where nothing else needs @err. Coccinelle script:
@rule1 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
binary operator op;
constant c1, c2;
symbol false;
@@
if (
(
- fun(args, &err, args2)
+ fun(args, errp, args2)
|
- !fun(args, &err, args2)
+ !fun(args, errp, args2)
|
- fun(args, &err, args2) op c1
+ fun(args, errp, args2) op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
)
}
@rule2 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
expression var;
binary operator op;
constant c1, c2;
symbol false;
@@
- var = fun(args, &err, args2);
+ var = fun(args, errp, args2);
... when != err
if (
(
var
|
!var
|
var op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
|
return var;
)
}
@depends on rule1 || rule2@
identifier err;
@@
- Error *err = NULL;
... when != err
Not exactly elegant, I'm afraid.
The "when != lbl:" is necessary to avoid transforming
if (fun(args, &err)) {
goto out
}
...
out:
error_propagate(errp, err);
even though other paths to label out still need the error_propagate().
For an actual example, see sclp_realize().
Without the "when strict", Coccinelle transforms vfio_msix_setup(),
incorrectly. I don't know what exactly "when strict" does, only that
it helps here.
The match of return is narrower than what I want, but I can't figure
out how to express "return where the operand doesn't use @err". For
an example where it's too narrow, see vfio_intx_enable().
Silently fails to convert hw/arm/armsse.c, because Coccinelle gets
confused by ARMSSE being used both as typedef and function-like macro
there. Converted manually.
Line breaks tidied up manually. One nested declaration of @local_err
deleted manually. Preexisting unwanted blank line dropped in
hw/riscv/sifive_e.c.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200707160613.848843-35-armbru@redhat.com>
2020-07-07 19:06:02 +03:00
|
|
|
if (!qdev_realize(DEVICE(pec), NULL, errp)) {
|
ppc/pnv: Add models for POWER9 PHB4 PCIe Host bridge
These changes introduces models for the PCIe Host Bridge (PHB4) of the
POWER9 processor. It includes the PowerBus logic interface (PBCQ),
IOMMU support, a single PCIe Gen.4 Root Complex, and support for MSI
and LSI interrupt sources as found on a POWER9 system using the XIVE
interrupt controller.
POWER9 processor comes with 3 PHB4 PEC (PCI Express Controller) and
each PEC can have several PHBs. By default,
* PEC0 provides 1 PHB (PHB0)
* PEC1 provides 2 PHBs (PHB1 and PHB2)
* PEC2 provides 3 PHBs (PHB3, PHB4 and PHB5)
Each PEC has a set "global" registers and some "per-stack" (per-PHB)
registers. Those are organized in two XSCOM ranges, the "Nest" range
and the "PCI" range, each range contains both some "PEC" registers and
some "per-stack" registers.
No default device layout is provided and PCI devices can be added on
any of the available PCIe Root Port (pcie.0 .. 2 of a Power9 chip)
with address 0x0 as the firwware (skiboot) only accepts a single
device per root port. To run a simple system with a network and a
storage adapters, use a command line options such as :
-device e1000e,netdev=net0,mac=C0:FF:EE:00:00:02,bus=pcie.0,addr=0x0
-netdev bridge,id=net0,helper=/usr/libexec/qemu-bridge-helper,br=virbr0,id=hostnet0
-device megasas,id=scsi0,bus=pcie.1,addr=0x0
-drive file=$disk,if=none,id=drive-scsi0-0-0-0,format=qcow2,cache=none
-device scsi-hd,bus=scsi0.0,channel=0,scsi-id=0,lun=0,drive=drive-scsi0-0-0-0,id=scsi0-0-0-0,bootindex=2
If more are needed, include a bridge.
Multi chip is supported, each chip adding its set of PHB4 controllers
and its PCI busses. The model doesn't emulate the EEH error handling.
This model is not ready for hotplug yet.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[ clg: - numerous cleanups
- commit log
- fix for broken LSI support
- PHB pic printinfo
- large QOM rework ]
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20200127144506.11132-2-clg@kaod.org>
[dwg: Use device_class_set_props()]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2020-01-27 17:45:05 +03:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
pec_nest_base = pecc->xscom_nest_base(pec);
|
|
|
|
pec_pci_base = pecc->xscom_pci_base(pec);
|
|
|
|
|
|
|
|
pnv_xscom_add_subregion(chip, pec_nest_base, &pec->nest_regs_mr);
|
|
|
|
pnv_xscom_add_subregion(chip, pec_pci_base, &pec->pci_regs_mr);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-06-18 20:05:39 +03:00
|
|
|
static void pnv_chip_power9_realize(DeviceState *dev, Error **errp)
|
|
|
|
{
|
|
|
|
PnvChipClass *pcc = PNV_CHIP_GET_CLASS(dev);
|
2019-03-06 11:50:11 +03:00
|
|
|
Pnv9Chip *chip9 = PNV9_CHIP(dev);
|
|
|
|
PnvChip *chip = PNV_CHIP(dev);
|
2019-03-08 01:35:35 +03:00
|
|
|
Pnv9Psi *psi9 = &chip9->psi;
|
2018-06-18 20:05:39 +03:00
|
|
|
Error *local_err = NULL;
|
|
|
|
|
2019-06-12 20:43:44 +03:00
|
|
|
/* XSCOM bridge is first */
|
|
|
|
pnv_xscom_realize(chip, PNV9_XSCOM_SIZE, &local_err);
|
|
|
|
if (local_err) {
|
|
|
|
error_propagate(errp, local_err);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
sysbus_mmio_map(SYS_BUS_DEVICE(chip), 0, PNV9_XSCOM_BASE(chip));
|
|
|
|
|
2018-06-18 20:05:39 +03:00
|
|
|
pcc->parent_realize(dev, &local_err);
|
|
|
|
if (local_err) {
|
|
|
|
error_propagate(errp, local_err);
|
|
|
|
return;
|
|
|
|
}
|
2019-03-06 11:50:11 +03:00
|
|
|
|
2019-03-08 01:35:44 +03:00
|
|
|
pnv_chip_quad_realize(chip9, &local_err);
|
|
|
|
if (local_err) {
|
|
|
|
error_propagate(errp, local_err);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2019-03-06 11:50:11 +03:00
|
|
|
/* XIVE interrupt controller (POWER9) */
|
qom: Put name parameter before value / visitor parameter
The object_property_set_FOO() setters take property name and value in
an unusual order:
void object_property_set_FOO(Object *obj, FOO_TYPE value,
const char *name, Error **errp)
Having to pass value before name feels grating. Swap them.
Same for object_property_set(), object_property_get(), and
object_property_parse().
Convert callers with this Coccinelle script:
@@
identifier fun = {
object_property_get, object_property_parse, object_property_set_str,
object_property_set_link, object_property_set_bool,
object_property_set_int, object_property_set_uint, object_property_set,
object_property_set_qobject
};
expression obj, v, name, errp;
@@
- fun(obj, v, name, errp)
+ fun(obj, name, v, errp)
Chokes on hw/arm/musicpal.c's lcd_refresh() with the unhelpful error
message "no position information". Convert that one manually.
Fails to convert hw/arm/armsse.c, because Coccinelle gets confused by
ARMSSE being used both as typedef and function-like macro there.
Convert manually.
Fails to convert hw/rx/rx-gdbsim.c, because Coccinelle gets confused
by RXCPU being used both as typedef and function-like macro there.
Convert manually. The other files using RXCPU that way don't need
conversion.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20200707160613.848843-27-armbru@redhat.com>
[Straightforwad conflict with commit 2336172d9b "audio: set default
value for pcspk.iobase property" resolved]
2020-07-07 19:05:54 +03:00
|
|
|
object_property_set_int(OBJECT(&chip9->xive), "ic-bar",
|
|
|
|
PNV9_XIVE_IC_BASE(chip), &error_fatal);
|
|
|
|
object_property_set_int(OBJECT(&chip9->xive), "vc-bar",
|
|
|
|
PNV9_XIVE_VC_BASE(chip), &error_fatal);
|
|
|
|
object_property_set_int(OBJECT(&chip9->xive), "pc-bar",
|
|
|
|
PNV9_XIVE_PC_BASE(chip), &error_fatal);
|
|
|
|
object_property_set_int(OBJECT(&chip9->xive), "tm-bar",
|
|
|
|
PNV9_XIVE_TM_BASE(chip), &error_fatal);
|
|
|
|
object_property_set_link(OBJECT(&chip9->xive), "chip", OBJECT(chip),
|
2019-11-15 18:56:05 +03:00
|
|
|
&error_abort);
|
error: Eliminate error_propagate() with Coccinelle, part 1
When all we do with an Error we receive into a local variable is
propagating to somewhere else, we can just as well receive it there
right away. Convert
if (!foo(..., &err)) {
...
error_propagate(errp, err);
...
return ...
}
to
if (!foo(..., errp)) {
...
...
return ...
}
where nothing else needs @err. Coccinelle script:
@rule1 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
binary operator op;
constant c1, c2;
symbol false;
@@
if (
(
- fun(args, &err, args2)
+ fun(args, errp, args2)
|
- !fun(args, &err, args2)
+ !fun(args, errp, args2)
|
- fun(args, &err, args2) op c1
+ fun(args, errp, args2) op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
)
}
@rule2 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
expression var;
binary operator op;
constant c1, c2;
symbol false;
@@
- var = fun(args, &err, args2);
+ var = fun(args, errp, args2);
... when != err
if (
(
var
|
!var
|
var op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
|
return var;
)
}
@depends on rule1 || rule2@
identifier err;
@@
- Error *err = NULL;
... when != err
Not exactly elegant, I'm afraid.
The "when != lbl:" is necessary to avoid transforming
if (fun(args, &err)) {
goto out
}
...
out:
error_propagate(errp, err);
even though other paths to label out still need the error_propagate().
For an actual example, see sclp_realize().
Without the "when strict", Coccinelle transforms vfio_msix_setup(),
incorrectly. I don't know what exactly "when strict" does, only that
it helps here.
The match of return is narrower than what I want, but I can't figure
out how to express "return where the operand doesn't use @err". For
an example where it's too narrow, see vfio_intx_enable().
Silently fails to convert hw/arm/armsse.c, because Coccinelle gets
confused by ARMSSE being used both as typedef and function-like macro
there. Converted manually.
Line breaks tidied up manually. One nested declaration of @local_err
deleted manually. Preexisting unwanted blank line dropped in
hw/riscv/sifive_e.c.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200707160613.848843-35-armbru@redhat.com>
2020-07-07 19:06:02 +03:00
|
|
|
if (!sysbus_realize(SYS_BUS_DEVICE(&chip9->xive), errp)) {
|
2019-03-06 11:50:11 +03:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
pnv_xscom_add_subregion(chip, PNV9_XSCOM_XIVE_BASE,
|
|
|
|
&chip9->xive.xscom_regs);
|
2019-03-08 01:35:35 +03:00
|
|
|
|
|
|
|
/* Processor Service Interface (PSI) Host Bridge */
|
qom: Put name parameter before value / visitor parameter
The object_property_set_FOO() setters take property name and value in
an unusual order:
void object_property_set_FOO(Object *obj, FOO_TYPE value,
const char *name, Error **errp)
Having to pass value before name feels grating. Swap them.
Same for object_property_set(), object_property_get(), and
object_property_parse().
Convert callers with this Coccinelle script:
@@
identifier fun = {
object_property_get, object_property_parse, object_property_set_str,
object_property_set_link, object_property_set_bool,
object_property_set_int, object_property_set_uint, object_property_set,
object_property_set_qobject
};
expression obj, v, name, errp;
@@
- fun(obj, v, name, errp)
+ fun(obj, name, v, errp)
Chokes on hw/arm/musicpal.c's lcd_refresh() with the unhelpful error
message "no position information". Convert that one manually.
Fails to convert hw/arm/armsse.c, because Coccinelle gets confused by
ARMSSE being used both as typedef and function-like macro there.
Convert manually.
Fails to convert hw/rx/rx-gdbsim.c, because Coccinelle gets confused
by RXCPU being used both as typedef and function-like macro there.
