qemu/hw/ppc/spapr_iommu.c

660 lines
18 KiB
C
Raw Normal View History

/*
* QEMU sPAPR IOMMU (TCE) code
*
* Copyright (c) 2010 David Gibson, IBM Corporation <dwg@au1.ibm.com>
*
* 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
* version 2 of the License, or (at your option) any later version.
*
* 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"
spapr_iommu: Introduce "enabled" state for TCE table Currently TCE tables are created once at start and their sizes never change. We are going to change that by introducing a Dynamic DMA windows support where DMA configuration may change during the guest execution. This changes spapr_tce_new_table() to create an empty zero-size IOMMU memory region (IOMMU MR). Only LIOBN is assigned by the time of creation. It still will be called once at the owner object (VIO or PHB) creation. This introduces an "enabled" state for TCE table objects, some helper functions are added: - spapr_tce_table_enable() receives TCE table parameters, stores in sPAPRTCETable and allocates a guest view of the TCE table (in the user space or KVM) and sets the correct size on the IOMMU MR; - spapr_tce_table_disable() disposes the table and resets the IOMMU MR size; it is made public as the following DDW code will be using it. This changes the PHB reset handler to do the default DMA initialization instead of spapr_phb_realize(). This does not make differenct now but later with more than just one DMA window, we will have to remove them all and create the default one on a system reset. No visible change in behaviour is expected except the actual table will be reallocated every reset. We might optimize this later. The other way to implement this would be dynamically create/remove the TCE table QOM objects but this would make migration impossible as the migration code expects all QOM objects to exist at the receiver so we have to have TCE table objects created when migration begins. Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2016-06-01 11:57:33 +03:00
#include "qemu/error-report.h"
#include "hw/hw.h"
#include "qemu/log.h"
#include "sysemu/kvm.h"
#include "hw/qdev.h"
#include "kvm_ppc.h"
#include "sysemu/dma.h"
#include "exec/address-spaces.h"
#include "trace.h"
#include "hw/ppc/spapr.h"
#include "hw/ppc/spapr_vio.h"
#include <libfdt.h>
enum sPAPRTCEAccess {
SPAPR_TCE_FAULT = 0,
SPAPR_TCE_RO = 1,
SPAPR_TCE_WO = 2,
SPAPR_TCE_RW = 3,
};
#define IOMMU_PAGE_SIZE(shift) (1ULL << (shift))
#define IOMMU_PAGE_MASK(shift) (~(IOMMU_PAGE_SIZE(shift) - 1))
static QLIST_HEAD(spapr_tce_tables, sPAPRTCETable) spapr_tce_tables;
sPAPRTCETable *spapr_tce_find_by_liobn(target_ulong liobn)
{
sPAPRTCETable *tcet;
if (liobn & 0xFFFFFFFF00000000ULL) {
hcall_dprintf("Request for out-of-bounds LIOBN 0x" TARGET_FMT_lx "\n",
liobn);
return NULL;
}
QLIST_FOREACH(tcet, &spapr_tce_tables, list) {
if (tcet->liobn == (uint32_t)liobn) {
return tcet;
}
}
return NULL;
}
static IOMMUAccessFlags spapr_tce_iommu_access_flags(uint64_t tce)
{
switch (tce & SPAPR_TCE_RW) {
case SPAPR_TCE_FAULT:
return IOMMU_NONE;
case SPAPR_TCE_RO:
return IOMMU_RO;
case SPAPR_TCE_WO:
return IOMMU_WO;
default: /* SPAPR_TCE_RW */
return IOMMU_RW;
}
}
static uint64_t *spapr_tce_alloc_table(uint32_t liobn,
uint32_t page_shift,
uint64_t bus_offset,
uint32_t nb_table,
int *fd,
bool need_vfio)
{
uint64_t *table = NULL;
if (kvm_enabled()) {
table = kvmppc_create_spapr_tce(liobn, page_shift, bus_offset, nb_table,
fd, need_vfio);
}
if (!table) {
*fd = -1;
table = g_malloc0(nb_table * sizeof(uint64_t));
}
trace_spapr_iommu_new_table(liobn, table, *fd);
return table;
}
static void spapr_tce_free_table(uint64_t *table, int fd, uint32_t nb_table)
{
if (!kvm_enabled() ||
(kvmppc_remove_spapr_tce(table, fd, nb_table) != 0)) {
g_free(table);
}
}
/* Called from RCU critical section */
static IOMMUTLBEntry spapr_tce_translate_iommu(IOMMUMemoryRegion *iommu,
hwaddr addr,
IOMMUAccessFlags flag)
{
sPAPRTCETable *tcet = container_of(iommu, sPAPRTCETable, iommu);
uint64_t tce;
IOMMUTLBEntry ret = {
.target_as = &address_space_memory,
.iova = 0,
.translated_addr = 0,
.addr_mask = ~(hwaddr)0,
.perm = IOMMU_NONE,
};
if ((addr >> tcet->page_shift) < tcet->nb_table) {