Convert manually. The other files using RXCPU that way don't need
conversion.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20200707160613.848843-27-armbru@redhat.com>
[Straightforwad conflict with commit 2336172d9b "audio: set default
value for pcspk.iobase property" resolved]
2020-07-07 19:05:54 +03:00
|
|
|
object_property_set_int(OBJECT(&chip9->psi), "bar", PNV9_PSIHB_BASE(chip),
|
|
|
|
&error_fatal);
|
2022-03-02 08:51:39 +03:00
|
|
|
/* This is the only device with 4k ESB pages */
|
|
|
|
object_property_set_int(OBJECT(&chip9->psi), "shift", XIVE_ESB_4K,
|
|
|
|
&error_fatal);
|
error: Eliminate error_propagate() with Coccinelle, part 1
When all we do with an Error we receive into a local variable is
propagating to somewhere else, we can just as well receive it there
right away. Convert
if (!foo(..., &err)) {
...
error_propagate(errp, err);
...
return ...
}
to
if (!foo(..., errp)) {
...
...
return ...
}
where nothing else needs @err. Coccinelle script:
@rule1 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
binary operator op;
constant c1, c2;
symbol false;
@@
if (
(
- fun(args, &err, args2)
+ fun(args, errp, args2)
|
- !fun(args, &err, args2)
+ !fun(args, errp, args2)
|
- fun(args, &err, args2) op c1
+ fun(args, errp, args2) op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
)
}
@rule2 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
expression var;
binary operator op;
constant c1, c2;
symbol false;
@@
- var = fun(args, &err, args2);
+ var = fun(args, errp, args2);
... when != err
if (
(
var
|
!var
|
var op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
|
return var;
)
}
@depends on rule1 || rule2@
identifier err;
@@
- Error *err = NULL;
... when != err
Not exactly elegant, I'm afraid.
The "when != lbl:" is necessary to avoid transforming
if (fun(args, &err)) {
goto out
}
...
out:
error_propagate(errp, err);
even though other paths to label out still need the error_propagate().
For an actual example, see sclp_realize().
Without the "when strict", Coccinelle transforms vfio_msix_setup(),
incorrectly. I don't know what exactly "when strict" does, only that
it helps here.
The match of return is narrower than what I want, but I can't figure
out how to express "return where the operand doesn't use @err". For
an example where it's too narrow, see vfio_intx_enable().
Silently fails to convert hw/arm/armsse.c, because Coccinelle gets
confused by ARMSSE being used both as typedef and function-like macro
there. Converted manually.
Line breaks tidied up manually. One nested declaration of @local_err
deleted manually. Preexisting unwanted blank line dropped in
hw/riscv/sifive_e.c.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200707160613.848843-35-armbru@redhat.com>
2020-07-07 19:06:02 +03:00
|
|
|
if (!qdev_realize(DEVICE(&chip9->psi), NULL, errp)) {
|
2019-03-08 01:35:35 +03:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
pnv_xscom_add_subregion(chip, PNV9_XSCOM_PSIHB_BASE,
|
|
|
|
&PNV_PSI(psi9)->xscom_regs);
|
2019-03-08 01:35:39 +03:00
|
|
|
|
|
|
|
/* LPC */
|
error: Eliminate error_propagate() with Coccinelle, part 1
When all we do with an Error we receive into a local variable is
propagating to somewhere else, we can just as well receive it there
right away. Convert
if (!foo(..., &err)) {
...
error_propagate(errp, err);
...
return ...
}
to
if (!foo(..., errp)) {
...
...
return ...
}
where nothing else needs @err. Coccinelle script:
@rule1 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
binary operator op;
constant c1, c2;
symbol false;
@@
if (
(
- fun(args, &err, args2)
+ fun(args, errp, args2)
|
- !fun(args, &err, args2)
+ !fun(args, errp, args2)
|
- fun(args, &err, args2) op c1
+ fun(args, errp, args2) op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
)
}
@rule2 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
expression var;
binary operator op;
constant c1, c2;
symbol false;
@@
- var = fun(args, &err, args2);
+ var = fun(args, errp, args2);
... when != err
if (
(
var
|
!var
|
var op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
|
return var;
)
}
@depends on rule1 || rule2@
identifier err;
@@
- Error *err = NULL;
... when != err
Not exactly elegant, I'm afraid.
The "when != lbl:" is necessary to avoid transforming
if (fun(args, &err)) {
goto out
}
...
out:
error_propagate(errp, err);
even though other paths to label out still need the error_propagate().
For an actual example, see sclp_realize().
Without the "when strict", Coccinelle transforms vfio_msix_setup(),
incorrectly. I don't know what exactly "when strict" does, only that
it helps here.
The match of return is narrower than what I want, but I can't figure
out how to express "return where the operand doesn't use @err". For
an example where it's too narrow, see vfio_intx_enable().
Silently fails to convert hw/arm/armsse.c, because Coccinelle gets
confused by ARMSSE being used both as typedef and function-like macro
there. Converted manually.
Line breaks tidied up manually. One nested declaration of @local_err
deleted manually. Preexisting unwanted blank line dropped in
hw/riscv/sifive_e.c.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200707160613.848843-35-armbru@redhat.com>
2020-07-07 19:06:02 +03:00
|
|
|
if (!qdev_realize(DEVICE(&chip9->lpc), NULL, errp)) {
|
2019-03-08 01:35:39 +03:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
memory_region_add_subregion(get_system_memory(), PNV9_LPCM_BASE(chip),
|
|
|
|
&chip9->lpc.xscom_regs);
|
|
|
|
|
2021-01-26 20:10:59 +03:00
|
|
|
chip->fw_mr = &chip9->lpc.isa_fw;
|
2019-03-08 01:35:39 +03:00
|
|
|
chip->dt_isa_nodename = g_strdup_printf("/lpcm-opb@%" PRIx64 "/lpc@0",
|
|
|
|
(uint64_t) PNV9_LPCM_BASE(chip));
|
2019-03-08 01:35:42 +03:00
|
|
|
|
|
|
|
/* Create the simplified OCC model */
|
error: Eliminate error_propagate() with Coccinelle, part 1
When all we do with an Error we receive into a local variable is
propagating to somewhere else, we can just as well receive it there
right away. Convert
if (!foo(..., &err)) {
...
error_propagate(errp, err);
...
return ...
}
to
if (!foo(..., errp)) {
...
...
return ...
}
where nothing else needs @err. Coccinelle script:
@rule1 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
binary operator op;
constant c1, c2;
symbol false;
@@
if (
(
- fun(args, &err, args2)
+ fun(args, errp, args2)
|
- !fun(args, &err, args2)
+ !fun(args, errp, args2)
|
- fun(args, &err, args2) op c1
+ fun(args, errp, args2) op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
)
}
@rule2 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
expression var;
binary operator op;
constant c1, c2;
symbol false;
@@
- var = fun(args, &err, args2);
+ var = fun(args, errp, args2);
... when != err
if (
(
var
|
!var
|
var op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
|
return var;
)
}
@depends on rule1 || rule2@
identifier err;
@@
- Error *err = NULL;
... when != err
Not exactly elegant, I'm afraid.
The "when != lbl:" is necessary to avoid transforming
if (fun(args, &err)) {
goto out
}
...
out:
error_propagate(errp, err);
even though other paths to label out still need the error_propagate().
For an actual example, see sclp_realize().
Without the "when strict", Coccinelle transforms vfio_msix_setup(),
incorrectly. I don't know what exactly "when strict" does, only that
it helps here.
The match of return is narrower than what I want, but I can't figure
out how to express "return where the operand doesn't use @err". For
an example where it's too narrow, see vfio_intx_enable().
Silently fails to convert hw/arm/armsse.c, because Coccinelle gets
confused by ARMSSE being used both as typedef and function-like macro
there. Converted manually.
Line breaks tidied up manually. One nested declaration of @local_err
deleted manually. Preexisting unwanted blank line dropped in
hw/riscv/sifive_e.c.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200707160613.848843-35-armbru@redhat.com>
2020-07-07 19:06:02 +03:00
|
|
|
if (!qdev_realize(DEVICE(&chip9->occ), NULL, errp)) {
|
2019-03-08 01:35:42 +03:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
pnv_xscom_add_subregion(chip, PNV9_XSCOM_OCC_BASE, &chip9->occ.xscom_regs);
|
2022-03-23 10:28:44 +03:00
|
|
|
qdev_connect_gpio_out(DEVICE(&chip9->occ), 0, qdev_get_gpio_in(
|
|
|
|
DEVICE(&chip9->psi), PSIHB9_IRQ_OCC));
|
2019-09-12 12:30:53 +03:00
|
|
|
|
|
|
|
/* OCC SRAM model */
|
2019-12-11 11:29:12 +03:00
|
|
|
memory_region_add_subregion(get_system_memory(), PNV9_OCC_SENSOR_BASE(chip),
|
2019-09-12 12:30:53 +03:00
|
|
|
&chip9->occ.sram_regs);
|
2019-09-12 12:30:54 +03:00
|
|
|
|
2022-08-11 12:37:26 +03:00
|
|
|
/* SBE */
|
|
|
|
if (!qdev_realize(DEVICE(&chip9->sbe), NULL, errp)) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
pnv_xscom_add_subregion(chip, PNV9_XSCOM_SBE_CTRL_BASE,
|
|
|
|
&chip9->sbe.xscom_ctrl_regs);
|
|
|
|
pnv_xscom_add_subregion(chip, PNV9_XSCOM_SBE_MBOX_BASE,
|
|
|
|
&chip9->sbe.xscom_mbox_regs);
|
|
|
|
qdev_connect_gpio_out(DEVICE(&chip9->sbe), 0, qdev_get_gpio_in(
|
|
|
|
DEVICE(&chip9->psi), PSIHB9_IRQ_PSU));
|
|
|
|
|
2019-09-12 12:30:54 +03:00
|
|
|
/* HOMER */
|
qom: Put name parameter before value / visitor parameter
The object_property_set_FOO() setters take property name and value in
an unusual order:
void object_property_set_FOO(Object *obj, FOO_TYPE value,
const char *name, Error **errp)
Having to pass value before name feels grating. Swap them.
Same for object_property_set(), object_property_get(), and
object_property_parse().
Convert callers with this Coccinelle script:
@@
identifier fun = {
object_property_get, object_property_parse, object_property_set_str,
object_property_set_link, object_property_set_bool,
object_property_set_int, object_property_set_uint, object_property_set,
object_property_set_qobject
};
expression obj, v, name, errp;
@@
- fun(obj, v, name, errp)
+ fun(obj, name, v, errp)
Chokes on hw/arm/musicpal.c's lcd_refresh() with the unhelpful error
message "no position information". Convert that one manually.
Fails to convert hw/arm/armsse.c, because Coccinelle gets confused by
ARMSSE being used both as typedef and function-like macro there.
Convert manually.
Fails to convert hw/rx/rx-gdbsim.c, because Coccinelle gets confused
by RXCPU being used both as typedef and function-like macro there.
Convert manually. The other files using RXCPU that way don't need
conversion.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20200707160613.848843-27-armbru@redhat.com>
[Straightforwad conflict with commit 2336172d9b "audio: set default
value for pcspk.iobase property" resolved]
2020-07-07 19:05:54 +03:00
|
|
|
object_property_set_link(OBJECT(&chip9->homer), "chip", OBJECT(chip),
|
2019-11-15 18:55:54 +03:00
|
|
|
&error_abort);
|
error: Eliminate error_propagate() with Coccinelle, part 1
When all we do with an Error we receive into a local variable is
propagating to somewhere else, we can just as well receive it there
right away. Convert
if (!foo(..., &err)) {
...
error_propagate(errp, err);
...
return ...
}
to
if (!foo(..., errp)) {
...
...
return ...
}
where nothing else needs @err. Coccinelle script:
@rule1 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
binary operator op;
constant c1, c2;
symbol false;
@@
if (
(
- fun(args, &err, args2)
+ fun(args, errp, args2)
|
- !fun(args, &err, args2)
+ !fun(args, errp, args2)
|
- fun(args, &err, args2) op c1
+ fun(args, errp, args2) op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
)
}
@rule2 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
expression var;
binary operator op;
constant c1, c2;
symbol false;
@@
- var = fun(args, &err, args2);
+ var = fun(args, errp, args2);
... when != err
if (
(
var
|
!var
|
var op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
|
return var;
)
}
@depends on rule1 || rule2@
identifier err;
@@
- Error *err = NULL;
... when != err
Not exactly elegant, I'm afraid.
The "when != lbl:" is necessary to avoid transforming
if (fun(args, &err)) {
goto out
}
...
out:
error_propagate(errp, err);
even though other paths to label out still need the error_propagate().
For an actual example, see sclp_realize().
Without the "when strict", Coccinelle transforms vfio_msix_setup(),
incorrectly. I don't know what exactly "when strict" does, only that
it helps here.
The match of return is narrower than what I want, but I can't figure
out how to express "return where the operand doesn't use @err". For
an example where it's too narrow, see vfio_intx_enable().
Silently fails to convert hw/arm/armsse.c, because Coccinelle gets
confused by ARMSSE being used both as typedef and function-like macro
there. Converted manually.
Line breaks tidied up manually. One nested declaration of @local_err
deleted manually. Preexisting unwanted blank line dropped in
hw/riscv/sifive_e.c.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200707160613.848843-35-armbru@redhat.com>
2020-07-07 19:06:02 +03:00
|
|
|
if (!qdev_realize(DEVICE(&chip9->homer), NULL, errp)) {
|
2019-09-12 12:30:54 +03:00
|
|
|
return;
|
|
|
|
}
|
2019-12-11 11:29:11 +03:00
|
|
|
/* Homer Xscom region */
|
|
|
|
pnv_xscom_add_subregion(chip, PNV9_XSCOM_PBA_BASE, &chip9->homer.pba_regs);
|
|
|
|
|
|
|
|
/* Homer mmio region */
|
2019-09-12 12:30:54 +03:00
|
|
|
memory_region_add_subregion(get_system_memory(), PNV9_HOMER_BASE(chip),
|
|
|
|
&chip9->homer.regs);
|
ppc/pnv: Add models for POWER9 PHB4 PCIe Host bridge
These changes introduces models for the PCIe Host Bridge (PHB4) of the
POWER9 processor. It includes the PowerBus logic interface (PBCQ),
IOMMU support, a single PCIe Gen.4 Root Complex, and support for MSI
and LSI interrupt sources as found on a POWER9 system using the XIVE
interrupt controller.
POWER9 processor comes with 3 PHB4 PEC (PCI Express Controller) and
each PEC can have several PHBs. By default,
* PEC0 provides 1 PHB (PHB0)
* PEC1 provides 2 PHBs (PHB1 and PHB2)
* PEC2 provides 3 PHBs (PHB3, PHB4 and PHB5)
Each PEC has a set "global" registers and some "per-stack" (per-PHB)
registers. Those are organized in two XSCOM ranges, the "Nest" range
and the "PCI" range, each range contains both some "PEC" registers and
some "per-stack" registers.
No default device layout is provided and PCI devices can be added on
any of the available PCIe Root Port (pcie.0 .. 2 of a Power9 chip)
with address 0x0 as the firwware (skiboot) only accepts a single
device per root port. To run a simple system with a network and a
storage adapters, use a command line options such as :
-device e1000e,netdev=net0,mac=C0:FF:EE:00:00:02,bus=pcie.0,addr=0x0
-netdev bridge,id=net0,helper=/usr/libexec/qemu-bridge-helper,br=virbr0,id=hostnet0
-device megasas,id=scsi0,bus=pcie.1,addr=0x0
-drive file=$disk,if=none,id=drive-scsi0-0-0-0,format=qcow2,cache=none
-device scsi-hd,bus=scsi0.0,channel=0,scsi-id=0,lun=0,drive=drive-scsi0-0-0-0,id=scsi0-0-0-0,bootindex=2
If more are needed, include a bridge.
Multi chip is supported, each chip adding its set of PHB4 controllers
and its PCI busses. The model doesn't emulate the EEH error handling.
This model is not ready for hotplug yet.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[ clg: - numerous cleanups
- commit log
- fix for broken LSI support
- PHB pic printinfo
- large QOM rework ]
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20200127144506.11132-2-clg@kaod.org>
[dwg: Use device_class_set_props()]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2020-01-27 17:45:05 +03:00
|
|
|
|
2021-12-17 19:57:19 +03:00
|
|
|
/* PEC PHBs */
|
|
|
|
pnv_chip_power9_pec_realize(chip, &local_err);
|
ppc/pnv: Add models for POWER9 PHB4 PCIe Host bridge
These changes introduces models for the PCIe Host Bridge (PHB4) of the
POWER9 processor. It includes the PowerBus logic interface (PBCQ),
IOMMU support, a single PCIe Gen.4 Root Complex, and support for MSI
and LSI interrupt sources as found on a POWER9 system using the XIVE
interrupt controller.
POWER9 processor comes with 3 PHB4 PEC (PCI Express Controller) and
each PEC can have several PHBs. By default,
* PEC0 provides 1 PHB (PHB0)
* PEC1 provides 2 PHBs (PHB1 and PHB2)
* PEC2 provides 3 PHBs (PHB3, PHB4 and PHB5)
Each PEC has a set "global" registers and some "per-stack" (per-PHB)
registers. Those are organized in two XSCOM ranges, the "Nest" range
and the "PCI" range, each range contains both some "PEC" registers and
some "per-stack" registers.
No default device layout is provided and PCI devices can be added on
any of the available PCIe Root Port (pcie.0 .. 2 of a Power9 chip)
with address 0x0 as the firwware (skiboot) only accepts a single
device per root port. To run a simple system with a network and a
storage adapters, use a command line options such as :
-device e1000e,netdev=net0,mac=C0:FF:EE:00:00:02,bus=pcie.0,addr=0x0
-netdev bridge,id=net0,helper=/usr/libexec/qemu-bridge-helper,br=virbr0,id=hostnet0
-device megasas,id=scsi0,bus=pcie.1,addr=0x0
-drive file=$disk,if=none,id=drive-scsi0-0-0-0,format=qcow2,cache=none
-device scsi-hd,bus=scsi0.0,channel=0,scsi-id=0,lun=0,drive=drive-scsi0-0-0-0,id=scsi0-0-0-0,bootindex=2
If more are needed, include a bridge.
Multi chip is supported, each chip adding its set of PHB4 controllers
and its PCI busses. The model doesn't emulate the EEH error handling.
This model is not ready for hotplug yet.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[ clg: - numerous cleanups
- commit log
- fix for broken LSI support
- PHB pic printinfo
- large QOM rework ]
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20200127144506.11132-2-clg@kaod.org>
[dwg: Use device_class_set_props()]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2020-01-27 17:45:05 +03:00
|
|
|
if (local_err) {
|
|
|
|
error_propagate(errp, local_err);
|
|
|
|
return;
|
|
|
|
}
|
2016-10-22 12:46:36 +03:00
|
|
|
}
|
|
|
|
|
2019-12-13 15:00:35 +03:00
|
|
|
static uint32_t pnv_chip_power9_xscom_pcba(PnvChip *chip, uint64_t addr)
|
|
|
|
{
|
|
|
|
addr &= (PNV9_XSCOM_SIZE - 1);
|
|
|
|
return addr >> 3;
|
|
|
|
}
|
|
|
|
|
2016-10-22 12:46:36 +03:00
|
|
|
static void pnv_chip_power9_class_init(ObjectClass *klass, void *data)
|
|
|
|
{
|
|
|
|
DeviceClass *dc = DEVICE_CLASS(klass);
|
|
|
|
PnvChipClass *k = PNV_CHIP_CLASS(klass);
|
|
|
|
|
2018-01-15 21:04:01 +03:00
|
|
|
k->chip_cfam_id = 0x220d104900008000ull; /* P9 Nimbus DD2.0 */
|
2016-10-22 12:46:37 +03:00
|
|
|
k->cores_mask = POWER9_CORE_MASK;
|
2016-10-22 12:46:38 +03:00
|
|
|
k->core_pir = pnv_chip_core_pir_p9;
|
2018-06-15 18:25:33 +03:00
|
|
|
k->intc_create = pnv_chip_power9_intc_create;
|
2019-10-22 19:38:10 +03:00
|
|
|
k->intc_reset = pnv_chip_power9_intc_reset;
|
2019-10-24 17:27:22 +03:00
|
|
|
k->intc_destroy = pnv_chip_power9_intc_destroy;
|
2019-12-13 15:00:07 +03:00
|
|
|
k->intc_print_info = pnv_chip_power9_intc_print_info;
|
2018-06-15 18:25:34 +03:00
|
|
|
k->isa_create = pnv_chip_power9_isa_create;
|
2019-03-06 11:50:12 +03:00
|
|
|
k->dt_populate = pnv_chip_power9_dt_populate;
|
2019-03-06 11:50:13 +03:00
|
|
|
k->pic_print_info = pnv_chip_power9_pic_print_info;
|
2019-12-13 15:00:13 +03:00
|
|
|
k->xscom_core_base = pnv_chip_power9_xscom_core_base;
|
2019-12-13 15:00:35 +03:00
|
|
|
k->xscom_pcba = pnv_chip_power9_xscom_pcba;
|
2016-10-22 12:46:36 +03:00
|
|
|
dc->desc = "PowerNV Chip POWER9";
|
2021-12-17 19:57:19 +03:00
|
|
|
k->num_pecs = PNV9_CHIP_MAX_PEC;
|
2018-06-18 20:05:39 +03:00
|
|
|
|
|
|
|
device_class_set_parent_realize(dc, pnv_chip_power9_realize,
|
|
|
|
&k->parent_realize);
|
2016-10-22 12:46:36 +03:00
|
|
|
}
|
|
|
|
|
2019-12-05 21:44:51 +03:00
|
|
|
static void pnv_chip_power10_instance_init(Object *obj)
|
|
|
|
{
|
2022-03-02 08:51:39 +03:00
|
|
|
PnvChip *chip = PNV_CHIP(obj);
|
2019-12-05 21:44:53 +03:00
|
|
|
Pnv10Chip *chip10 = PNV10_CHIP(obj);
|
2022-03-02 08:51:39 +03:00
|
|
|
PnvChipClass *pcc = PNV_CHIP_GET_CLASS(obj);
|
|
|
|
int i;
|
2019-12-05 21:44:53 +03:00
|
|
|
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 08:51:38 +03:00
|
|
|
object_initialize_child(obj, "xive", &chip10->xive, TYPE_PNV_XIVE2);
|
|
|
|
object_property_add_alias(obj, "xive-fabric", OBJECT(&chip10->xive),
|
|
|
|
"xive-fabric");
|
qom: Less verbose object_initialize_child()
All users of object_initialize_child() pass the obvious child size
argument. Almost all pass &error_abort and no properties. Tiresome.
Rename object_initialize_child() to
object_initialize_child_with_props() to free the name. New
convenience wrapper object_initialize_child() automates the size
argument, and passes &error_abort and no properties.
Rename object_initialize_childv() to
object_initialize_child_with_propsv() for consistency.
Convert callers with this Coccinelle script:
@@
expression parent, propname, type;
expression child, size;
symbol error_abort;
@@
- object_initialize_child(parent, propname, OBJECT(child), size, type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, size, type, &error_abort, NULL)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, child, sizeof(*child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, child, type)
@@
expression parent, propname, type;
expression child;
symbol error_abort;
@@
- object_initialize_child(parent, propname, &child, sizeof(child), type, &error_abort, NULL)
+ object_initialize_child(parent, propname, &child, type)
@@
expression parent, propname, type;
expression child, size, err;
expression list props;
@@
- object_initialize_child(parent, propname, child, size, type, err, props)
+ object_initialize_child_with_props(parent, propname, child, size, type, err, props)
Note that Coccinelle chokes on ARMSSE typedef vs. macro in
hw/arm/armsse.c. Worked around by temporarily renaming the macro for
the spatch run.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
[Rebased: machine opentitan is new (commit fe0fe4735e7)]
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200610053247.1583243-37-armbru@redhat.com>
2020-06-10 08:32:25 +03:00
|
|
|
object_initialize_child(obj, "psi", &chip10->psi, TYPE_PNV10_PSI);
|
|
|
|
object_initialize_child(obj, "lpc", &chip10->lpc, TYPE_PNV10_LPC);
|
2022-03-02 08:51:39 +03:00
|
|
|
object_initialize_child(obj, "occ", &chip10->occ, TYPE_PNV10_OCC);
|
2022-08-11 12:37:26 +03:00
|
|
|
object_initialize_child(obj, "sbe", &chip10->sbe, TYPE_PNV10_SBE);
|
2022-03-02 08:51:39 +03:00
|
|
|
object_initialize_child(obj, "homer", &chip10->homer, TYPE_PNV10_HOMER);
|
2022-03-02 08:51:39 +03:00
|
|
|
|
2022-03-14 17:57:17 +03:00
|
|
|
chip->num_pecs = pcc->num_pecs;
|
2022-03-02 08:51:39 +03:00
|
|
|
|
|
|
|
for (i = 0; i < chip->num_pecs; i++) {
|
|
|
|
object_initialize_child(obj, "pec[*]", &chip10->pecs[i],
|
|
|
|
TYPE_PNV_PHB5_PEC);
|
|
|
|
}
|
2019-12-05 21:44:51 +03:00
|
|
|
}
|
|
|
|
|
2022-03-02 08:51:39 +03:00
|
|
|
static void pnv_chip_power10_quad_realize(Pnv10Chip *chip10, Error **errp)
|
|
|
|
{
|
|
|
|
PnvChip *chip = PNV_CHIP(chip10);
|
|
|
|
int i;
|
|
|
|
|
|
|
|
chip10->nr_quads = DIV_ROUND_UP(chip->nr_cores, 4);
|
|
|
|
chip10->quads = g_new0(PnvQuad, chip10->nr_quads);
|
|
|
|
|
|
|
|
for (i = 0; i < chip10->nr_quads; i++) {
|
|
|
|
PnvQuad *eq = &chip10->quads[i];
|
|
|
|
|
|
|
|
pnv_chip_quad_realize_one(chip, eq, chip->cores[i * 4]);
|
|
|
|
|
|
|
|
pnv_xscom_add_subregion(chip, PNV10_XSCOM_EQ_BASE(eq->quad_id),
|
|
|
|
&eq->xscom_regs);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-03-02 08:51:39 +03:00
|
|
|
static void pnv_chip_power10_phb_realize(PnvChip *chip, Error **errp)
|
|
|
|
{
|
|
|
|
Pnv10Chip *chip10 = PNV10_CHIP(chip);
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = 0; i < chip->num_pecs; i++) {
|
|
|
|
PnvPhb4PecState *pec = &chip10->pecs[i];
|
|
|
|
PnvPhb4PecClass *pecc = PNV_PHB4_PEC_GET_CLASS(pec);
|
|
|
|
uint32_t pec_nest_base;
|
|
|
|
uint32_t pec_pci_base;
|
|
|
|
|
|
|
|
object_property_set_int(OBJECT(pec), "index", i, &error_fatal);
|
|
|
|
object_property_set_int(OBJECT(pec), "chip-id", chip->chip_id,
|
|
|
|
&error_fatal);
|
|
|
|
object_property_set_link(OBJECT(pec), "chip", OBJECT(chip),
|
|
|
|
&error_fatal);
|
|
|
|
if (!qdev_realize(DEVICE(pec), NULL, errp)) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
pec_nest_base = pecc->xscom_nest_base(pec);
|
|
|
|
pec_pci_base = pecc->xscom_pci_base(pec);
|
|
|
|
|
|
|
|
pnv_xscom_add_subregion(chip, pec_nest_base, &pec->nest_regs_mr);
|
|
|
|
pnv_xscom_add_subregion(chip, pec_pci_base, &pec->pci_regs_mr);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2019-12-05 21:44:51 +03:00
|
|
|
static void pnv_chip_power10_realize(DeviceState *dev, Error **errp)
|
|
|
|
{
|
|
|
|
PnvChipClass *pcc = PNV_CHIP_GET_CLASS(dev);
|
|
|
|
PnvChip *chip = PNV_CHIP(dev);
|
2019-12-05 21:44:53 +03:00
|
|
|
Pnv10Chip *chip10 = PNV10_CHIP(dev);
|
2019-12-05 21:44:51 +03:00
|
|
|
Error *local_err = NULL;
|
|
|
|
|
|
|
|
/* XSCOM bridge is first */
|
|
|
|
pnv_xscom_realize(chip, PNV10_XSCOM_SIZE, &local_err);
|
|
|
|
if (local_err) {
|
|
|
|
error_propagate(errp, local_err);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
sysbus_mmio_map(SYS_BUS_DEVICE(chip), 0, PNV10_XSCOM_BASE(chip));
|
|
|
|
|
|
|
|
pcc->parent_realize(dev, &local_err);
|
|
|
|
if (local_err) {
|
|
|
|
error_propagate(errp, local_err);
|
|
|
|
return;
|
|
|
|
}
|
2019-12-05 21:44:53 +03:00
|
|
|
|
2022-03-02 08:51:39 +03:00
|
|
|
pnv_chip_power10_quad_realize(chip10, &local_err);
|
|
|
|
if (local_err) {
|
|
|
|
error_propagate(errp, local_err);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 08:51:38 +03:00
|
|
|
/* XIVE2 interrupt controller (POWER10) */
|
|
|
|
object_property_set_int(OBJECT(&chip10->xive), "ic-bar",
|
|
|
|
PNV10_XIVE2_IC_BASE(chip), &error_fatal);
|
|
|
|
object_property_set_int(OBJECT(&chip10->xive), "esb-bar",
|
|
|
|
PNV10_XIVE2_ESB_BASE(chip), &error_fatal);
|
|
|
|
object_property_set_int(OBJECT(&chip10->xive), "end-bar",
|
|
|
|
PNV10_XIVE2_END_BASE(chip), &error_fatal);
|
|
|
|
object_property_set_int(OBJECT(&chip10->xive), "nvpg-bar",
|
|
|
|
PNV10_XIVE2_NVPG_BASE(chip), &error_fatal);
|
|
|
|
object_property_set_int(OBJECT(&chip10->xive), "nvc-bar",
|
|
|
|
PNV10_XIVE2_NVC_BASE(chip), &error_fatal);
|
|
|
|
object_property_set_int(OBJECT(&chip10->xive), "tm-bar",
|
|
|
|
PNV10_XIVE2_TM_BASE(chip), &error_fatal);
|
|
|
|
object_property_set_link(OBJECT(&chip10->xive), "chip", OBJECT(chip),
|
|
|
|
&error_abort);
|
|
|
|
if (!sysbus_realize(SYS_BUS_DEVICE(&chip10->xive), errp)) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
pnv_xscom_add_subregion(chip, PNV10_XSCOM_XIVE2_BASE,
|
|
|
|
&chip10->xive.xscom_regs);
|
|
|
|
|
2019-12-05 21:44:53 +03:00
|
|
|
/* Processor Service Interface (PSI) Host Bridge */
|
qom: Put name parameter before value / visitor parameter
The object_property_set_FOO() setters take property name and value in
an unusual order:
void object_property_set_FOO(Object *obj, FOO_TYPE value,
const char *name, Error **errp)
Having to pass value before name feels grating. Swap them.
Same for object_property_set(), object_property_get(), and
object_property_parse().
Convert callers with this Coccinelle script:
@@
identifier fun = {
object_property_get, object_property_parse, object_property_set_str,
object_property_set_link, object_property_set_bool,
object_property_set_int, object_property_set_uint, object_property_set,
object_property_set_qobject
};
expression obj, v, name, errp;
@@
- fun(obj, v, name, errp)
+ fun(obj, name, v, errp)
Chokes on hw/arm/musicpal.c's lcd_refresh() with the unhelpful error
message "no position information". Convert that one manually.
Fails to convert hw/arm/armsse.c, because Coccinelle gets confused by
ARMSSE being used both as typedef and function-like macro there.
Convert manually.
Fails to convert hw/rx/rx-gdbsim.c, because Coccinelle gets confused
by RXCPU being used both as typedef and function-like macro there.
Convert manually. The other files using RXCPU that way don't need
conversion.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20200707160613.848843-27-armbru@redhat.com>
[Straightforwad conflict with commit 2336172d9b "audio: set default
value for pcspk.iobase property" resolved]
2020-07-07 19:05:54 +03:00
|
|
|
object_property_set_int(OBJECT(&chip10->psi), "bar",
|
|
|
|
PNV10_PSIHB_BASE(chip), &error_fatal);
|
2022-03-02 08:51:39 +03:00
|
|
|
/* PSI can now be configured to use 64k ESB pages on POWER10 */
|
|
|
|
object_property_set_int(OBJECT(&chip10->psi), "shift", XIVE_ESB_64K,
|
|
|
|
&error_fatal);
|
error: Eliminate error_propagate() with Coccinelle, part 1
When all we do with an Error we receive into a local variable is
propagating to somewhere else, we can just as well receive it there
right away. Convert
if (!foo(..., &err)) {
...
error_propagate(errp, err);
...
return ...
}
to
if (!foo(..., errp)) {
...
...
return ...
}
where nothing else needs @err. Coccinelle script:
@rule1 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
binary operator op;
constant c1, c2;
symbol false;
@@
if (
(
- fun(args, &err, args2)
+ fun(args, errp, args2)
|
- !fun(args, &err, args2)
+ !fun(args, errp, args2)
|
- fun(args, &err, args2) op c1
+ fun(args, errp, args2) op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
)
}
@rule2 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
expression var;
binary operator op;
constant c1, c2;
symbol false;
@@
- var = fun(args, &err, args2);
+ var = fun(args, errp, args2);
... when != err
if (
(
var
|
!var
|
var op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
|
return var;
)
}
@depends on rule1 || rule2@
identifier err;
@@
- Error *err = NULL;
... when != err
Not exactly elegant, I'm afraid.
The "when != lbl:" is necessary to avoid transforming
if (fun(args, &err)) {
goto out
}
...
out:
error_propagate(errp, err);
even though other paths to label out still need the error_propagate().
For an actual example, see sclp_realize().
Without the "when strict", Coccinelle transforms vfio_msix_setup(),
incorrectly. I don't know what exactly "when strict" does, only that
it helps here.
The match of return is narrower than what I want, but I can't figure
out how to express "return where the operand doesn't use @err". For
an example where it's too narrow, see vfio_intx_enable().
Silently fails to convert hw/arm/armsse.c, because Coccinelle gets
confused by ARMSSE being used both as typedef and function-like macro
there. Converted manually.
Line breaks tidied up manually. One nested declaration of @local_err
deleted manually. Preexisting unwanted blank line dropped in
hw/riscv/sifive_e.c.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200707160613.848843-35-armbru@redhat.com>
2020-07-07 19:06:02 +03:00
|
|
|
if (!qdev_realize(DEVICE(&chip10->psi), NULL, errp)) {
|
2019-12-05 21:44:53 +03:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
pnv_xscom_add_subregion(chip, PNV10_XSCOM_PSIHB_BASE,
|
|
|
|
&PNV_PSI(&chip10->psi)->xscom_regs);
|
2019-12-05 21:44:54 +03:00
|
|
|
|
|
|
|
/* LPC */
|
error: Eliminate error_propagate() with Coccinelle, part 1
When all we do with an Error we receive into a local variable is
propagating to somewhere else, we can just as well receive it there
right away. Convert
if (!foo(..., &err)) {
...
error_propagate(errp, err);
...
return ...
}
to
if (!foo(..., errp)) {
...
...
return ...
}
where nothing else needs @err. Coccinelle script:
@rule1 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
binary operator op;
constant c1, c2;
symbol false;
@@
if (
(
- fun(args, &err, args2)
+ fun(args, errp, args2)
|
- !fun(args, &err, args2)
+ !fun(args, errp, args2)
|
- fun(args, &err, args2) op c1
+ fun(args, errp, args2) op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
)
}
@rule2 forall@
identifier fun, err, errp, lbl;
expression list args, args2;
expression var;
binary operator op;
constant c1, c2;
symbol false;
@@
- var = fun(args, &err, args2);
+ var = fun(args, errp, args2);
... when != err
if (
(
var
|
!var
|
var op c1
)
)
{
... when != err
when != lbl:
when strict
- error_propagate(errp, err);
... when != err
(
return;
|
return c2;
|
return false;
|
return var;
)
}
@depends on rule1 || rule2@
identifier err;
@@
- Error *err = NULL;
... when != err
Not exactly elegant, I'm afraid.
The "when != lbl:" is necessary to avoid transforming
if (fun(args, &err)) {
goto out
}
...
out:
error_propagate(errp, err);
even though other paths to label out still need the error_propagate().
For an actual example, see sclp_realize().
Without the "when strict", Coccinelle transforms vfio_msix_setup(),
incorrectly. I don't know what exactly "when strict" does, only that
it helps here.
The match of return is narrower than what I want, but I can't figure
out how to express "return where the operand doesn't use @err". For
an example where it's too narrow, see vfio_intx_enable().
Silently fails to convert hw/arm/armsse.c, because Coccinelle gets
confused by ARMSSE being used both as typedef and function-like macro
there. Converted manually.
Line breaks tidied up manually. One nested declaration of @local_err
deleted manually. Preexisting unwanted blank line dropped in
hw/riscv/sifive_e.c.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200707160613.848843-35-armbru@redhat.com>
2020-07-07 19:06:02 +03:00
|
|
|
if (!qdev_realize(DEVICE(&chip10->lpc), NULL, errp)) {
|
2019-12-05 21:44:54 +03:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
memory_region_add_subregion(get_system_memory(), PNV10_LPCM_BASE(chip),
|
|
|
|
&chip10->lpc.xscom_regs);
|
|
|
|
|
2021-01-26 20:10:59 +03:00
|
|
|
chip->fw_mr = &chip10->lpc.isa_fw;
|
2019-12-05 21:44:54 +03:00
|
|
|
chip->dt_isa_nodename = g_strdup_printf("/lpcm-opb@%" PRIx64 "/lpc@0",
|
|
|
|
(uint64_t) PNV10_LPCM_BASE(chip));
|
2022-03-02 08:51:39 +03:00
|
|
|
|
|
|
|
/* Create the simplified OCC model */
|
|
|
|
if (!qdev_realize(DEVICE(&chip10->occ), NULL, errp)) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
pnv_xscom_add_subregion(chip, PNV10_XSCOM_OCC_BASE,
|
|
|
|
&chip10->occ.xscom_regs);
|
2022-03-23 10:28:44 +03:00
|
|
|
qdev_connect_gpio_out(DEVICE(&chip10->occ), 0, qdev_get_gpio_in(
|
|
|
|
DEVICE(&chip10->psi), PSIHB9_IRQ_OCC));
|
2022-03-02 08:51:39 +03:00
|
|
|
|
2022-03-02 08:51:39 +03:00
|
|
|
/* OCC SRAM model */
|
|
|
|
memory_region_add_subregion(get_system_memory(),
|
|
|
|
PNV10_OCC_SENSOR_BASE(chip),
|
|
|
|
&chip10->occ.sram_regs);
|
|
|
|
|
2022-08-11 12:37:26 +03:00
|
|
|
/* SBE */
|
|
|
|
if (!qdev_realize(DEVICE(&chip10->sbe), NULL, errp)) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
pnv_xscom_add_subregion(chip, PNV10_XSCOM_SBE_CTRL_BASE,
|
|
|
|
&chip10->sbe.xscom_ctrl_regs);
|
|
|
|
pnv_xscom_add_subregion(chip, PNV10_XSCOM_SBE_MBOX_BASE,
|
|
|
|
&chip10->sbe.xscom_mbox_regs);
|
|
|
|
qdev_connect_gpio_out(DEVICE(&chip10->sbe), 0, qdev_get_gpio_in(
|
|
|
|
DEVICE(&chip10->psi), PSIHB9_IRQ_PSU));
|
|
|
|
|
2022-03-02 08:51:39 +03:00
|
|
|
/* HOMER */
|
|
|
|
object_property_set_link(OBJECT(&chip10->homer), "chip", OBJECT(chip),
|
|
|
|
&error_abort);
|
|
|
|
if (!qdev_realize(DEVICE(&chip10->homer), NULL, errp)) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
/* Homer Xscom region */
|
|
|
|
pnv_xscom_add_subregion(chip, PNV10_XSCOM_PBA_BASE,
|
|
|
|
&chip10->homer.pba_regs);
|
|
|
|
|
|
|
|
/* Homer mmio region */
|
|
|
|
memory_region_add_subregion(get_system_memory(), PNV10_HOMER_BASE(chip),
|
|
|
|
&chip10->homer.regs);
|
|
|
|
|
2022-03-02 08:51:39 +03:00
|
|
|
/* PHBs */
|
|
|
|
pnv_chip_power10_phb_realize(chip, &local_err);
|
|
|
|
if (local_err) {
|
|
|
|
error_propagate(errp, local_err);
|
|
|
|
return;
|
|
|
|
}
|
2019-12-05 21:44:51 +03:00
|
|
|
}
|
|
|
|
|
2019-12-13 15:00:35 +03:00
|
|
|
static uint32_t pnv_chip_power10_xscom_pcba(PnvChip *chip, uint64_t addr)
|
|
|
|
{
|
|
|
|
addr &= (PNV10_XSCOM_SIZE - 1);
|
|
|
|
return addr >> 3;
|
|
|
|
}
|
|
|
|
|
2019-12-05 21:44:51 +03:00
|
|
|
static void pnv_chip_power10_class_init(ObjectClass *klass, void *data)
|
|
|
|
{
|
|
|
|
DeviceClass *dc = DEVICE_CLASS(klass);
|
|
|
|
PnvChipClass *k = PNV_CHIP_CLASS(klass);
|
|
|
|
|
|
|
|
k->chip_cfam_id = 0x120da04900008000ull; /* P10 DD1.0 (with NX) */
|
|
|
|
k->cores_mask = POWER10_CORE_MASK;
|
|
|
|
k->core_pir = pnv_chip_core_pir_p10;
|
|
|
|
k->intc_create = pnv_chip_power10_intc_create;
|
|
|
|
k->intc_reset = pnv_chip_power10_intc_reset;
|
|
|
|
k->intc_destroy = pnv_chip_power10_intc_destroy;
|
2019-12-13 15:00:07 +03:00
|
|
|
k->intc_print_info = pnv_chip_power10_intc_print_info;
|
2019-12-05 21:44:51 +03:00
|
|
|
k->isa_create = pnv_chip_power10_isa_create;
|
|
|
|
k->dt_populate = pnv_chip_power10_dt_populate;
|
|
|
|
k->pic_print_info = pnv_chip_power10_pic_print_info;
|
2019-12-13 15:00:13 +03:00
|
|
|
k->xscom_core_base = pnv_chip_power10_xscom_core_base;
|
2019-12-13 15:00:35 +03:00
|
|
|
k->xscom_pcba = pnv_chip_power10_xscom_pcba;
|
2019-12-05 21:44:51 +03:00
|
|
|
dc->desc = "PowerNV Chip POWER10";
|
2022-03-02 08:51:39 +03:00
|
|
|
k->num_pecs = PNV10_CHIP_MAX_PEC;
|
2019-12-05 21:44:51 +03:00
|
|
|
|
|
|
|
device_class_set_parent_realize(dc, pnv_chip_power10_realize,
|
|
|
|
&k->parent_realize);
|
|
|
|
}
|
|
|
|
|
2016-10-22 12:46:37 +03:00
|
|
|
static void pnv_chip_core_sanitize(PnvChip *chip, Error **errp)
|
|
|
|
{
|
|
|
|
PnvChipClass *pcc = PNV_CHIP_GET_CLASS(chip);
|
|
|
|
int cores_max;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* No custom mask for this chip, let's use the default one from *
|
|
|
|
* the chip class
|
|
|
|
*/
|
|
|
|
if (!chip->cores_mask) {
|
|
|
|
chip->cores_mask = pcc->cores_mask;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* filter alien core ids ! some are reserved */
|
|
|
|
if ((chip->cores_mask & pcc->cores_mask) != chip->cores_mask) {
|
|
|
|
error_setg(errp, "warning: invalid core mask for chip Ox%"PRIx64" !",
|
|
|
|
chip->cores_mask);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
chip->cores_mask &= pcc->cores_mask;
|
|
|
|
|
|
|
|
/* now that we have a sane layout, let check the number of cores */
|
2016-11-14 12:12:57 +03:00
|
|
|
cores_max = ctpop64(chip->cores_mask);
|
2016-10-22 12:46:37 +03:00
|
|
|
if (chip->nr_cores > cores_max) {
|
|
|
|
error_setg(errp, "warning: too many cores for chip ! Limit is %d",
|
|
|
|
cores_max);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-06-14 17:00:41 +03:00
|
|
|
static void pnv_chip_core_realize(PnvChip *chip, Error **errp)
|
2016-10-22 12:46:36 +03:00
|
|
|
{
|
2016-10-22 12:46:37 +03:00
|
|
|
Error *error = NULL;
|
2016-10-22 12:46:39 +03:00
|
|
|
PnvChipClass *pcc = PNV_CHIP_GET_CLASS(chip);
|
2017-10-09 22:51:10 +03:00
|
|
|
const char *typename = pnv_chip_core_typename(chip);
|
2016-10-22 12:46:39 +03:00
|
|
|
int i, core_hwid;
|
2020-01-27 17:41:54 +03:00
|
|
|
PnvMachineState *pnv = PNV_MACHINE(qdev_get_machine());
|
2016-10-22 12:46:39 +03:00
|
|
|
|
|
|
|
if (!object_class_by_name(typename)) {
|
|
|
|
error_setg(errp, "Unable to find PowerNV CPU Core '%s'", typename);
|
|
|
|
return;
|
|
|
|
}
|
2016-10-22 12:46:37 +03:00
|
|
|
|
2016-10-22 12:46:39 +03:00
|
|
|
/* Cores */
|
2016-10-22 12:46:37 +03:00
|
|
|
pnv_chip_core_sanitize(chip, &error);
|
|
|
|
if (error) {
|
|
|
|
error_propagate(errp, error);
|
|
|
|
return;
|
|
|
|
}
|
2016-10-22 12:46:39 +03:00
|
|
|
|
2019-11-25 09:58:03 +03:00
|
|
|
chip->cores = g_new0(PnvCore *, chip->nr_cores);
|
2016-10-22 12:46:39 +03:00
|
|
|
|
|
|
|
for (i = 0, core_hwid = 0; (core_hwid < sizeof(chip->cores_mask) * 8)
|
|
|
|
&& (i < chip->nr_cores); core_hwid++) {
|
|
|
|
char core_name[32];
|
2019-11-25 09:58:03 +03:00
|
|
|
PnvCore *pnv_core;
|
2018-01-15 21:04:04 +03:00
|
|
|
uint64_t xscom_core_base;
|
2016-10-22 12:46:39 +03:00
|
|
|
|
|
|
|
if (!(chip->cores_mask & (1ull << core_hwid))) {
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2019-11-25 09:58:03 +03:00
|
|
|
pnv_core = PNV_CORE(object_new(typename));
|
|
|
|
|
2016-10-22 12:46:39 +03:00
|
|
|
snprintf(core_name, sizeof(core_name), "core[%d]", core_hwid);
|
qom: Drop parameter @errp of object_property_add() & friends
The only way object_property_add() can fail is when a property with
the same name already exists. Since our property names are all
hardcoded, failure is a programming error, and the appropriate way to
handle it is passing &error_abort.
Same for its variants, except for object_property_add_child(), which
additionally fails when the child already has a parent. Parentage is
also under program control, so this is a programming error, too.
We have a bit over 500 callers. Almost half of them pass
&error_abort, slightly fewer ignore errors, one test case handles
errors, and the remaining few callers pass them to their own callers.
The previous few commits demonstrated once again that ignoring
programming errors is a bad idea.
Of the few ones that pass on errors, several violate the Error API.
The Error ** argument must be NULL, &error_abort, &error_fatal, or a
pointer to a variable containing NULL. Passing an argument of the
latter kind twice without clearing it in between is wrong: if the
first call sets an error, it no longer points to NULL for the second
call. ich9_pm_add_properties(), sparc32_ledma_realize(),
sparc32_dma_realize(), xilinx_axidma_realize(), xilinx_enet_realize()
are wrong that way.
When the one appropriate choice of argument is &error_abort, letting
users pick the argument is a bad idea.
Drop parameter @errp and assert the preconditions instead.
There's one exception to "duplicate property name is a programming
error": the way object_property_add() implements the magic (and
undocumented) "automatic arrayification". Don't drop @errp there.
Instead, rename object_property_add() to object_property_try_add(),
and add the obvious wrapper object_property_add().
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200505152926.18877-15-armbru@redhat.com>
[Two semantic rebase conflicts resolved]
2020-05-05 18:29:22 +03:00
|
|
|
object_property_add_child(OBJECT(chip), core_name, OBJECT(pnv_core));
|
2019-11-25 09:58:03 +03:00
|
|
|
chip->cores[i] = pnv_core;
|
qom: Put name parameter before value / visitor parameter
The object_property_set_FOO() setters take property name and value in
an unusual order:
void object_property_set_FOO(Object *obj, FOO_TYPE value,
const char *name, Error **errp)
Having to pass value before name feels grating. Swap them.
Same for object_property_set(), object_property_get(), and
object_property_parse().
Convert callers with this Coccinelle script:
@@
identifier fun = {
object_property_get, object_property_parse, object_property_set_str,
object_property_set_link, object_property_set_bool,
object_property_set_int, object_property_set_uint, object_property_set,
object_property_set_qobject
};
expression obj, v, name, errp;
@@
- fun(obj, v, name, errp)
+ fun(obj, name, v, errp)
Chokes on hw/arm/musicpal.c's lcd_refresh() with the unhelpful error
message "no position information". Convert that one manually.
Fails to convert hw/arm/armsse.c, because Coccinelle gets confused by
ARMSSE being used both as typedef and function-like macro there.
Convert manually.
Fails to convert hw/rx/rx-gdbsim.c, because Coccinelle gets confused
by RXCPU being used both as typedef and function-like macro there.
Convert manually. The other files using RXCPU that way don't need
conversion.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20200707160613.848843-27-armbru@redhat.com>
[Straightforwad conflict with commit 2336172d9b "audio: set default
value for pcspk.iobase property" resolved]
2020-07-07 19:05:54 +03:00
|
|
|
object_property_set_int(OBJECT(pnv_core), "nr-threads",
|
|
|
|
chip->nr_threads, &error_fatal);
|
|
|
|
object_property_set_int(OBJECT(pnv_core), CPU_CORE_PROP_CORE_ID,
|
|
|
|
core_hwid, &error_fatal);
|
|
|
|
object_property_set_int(OBJECT(pnv_core), "pir",
|
|
|
|
pcc->core_pir(chip, core_hwid), &error_fatal);
|
|
|
|
object_property_set_int(OBJECT(pnv_core), "hrmor", pnv->fw_load_addr,
|
|
|
|
&error_fatal);
|
|
|
|
object_property_set_link(OBJECT(pnv_core), "chip", OBJECT(chip),
|
2019-11-15 18:56:00 +03:00
|
|
|
&error_abort);
|
qdev: Convert bus-less devices to qdev_realize() with Coccinelle
All remaining conversions to qdev_realize() are for bus-less devices.
Coccinelle script:
// only correct for bus-less @dev!
@@
expression errp;
expression dev;
@@
- qdev_init_nofail(dev);
+ qdev_realize(dev, NULL, &error_fatal);
@ depends on !(file in "hw/core/qdev.c") && !(file in "hw/core/bus.c")@
expression errp;
expression dev;
symbol true;
@@
- object_property_set_bool(OBJECT(dev), true, "realized", errp);
+ qdev_realize(DEVICE(dev), NULL, errp);
@ depends on !(file in "hw/core/qdev.c") && !(file in "hw/core/bus.c")@
expression errp;
expression dev;
symbol true;
@@
- object_property_set_bool(dev, true, "realized", errp);
+ qdev_realize(DEVICE(dev), NULL, errp);
Note that Coccinelle chokes on ARMSSE typedef vs. macro in
hw/arm/armsse.c. Worked around by temporarily renaming the macro for
the spatch run.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200610053247.1583243-57-armbru@redhat.com>
2020-06-10 08:32:45 +03:00
|
|
|
qdev_realize(DEVICE(pnv_core), NULL, &error_fatal);
|
2016-10-22 12:46:41 +03:00
|
|
|
|
|
|
|
/* Each core has an XSCOM MMIO region */
|
2019-12-13 15:00:13 +03:00
|
|
|
xscom_core_base = pcc->xscom_core_base(chip, core_hwid);
|
2018-01-15 21:04:04 +03:00
|
|
|
|
|
|
|
pnv_xscom_add_subregion(chip, xscom_core_base,
|
2019-11-25 09:58:03 +03:00
|
|
|
&pnv_core->xscom_regs);
|
2016-10-22 12:46:39 +03:00
|
|
|
i++;
|
|
|
|
}
|
2018-06-14 17:00:41 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_chip_realize(DeviceState *dev, Error **errp)
|
|
|
|
{
|
|
|
|
PnvChip *chip = PNV_CHIP(dev);
|
|
|
|
Error *error = NULL;
|
|
|
|
|
|
|
|
/* Cores */
|
|
|
|
pnv_chip_core_realize(chip, &error);
|
|
|
|
if (error) {
|
|
|
|
error_propagate(errp, error);
|
|
|
|
return;
|
|
|
|
}
|
2016-10-22 12:46:36 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
static Property pnv_chip_properties[] = {
|
|
|
|
DEFINE_PROP_UINT32("chip-id", PnvChip, chip_id, 0),
|
|
|
|
DEFINE_PROP_UINT64("ram-start", PnvChip, ram_start, 0),
|
|
|
|
DEFINE_PROP_UINT64("ram-size", PnvChip, ram_size, 0),
|
2016-10-22 12:46:37 +03:00
|
|
|
DEFINE_PROP_UINT32("nr-cores", PnvChip, nr_cores, 1),
|
|
|
|
DEFINE_PROP_UINT64("cores-mask", PnvChip, cores_mask, 0x0),
|
2020-01-06 17:56:39 +03:00
|
|
|
DEFINE_PROP_UINT32("nr-threads", PnvChip, nr_threads, 1),
|
2016-10-22 12:46:36 +03:00
|
|
|
DEFINE_PROP_END_OF_LIST(),
|
|
|
|
};
|
|
|
|
|
|
|
|
static void pnv_chip_class_init(ObjectClass *klass, void *data)
|
|
|
|
{
|
|
|
|
DeviceClass *dc = DEVICE_CLASS(klass);
|
|
|
|
|
2017-03-09 21:37:53 +03:00
|
|
|
set_bit(DEVICE_CATEGORY_CPU, dc->categories);
|
2016-10-22 12:46:36 +03:00
|
|
|
dc->realize = pnv_chip_realize;
|
2020-01-10 18:30:32 +03:00
|
|
|
device_class_set_props(dc, pnv_chip_properties);
|
2016-10-22 12:46:36 +03:00
|
|
|
dc->desc = "PowerNV Chip";
|
|
|
|
}
|
|
|
|
|
2019-11-25 09:58:07 +03:00
|
|
|
PowerPCCPU *pnv_chip_find_cpu(PnvChip *chip, uint32_t pir)
|
|
|
|
{
|
|
|
|
int i, j;
|
|
|
|
|
|
|
|
for (i = 0; i < chip->nr_cores; i++) {
|
|
|
|
PnvCore *pc = chip->cores[i];
|
|
|
|
CPUCore *cc = CPU_CORE(pc);
|
|
|
|
|
|
|
|
for (j = 0; j < cc->nr_threads; j++) {
|
|
|
|
if (ppc_cpu_pir(pc->threads[j]) == pir) {
|
|
|
|
return pc->threads[j];
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
2017-04-05 15:41:26 +03:00
|
|
|
static ICSState *pnv_ics_get(XICSFabric *xi, int irq)
|
|
|
|
{
|
2017-12-15 16:56:01 +03:00
|
|
|
PnvMachineState *pnv = PNV_MACHINE(xi);
|
2022-06-21 20:34:32 +03:00
|
|
|
int i, j;
|
2017-04-05 15:41:26 +03:00
|
|
|
|
|
|
|
for (i = 0; i < pnv->num_chips; i++) {
|
2018-06-18 20:05:39 +03:00
|
|
|
Pnv8Chip *chip8 = PNV8_CHIP(pnv->chips[i]);
|
|
|
|
|
|
|
|
if (ics_valid_irq(&chip8->psi.ics, irq)) {
|
|
|
|
return &chip8->psi.ics;
|
2017-04-05 15:41:26 +03:00
|
|
|
}
|
2021-12-17 19:57:19 +03:00
|
|
|
|
2022-06-21 20:34:32 +03:00
|
|
|
for (j = 0; j < chip8->num_phbs; j++) {
|
2022-08-11 19:39:44 +03:00
|
|
|
PnvPHB *phb = chip8->phbs[j];
|
2022-06-24 11:49:12 +03:00
|
|
|
PnvPHB3 *phb3 = PNV_PHB3(phb->backend);
|
2022-06-21 20:34:32 +03:00
|
|
|
|
|
|
|
if (ics_valid_irq(&phb3->lsis, irq)) {
|
|
|
|
return &phb3->lsis;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (ics_valid_irq(ICS(&phb3->msis), irq)) {
|
|
|
|
return ICS(&phb3->msis);
|
|
|
|
}
|
2020-01-27 17:45:06 +03:00
|
|
|
}
|
2017-04-05 15:41:26 +03:00
|
|
|
}
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
2022-01-12 13:28:27 +03:00
|
|
|
PnvChip *pnv_get_chip(PnvMachineState *pnv, uint32_t chip_id)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = 0; i < pnv->num_chips; i++) {
|
|
|
|
PnvChip *chip = pnv->chips[i];
|
|
|
|
if (chip->chip_id == chip_id) {
|
|
|
|
return chip;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
2017-04-05 15:41:26 +03:00
|
|
|
static void pnv_ics_resend(XICSFabric *xi)
|
|
|
|
{
|
2017-12-15 16:56:01 +03:00
|
|
|
PnvMachineState *pnv = PNV_MACHINE(xi);
|
2022-06-21 20:34:33 +03:00
|
|
|
int i, j;
|
2017-04-05 15:41:26 +03:00
|
|
|
|
|
|
|
for (i = 0; i < pnv->num_chips; i++) {
|
2018-06-18 20:05:39 +03:00
|
|
|
Pnv8Chip *chip8 = PNV8_CHIP(pnv->chips[i]);
|
2020-01-27 17:45:06 +03:00
|
|
|
|
2018-06-18 20:05:39 +03:00
|
|
|
ics_resend(&chip8->psi.ics);
|
2022-06-21 20:34:33 +03:00
|
|
|
|
|
|
|
for (j = 0; j < chip8->num_phbs; j++) {
|
2022-08-11 19:39:44 +03:00
|
|
|
PnvPHB *phb = chip8->phbs[j];
|
2022-06-24 11:49:12 +03:00
|
|
|
PnvPHB3 *phb3 = PNV_PHB3(phb->backend);
|
2022-06-21 20:34:33 +03:00
|
|
|
|
|
|
|
ics_resend(&phb3->lsis);
|
|
|
|
ics_resend(ICS(&phb3->msis));
|
|
|
|
}
|
2017-04-05 15:41:26 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2017-04-03 10:46:01 +03:00
|
|
|
static ICPState *pnv_icp_get(XICSFabric *xi, int pir)
|
|
|
|
{
|
|
|
|
PowerPCCPU *cpu = ppc_get_vcpu_by_pir(pir);
|
|
|
|
|
2019-03-06 11:50:10 +03:00
|
|
|
return cpu ? ICP(pnv_cpu_state(cpu)->intc) : NULL;
|
2017-04-03 10:46:01 +03:00
|
|
|
}
|
|
|
|
|
2017-04-03 10:46:02 +03:00
|
|
|
static void pnv_pic_print_info(InterruptStatsProvider *obj,
|
|
|
|
Monitor *mon)
|
|
|
|
{
|
2017-12-15 16:56:01 +03:00
|
|
|
PnvMachineState *pnv = PNV_MACHINE(obj);
|
2017-04-05 15:41:26 +03:00
|
|
|
int i;
|
2017-04-03 10:46:02 +03:00
|
|
|
CPUState *cs;
|
|
|
|
|
|
|
|
CPU_FOREACH(cs) {
|
|
|
|
PowerPCCPU *cpu = POWERPC_CPU(cs);
|
|
|
|
|
2019-12-13 15:00:07 +03:00
|
|
|
/* XXX: loop on each chip/core/thread instead of CPU_FOREACH() */
|
|
|
|
PNV_CHIP_GET_CLASS(pnv->chips[0])->intc_print_info(pnv->chips[0], cpu,
|
|
|
|
mon);
|
2017-04-03 10:46:02 +03:00
|
|
|
}
|
2017-04-05 15:41:26 +03:00
|
|
|
|
|
|
|
for (i = 0; i < pnv->num_chips; i++) {
|
2019-03-06 11:50:13 +03:00
|
|
|
PNV_CHIP_GET_CLASS(pnv->chips[i])->pic_print_info(pnv->chips[i], mon);
|
2017-04-05 15:41:26 +03:00
|
|
|
}
|
2017-04-03 10:46:02 +03:00
|
|
|
}
|
|
|
|
|
2019-11-25 09:58:09 +03:00
|
|
|
static int pnv_match_nvt(XiveFabric *xfb, uint8_t format,
|
|
|
|
uint8_t nvt_blk, uint32_t nvt_idx,
|
|
|
|
bool cam_ignore, uint8_t priority,
|
|
|
|
uint32_t logic_serv,
|
|
|
|
XiveTCTXMatch *match)
|
|
|
|
{
|
|
|
|
PnvMachineState *pnv = PNV_MACHINE(xfb);
|
|
|
|
int total_count = 0;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = 0; i < pnv->num_chips; i++) {
|
|
|
|
Pnv9Chip *chip9 = PNV9_CHIP(pnv->chips[i]);
|
|
|
|
XivePresenter *xptr = XIVE_PRESENTER(&chip9->xive);
|
|
|
|
XivePresenterClass *xpc = XIVE_PRESENTER_GET_CLASS(xptr);
|
|
|
|
int count;
|
|
|
|
|
|
|
|
count = xpc->match_nvt(xptr, format, nvt_blk, nvt_idx, cam_ignore,
|
|
|
|
priority, logic_serv, match);
|
|
|
|
|
|
|
|
if (count < 0) {
|
|
|
|
return count;
|
|
|
|
}
|
|
|
|
|
|
|
|
total_count += count;
|
|
|
|
}
|
|
|
|
|
|
|
|
return total_count;
|
|
|
|
}
|
|
|
|
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 08:51:38 +03:00
|
|
|
static int pnv10_xive_match_nvt(XiveFabric *xfb, uint8_t format,
|
|
|
|
uint8_t nvt_blk, uint32_t nvt_idx,
|
|
|
|
bool cam_ignore, uint8_t priority,
|
|
|
|
uint32_t logic_serv,
|
|
|
|
XiveTCTXMatch *match)
|
|
|
|
{
|
|
|
|
PnvMachineState *pnv = PNV_MACHINE(xfb);
|
|
|
|
int total_count = 0;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = 0; i < pnv->num_chips; i++) {
|
|
|
|
Pnv10Chip *chip10 = PNV10_CHIP(pnv->chips[i]);
|
|
|
|
XivePresenter *xptr = XIVE_PRESENTER(&chip10->xive);
|
|
|
|
XivePresenterClass *xpc = XIVE_PRESENTER_GET_CLASS(xptr);
|
|
|
|
int count;
|
|
|
|
|
|
|
|
count = xpc->match_nvt(xptr, format, nvt_blk, nvt_idx, cam_ignore,
|
|
|
|
priority, logic_serv, match);
|
|
|
|
|
|
|
|
if (count < 0) {
|
|
|
|
return count;
|
|
|
|
}
|
|
|
|
|
|
|
|
total_count += count;
|
|
|
|
}
|
|
|
|
|
|
|
|
return total_count;
|
|
|
|
}
|
|
|
|
|
2019-07-31 17:12:16 +03:00
|
|
|
static void pnv_machine_power8_class_init(ObjectClass *oc, void *data)
|
2016-10-22 12:46:35 +03:00
|
|
|
{
|
|
|
|
MachineClass *mc = MACHINE_CLASS(oc);
|
2017-04-03 10:46:01 +03:00
|
|
|
XICSFabricClass *xic = XICS_FABRIC_CLASS(oc);
|
2019-12-13 14:59:50 +03:00
|
|
|
PnvMachineClass *pmc = PNV_MACHINE_CLASS(oc);
|
|
|
|
static const char compat[] = "qemu,powernv8\0qemu,powernv\0ibm,powernv";
|
2019-07-31 17:12:16 +03:00
|
|
|
|
2022-06-24 11:49:12 +03:00
|
|
|
static GlobalProperty phb_compat[] = {
|
|
|
|
{ TYPE_PNV_PHB, "version", "3" },
|
2022-06-24 11:49:16 +03:00
|
|
|
{ TYPE_PNV_PHB_ROOT_PORT, "version", "3" },
|
2022-06-24 11:49:12 +03:00
|
|
|
};
|
|
|
|
|
2019-07-31 17:12:16 +03:00
|
|
|
mc->desc = "IBM PowerNV (Non-Virtualized) POWER8";
|
|
|
|
mc->default_cpu_type = POWERPC_CPU_TYPE_NAME("power8_v2.0");
|
2022-06-24 11:49:12 +03:00
|
|
|
compat_props_add(mc->compat_props, phb_compat, G_N_ELEMENTS(phb_compat));
|
2019-07-31 17:12:16 +03:00
|
|
|
|
|
|
|
xic->icp_get = pnv_icp_get;
|
|
|
|
xic->ics_get = pnv_ics_get;
|
|
|
|
xic->ics_resend = pnv_ics_resend;
|
2019-12-13 14:59:50 +03:00
|
|
|
|
|
|
|
pmc->compat = compat;
|
|
|
|
pmc->compat_size = sizeof(compat);
|
2022-08-11 19:39:45 +03:00
|
|
|
|
|
|
|
machine_class_allow_dynamic_sysbus_dev(mc, TYPE_PNV_PHB);
|
2019-07-31 17:12:16 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_machine_power9_class_init(ObjectClass *oc, void *data)
|
|
|
|
{
|
|
|
|
MachineClass *mc = MACHINE_CLASS(oc);
|
2019-11-25 09:58:09 +03:00
|
|
|
XiveFabricClass *xfc = XIVE_FABRIC_CLASS(oc);
|
2019-12-13 14:59:50 +03:00
|
|
|
PnvMachineClass *pmc = PNV_MACHINE_CLASS(oc);
|
|
|
|
static const char compat[] = "qemu,powernv9\0ibm,powernv";
|
2019-07-31 17:12:16 +03:00
|
|
|
|
2022-06-24 11:49:14 +03:00
|
|
|
static GlobalProperty phb_compat[] = {
|
|
|
|
{ TYPE_PNV_PHB, "version", "4" },
|
2022-06-24 11:49:17 +03:00
|
|
|
{ TYPE_PNV_PHB_ROOT_PORT, "version", "4" },
|
2022-06-24 11:49:14 +03:00
|
|
|
};
|
|
|
|
|
2019-07-31 17:12:16 +03:00
|
|
|
mc->desc = "IBM PowerNV (Non-Virtualized) POWER9";
|
|
|
|
mc->default_cpu_type = POWERPC_CPU_TYPE_NAME("power9_v2.0");
|
2022-06-24 11:49:14 +03:00
|
|
|
compat_props_add(mc->compat_props, phb_compat, G_N_ELEMENTS(phb_compat));
|
|
|
|
|
2019-11-25 09:58:09 +03:00
|
|
|
xfc->match_nvt = pnv_match_nvt;
|
2019-07-31 17:12:16 +03:00
|
|
|
|
|
|
|
mc->alias = "powernv";
|
2019-12-13 14:59:50 +03:00
|
|
|
|
|
|
|
pmc->compat = compat;
|
|
|
|
pmc->compat_size = sizeof(compat);
|
2019-12-13 14:59:56 +03:00
|
|
|
pmc->dt_power_mgt = pnv_dt_power_mgt;
|
2022-08-11 19:39:47 +03:00
|
|
|
|
|
|
|
machine_class_allow_dynamic_sysbus_dev(mc, TYPE_PNV_PHB);
|
2019-07-31 17:12:16 +03:00
|
|
|
}
|
|
|
|
|
2019-12-05 21:44:51 +03:00
|
|
|
static void pnv_machine_power10_class_init(ObjectClass *oc, void *data)
|
|
|
|
{
|
|
|
|
MachineClass *mc = MACHINE_CLASS(oc);
|
2019-12-13 14:59:50 +03:00
|
|
|
PnvMachineClass *pmc = PNV_MACHINE_CLASS(oc);
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 08:51:38 +03:00
|
|
|
XiveFabricClass *xfc = XIVE_FABRIC_CLASS(oc);
|
2019-12-13 14:59:50 +03:00
|
|
|
static const char compat[] = "qemu,powernv10\0ibm,powernv";
|
2019-12-05 21:44:51 +03:00
|
|
|
|
2022-06-24 11:49:14 +03:00
|
|
|
static GlobalProperty phb_compat[] = {
|
|
|
|
{ TYPE_PNV_PHB, "version", "5" },
|
2022-06-24 11:49:17 +03:00
|
|
|
{ TYPE_PNV_PHB_ROOT_PORT, "version", "5" },
|
2022-06-24 11:49:14 +03:00
|
|
|
};
|
|
|
|
|
2019-12-05 21:44:51 +03:00
|
|
|
mc->desc = "IBM PowerNV (Non-Virtualized) POWER10";
|
2021-08-09 16:45:23 +03:00
|
|
|
mc->default_cpu_type = POWERPC_CPU_TYPE_NAME("power10_v2.0");
|
2022-06-24 11:49:14 +03:00
|
|
|
compat_props_add(mc->compat_props, phb_compat, G_N_ELEMENTS(phb_compat));
|
2019-12-13 14:59:50 +03:00
|
|
|
|
|
|
|
pmc->compat = compat;
|
|
|
|
pmc->compat_size = sizeof(compat);
|
2019-12-13 14:59:56 +03:00
|
|
|
pmc->dt_power_mgt = pnv_dt_power_mgt;
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 08:51:38 +03:00
|
|
|
|
|
|
|
xfc->match_nvt = pnv10_xive_match_nvt;
|
2022-08-11 19:39:49 +03:00
|
|
|
|
|
|
|
machine_class_allow_dynamic_sysbus_dev(mc, TYPE_PNV_PHB);
|
2019-12-05 21:44:51 +03:00
|
|
|
}
|
|
|
|
|
2020-01-27 17:41:54 +03:00
|
|
|
static bool pnv_machine_get_hb(Object *obj, Error **errp)
|
|
|
|
{
|
|
|
|
PnvMachineState *pnv = PNV_MACHINE(obj);
|
|
|
|
|
|
|
|
return !!pnv->fw_load_addr;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_machine_set_hb(Object *obj, bool value, Error **errp)
|
|
|
|
{
|
|
|
|
PnvMachineState *pnv = PNV_MACHINE(obj);
|
|
|
|
|
|
|
|
if (value) {
|
|
|
|
pnv->fw_load_addr = 0x8000000;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-03-25 17:41:44 +03:00
|
|
|
static void pnv_cpu_do_nmi_on_cpu(CPUState *cs, run_on_cpu_data arg)
|
|
|
|
{
|
|
|
|
PowerPCCPU *cpu = POWERPC_CPU(cs);
|
|
|
|
CPUPPCState *env = &cpu->env;
|
|
|
|
|
|
|
|
cpu_synchronize_state(cs);
|
|
|
|
ppc_cpu_do_system_reset(cs);
|
2020-05-11 23:02:02 +03:00
|
|
|
if (env->spr[SPR_SRR1] & SRR1_WAKESTATE) {
|
2020-05-07 14:48:24 +03:00
|
|
|
/*
|
|
|
|
* Power-save wakeups, as indicated by non-zero SRR1[46:47] put the
|
|
|
|
* wakeup reason in SRR1[42:45], system reset is indicated with 0b0100
|
|
|
|
* (PPC_BIT(43)).
|
|
|
|
*/
|
2020-05-11 23:02:02 +03:00
|
|
|
if (!(env->spr[SPR_SRR1] & SRR1_WAKERESET)) {
|
2020-05-07 14:48:24 +03:00
|
|
|
warn_report("ppc_cpu_do_system_reset does not set system reset wakeup reason");
|
2020-05-11 23:02:02 +03:00
|
|
|
env->spr[SPR_SRR1] |= SRR1_WAKERESET;
|
2020-05-07 14:48:24 +03:00
|
|
|
}
|
|
|
|
} else {
|
|
|
|
/*
|
|
|
|
* For non-powersave system resets, SRR1[42:45] are defined to be
|
|
|
|
* implementation-dependent. The POWER9 User Manual specifies that
|
|
|
|
* an external (SCOM driven, which may come from a BMC nmi command or
|
|
|
|
* another CPU requesting a NMI IPI) system reset exception should be
|
|
|
|
* 0b0010 (PPC_BIT(44)).
|
|
|
|
*/
|
2020-05-11 23:02:02 +03:00
|
|
|
env->spr[SPR_SRR1] |= SRR1_WAKESCOM;
|
2020-05-07 14:48:24 +03:00
|
|
|
}
|
2020-03-25 17:41:44 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_nmi(NMIState *n, int cpu_index, Error **errp)
|
|
|
|
{
|
|
|
|
CPUState *cs;
|
|
|
|
|
|
|
|
CPU_FOREACH(cs) {
|
|
|
|
async_run_on_cpu(cs, pnv_cpu_do_nmi_on_cpu, RUN_ON_CPU_NULL);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2019-07-31 17:12:16 +03:00
|
|
|
static void pnv_machine_class_init(ObjectClass *oc, void *data)
|
|
|
|
{
|
|
|
|
MachineClass *mc = MACHINE_CLASS(oc);
|
2017-04-03 10:46:02 +03:00
|
|
|
InterruptStatsProviderClass *ispc = INTERRUPT_STATS_PROVIDER_CLASS(oc);
|
2020-03-25 17:41:44 +03:00
|
|
|
NMIClass *nc = NMI_CLASS(oc);
|
2016-10-22 12:46:35 +03:00
|
|
|
|
|
|
|
mc->desc = "IBM PowerNV (Non-Virtualized)";
|
2017-12-15 16:56:01 +03:00
|
|
|
mc->init = pnv_init;
|
|
|
|
mc->reset = pnv_reset;
|
2016-10-22 12:46:35 +03:00
|
|
|
mc->max_cpus = MAX_CPUS;
|
2019-09-11 17:29:25 +03:00
|
|
|
/* Pnv provides a AHCI device for storage */
|
|
|
|
mc->block_default_type = IF_IDE;
|
2016-10-22 12:46:35 +03:00
|
|
|
mc->no_parallel = 1;
|
|
|
|
mc->default_boot_order = NULL;
|
2019-08-21 06:09:45 +03:00
|
|
|
/*
|
|
|
|
* RAM defaults to less than 2048 for 32-bit hosts, and large
|
|
|
|
* enough to fit the maximum initrd size at it's load address
|
|
|
|
*/
|
ppc/pnv: Set default RAM size to 1 GB
The memory layout of the PowerNV machine is defined as :
#define KERNEL_LOAD_BASE ((void *)0x20000000)
#define KERNEL_LOAD_SIZE 0x08000000
#define INITRAMFS_LOAD_BASE KERNEL_LOAD_BASE + KERNEL_LOAD_SIZE
#define INITRAMFS_LOAD_SIZE 0x08000000
#define SKIBOOT_BASE 0x30000000
#define SKIBOOT_SIZE 0x01c10000
#define CPU_STACKS_BASE (SKIBOOT_BASE + SKIBOOT_SIZE)
#define STACK_SHIFT 15
#define STACK_SIZE (1 << STACK_SHIFT)
The overall size of the CPU stacks is (max PIR + 1) * 32K and the
machine easily reaches 800MB of minimum required RAM.
Any value below will result in a skiboot crash :
[ 0.034949905,3] MEM: Partial overlap detected between regions:
[ 0.034959039,3] MEM: ibm,firmware-stacks [0x31c10000-0x3a450000] (new)
[ 0.034968576,3] MEM: ibm,firmware-allocs-memory@0 [0x31c10000-0x38400000]
[ 0.034980367,3] Out of memory adding skiboot reserved areas
[ 0.035074945,3] ***********************************************
[ 0.035093627,3] < assert failed at core/mem_region.c:1129 >
[ 0.035104247,3] .
[ 0.035108025,3] .
[ 0.035111651,3] .
[ 0.035115231,3] OO__)
[ 0.035119198,3] <"__/
[ 0.035122980,3] ^ ^
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20210129111719.790692-1-clg@kaod.org>
Reviewed-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2021-01-29 14:17:19 +03:00
|
|
|
mc->default_ram_size = 1 * GiB;
|
2020-02-19 19:09:36 +03:00
|
|
|
mc->default_ram_id = "pnv.ram";
|
2017-04-03 10:46:02 +03:00
|
|
|
ispc->print_info = pnv_pic_print_info;
|
2020-03-25 17:41:44 +03:00
|
|
|
nc->nmi_monitor_handler = pnv_nmi;
|
2020-01-27 17:41:54 +03:00
|
|
|
|
|
|
|
object_class_property_add_bool(oc, "hb-mode",
|
qom: Drop parameter @errp of object_property_add() & friends
The only way object_property_add() can fail is when a property with
the same name already exists. Since our property names are all
hardcoded, failure is a programming error, and the appropriate way to
handle it is passing &error_abort.
Same for its variants, except for object_property_add_child(), which
additionally fails when the child already has a parent. Parentage is
also under program control, so this is a programming error, too.
We have a bit over 500 callers. Almost half of them pass
&error_abort, slightly fewer ignore errors, one test case handles
errors, and the remaining few callers pass them to their own callers.
The previous few commits demonstrated once again that ignoring
programming errors is a bad idea.
Of the few ones that pass on errors, several violate the Error API.
The Error ** argument must be NULL, &error_abort, &error_fatal, or a
pointer to a variable containing NULL. Passing an argument of the
latter kind twice without clearing it in between is wrong: if the
first call sets an error, it no longer points to NULL for the second
call. ich9_pm_add_properties(), sparc32_ledma_realize(),
sparc32_dma_realize(), xilinx_axidma_realize(), xilinx_enet_realize()
are wrong that way.
When the one appropriate choice of argument is &error_abort, letting
users pick the argument is a bad idea.
Drop parameter @errp and assert the preconditions instead.
There's one exception to "duplicate property name is a programming
error": the way object_property_add() implements the magic (and
undocumented) "automatic arrayification". Don't drop @errp there.
Instead, rename object_property_add() to object_property_try_add(),
and add the obvious wrapper object_property_add().
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200505152926.18877-15-armbru@redhat.com>
[Two semantic rebase conflicts resolved]
2020-05-05 18:29:22 +03:00
|
|
|
pnv_machine_get_hb, pnv_machine_set_hb);
|
2020-01-27 17:41:54 +03:00
|
|
|
object_class_property_set_description(oc, "hb-mode",
|
2020-05-05 18:29:15 +03:00
|
|
|
"Use a hostboot like boot loader");
|
2016-10-22 12:46:35 +03:00
|
|
|
}
|
|
|
|
|
2018-06-18 20:05:39 +03:00
|
|
|
#define DEFINE_PNV8_CHIP_TYPE(type, class_initfn) \
|
|
|
|
{ \
|
|
|
|
.name = type, \
|
|
|
|
.class_init = class_initfn, \
|
|
|
|
.parent = TYPE_PNV8_CHIP, \
|
|
|
|
}
|
|
|
|
|
|
|
|
#define DEFINE_PNV9_CHIP_TYPE(type, class_initfn) \
|
|
|
|
{ \
|
|
|
|
.name = type, \
|
|
|
|
.class_init = class_initfn, \
|
|
|
|
.parent = TYPE_PNV9_CHIP, \
|
2017-10-09 22:51:11 +03:00
|
|
|
}
|
|
|
|
|
2019-12-05 21:44:51 +03:00
|
|
|
#define DEFINE_PNV10_CHIP_TYPE(type, class_initfn) \
|
|
|
|
{ \
|
|
|
|
.name = type, \
|
|
|
|
.class_init = class_initfn, \
|
|
|
|
.parent = TYPE_PNV10_CHIP, \
|
|
|
|
}
|
|
|
|
|
2017-10-09 22:51:11 +03:00
|
|
|
static const TypeInfo types[] = {
|
2019-12-05 21:44:51 +03:00
|
|
|
{
|
|
|
|
.name = MACHINE_TYPE_NAME("powernv10"),
|
|
|
|
.parent = TYPE_PNV_MACHINE,
|
|
|
|
.class_init = pnv_machine_power10_class_init,
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 08:51:38 +03:00
|
|
|
.interfaces = (InterfaceInfo[]) {
|
|
|
|
{ TYPE_XIVE_FABRIC },
|
|
|
|
{ },
|
|
|
|
},
|
2019-12-05 21:44:51 +03:00
|
|
|
},
|
2019-10-03 17:36:17 +03:00
|
|
|
{
|
|
|
|
.name = MACHINE_TYPE_NAME("powernv9"),
|
|
|
|
.parent = TYPE_PNV_MACHINE,
|
|
|
|
.class_init = pnv_machine_power9_class_init,
|
2019-11-25 09:58:09 +03:00
|
|
|
.interfaces = (InterfaceInfo[]) {
|
|
|
|
{ TYPE_XIVE_FABRIC },
|
|
|
|
{ },
|
|
|
|
},
|
2019-10-03 17:36:17 +03:00
|
|
|
},
|
|
|
|
{
|
|
|
|
.name = MACHINE_TYPE_NAME("powernv8"),
|
|
|
|
.parent = TYPE_PNV_MACHINE,
|
|
|
|
.class_init = pnv_machine_power8_class_init,
|
|
|
|
.interfaces = (InterfaceInfo[]) {
|
|
|
|
{ TYPE_XICS_FABRIC },
|
|
|
|
{ },
|
|
|
|
},
|
|
|
|
},
|
2017-10-09 22:51:11 +03:00
|
|
|
{
|
2017-12-15 16:56:01 +03:00
|
|
|
.name = TYPE_PNV_MACHINE,
|
2017-10-09 22:51:11 +03:00
|
|
|
.parent = TYPE_MACHINE,
|
2019-07-31 17:12:16 +03:00
|
|
|
.abstract = true,
|
2017-10-09 22:51:11 +03:00
|
|
|
.instance_size = sizeof(PnvMachineState),
|
2017-12-15 16:56:01 +03:00
|
|
|
.class_init = pnv_machine_class_init,
|
2019-12-13 14:59:50 +03:00
|
|
|
.class_size = sizeof(PnvMachineClass),
|
2017-10-09 22:51:11 +03:00
|
|
|
.interfaces = (InterfaceInfo[]) {
|
|
|
|
{ TYPE_INTERRUPT_STATS_PROVIDER },
|
2020-03-25 17:41:44 +03:00
|
|
|
{ TYPE_NMI },
|
2017-10-09 22:51:11 +03:00
|
|
|
{ },
|
|
|
|
},
|
2017-04-03 10:46:01 +03:00
|
|
|
},
|
2017-10-09 22:51:11 +03:00
|
|
|
{
|
|
|
|
.name = TYPE_PNV_CHIP,
|
|
|
|
.parent = TYPE_SYS_BUS_DEVICE,
|
|
|
|
.class_init = pnv_chip_class_init,
|
|
|
|
.instance_size = sizeof(PnvChip),
|
|
|
|
.class_size = sizeof(PnvChipClass),
|
|
|
|
.abstract = true,
|
|
|
|
},
|
2018-06-18 20:05:39 +03:00
|
|
|
|
2019-12-05 21:44:51 +03:00
|
|
|
/*
|
|
|
|
* P10 chip and variants
|
|
|
|
*/
|
|
|
|
{
|
|
|
|
.name = TYPE_PNV10_CHIP,
|
|
|
|
.parent = TYPE_PNV_CHIP,
|
|
|
|
.instance_init = pnv_chip_power10_instance_init,
|
|
|
|
.instance_size = sizeof(Pnv10Chip),
|
|
|
|
},
|
|
|
|
DEFINE_PNV10_CHIP_TYPE(TYPE_PNV_CHIP_POWER10, pnv_chip_power10_class_init),
|
|
|
|
|
2018-06-18 20:05:39 +03:00
|
|
|
/*
|
|
|
|
* P9 chip and variants
|
|
|
|
*/
|
|
|
|
{
|
|
|
|
.name = TYPE_PNV9_CHIP,
|
|
|
|
.parent = TYPE_PNV_CHIP,
|
|
|
|
.instance_init = pnv_chip_power9_instance_init,
|
|
|
|
.instance_size = sizeof(Pnv9Chip),
|
|
|
|
},
|
|
|
|
DEFINE_PNV9_CHIP_TYPE(TYPE_PNV_CHIP_POWER9, pnv_chip_power9_class_init),
|
|
|
|
|
|
|
|
/*
|
|
|
|
* P8 chip and variants
|
|
|
|
*/
|
|
|
|
{
|
|
|
|
.name = TYPE_PNV8_CHIP,
|
|
|
|
.parent = TYPE_PNV_CHIP,
|
|
|
|
.instance_init = pnv_chip_power8_instance_init,
|
|
|
|
.instance_size = sizeof(Pnv8Chip),
|
|
|
|
},
|
|
|
|
DEFINE_PNV8_CHIP_TYPE(TYPE_PNV_CHIP_POWER8, pnv_chip_power8_class_init),
|
|
|
|
DEFINE_PNV8_CHIP_TYPE(TYPE_PNV_CHIP_POWER8E, pnv_chip_power8e_class_init),
|
|
|
|
DEFINE_PNV8_CHIP_TYPE(TYPE_PNV_CHIP_POWER8NVL,
|
|
|
|
pnv_chip_power8nvl_class_init),
|
2016-10-22 12:46:35 +03:00
|
|
|
};
|
|
|
|
|
2017-10-09 22:51:11 +03:00
|
|
|
DEFINE_TYPES(types)
|