/* Check if we are in bound */
hwaddr page_mask = IOMMU_PAGE_MASK(tcet->page_shift);
tce = tcet->table[addr >> tcet->page_shift];
ret.iova = addr & page_mask;
ret.translated_addr = tce & page_mask;
ret.addr_mask = ~page_mask;
ret.perm = spapr_tce_iommu_access_flags(tce);
}
trace_spapr_iommu_xlate(tcet->liobn, addr, ret.iova, ret.perm,
ret.addr_mask);
return ret;
}
static void spapr_tce_table_pre_save(void *opaque)
{
sPAPRTCETable *tcet = SPAPR_TCE_TABLE(opaque);
tcet->mig_table = tcet->table;
tcet->mig_nb_table = tcet->nb_table;
trace_spapr_iommu_pre_save(tcet->liobn, tcet->mig_nb_table,
tcet->bus_offset, tcet->page_shift);
}
static uint64_t spapr_tce_get_min_page_size(IOMMUMemoryRegion *iommu)
{
sPAPRTCETable *tcet = container_of(iommu, sPAPRTCETable, iommu);
return 1ULL << tcet->page_shift;
}
static void spapr_tce_notify_flag_changed(IOMMUMemoryRegion *iommu,
IOMMUNotifierFlag old,
IOMMUNotifierFlag new)
{
struct sPAPRTCETable *tbl = container_of(iommu, sPAPRTCETable, iommu);
if (old == IOMMU_NOTIFIER_NONE && new != IOMMU_NOTIFIER_NONE) {
spapr_tce_set_need_vfio(tbl, true);
} else if (old != IOMMU_NOTIFIER_NONE && new == IOMMU_NOTIFIER_NONE) {
spapr_tce_set_need_vfio(tbl, false);
}
}
static int spapr_tce_table_post_load(void *opaque, int version_id)
{
sPAPRTCETable *tcet = SPAPR_TCE_TABLE(opaque);
uint32_t old_nb_table = tcet->nb_table;
uint64_t old_bus_offset = tcet->bus_offset;
uint32_t old_page_shift = tcet->page_shift;
if (tcet->vdev) {
spapr_vio_set_bypass(tcet->vdev, tcet->bypass);
}
if (tcet->mig_nb_table != tcet->nb_table) {
spapr_tce_table_disable(tcet);
}
if (tcet->mig_nb_table) {
if (!tcet->nb_table) {
spapr_tce_table_enable(tcet, old_page_shift, old_bus_offset,
tcet->mig_nb_table);
}
memcpy(tcet->table, tcet->mig_table,
tcet->nb_table * sizeof(tcet->table[0]));
free(tcet->mig_table);
tcet->mig_table = NULL;
}
trace_spapr_iommu_post_load(tcet->liobn, old_nb_table, tcet->nb_table,
tcet->bus_offset, tcet->page_shift);
return 0;
}
static bool spapr_tce_table_ex_needed(void *opaque)
{
sPAPRTCETable *tcet = opaque;
return tcet->bus_offset || tcet->page_shift != 0xC;
}
static const VMStateDescription vmstate_spapr_tce_table_ex = {
.name = "spapr_iommu_ex",
.version_id = 1,
.minimum_version_id = 1,
.needed = spapr_tce_table_ex_needed,
.fields = (VMStateField[]) {
VMSTATE_UINT64(bus_offset, sPAPRTCETable),
VMSTATE_UINT32(page_shift, sPAPRTCETable),
VMSTATE_END_OF_LIST()
},
};
static const VMStateDescription vmstate_spapr_tce_table = {
.name = "spapr_iommu",
.version_id = 2,
.minimum_version_id = 2,
.pre_save = spapr_tce_table_pre_save,
.post_load = spapr_tce_table_post_load,
.fields = (VMStateField []) {
/* Sanity check */
VMSTATE_UINT32_EQUAL(liobn, sPAPRTCETable, NULL),
/* IOMMU state */
VMSTATE_UINT32(mig_nb_table, sPAPRTCETable),
VMSTATE_BOOL(bypass, sPAPRTCETable),
VMSTATE_VARRAY_UINT32_ALLOC(mig_table, sPAPRTCETable, mig_nb_table, 0,
vmstate_info_uint64, uint64_t),
VMSTATE_END_OF_LIST()
},
.subsections = (const VMStateDescription*[]) {
&vmstate_spapr_tce_table_ex,
NULL
}
};
static int spapr_tce_table_realize(DeviceState *dev)
{
sPAPRTCETable *tcet = SPAPR_TCE_TABLE(dev);
Object *tcetobj = OBJECT(tcet);
char tmp[32];
tcet->fd = -1;
spapr_iommu: Introduce "enabled" state for TCE table Currently TCE tables are created once at start and their sizes never change. We are going to change that by introducing a Dynamic DMA windows support where DMA configuration may change during the guest execution. This changes spapr_tce_new_table() to create an empty zero-size IOMMU memory region (IOMMU MR). Only LIOBN is assigned by the time of creation. It still will be called once at the owner object (VIO or PHB) creation. This introduces an "enabled" state for TCE table objects, some helper functions are added: - spapr_tce_table_enable() receives TCE table parameters, stores in sPAPRTCETable and allocates a guest view of the TCE table (in the user space or KVM) and sets the correct size on the IOMMU MR; - spapr_tce_table_disable() disposes the table and resets the IOMMU MR size; it is made public as the following DDW code will be using it. This changes the PHB reset handler to do the default DMA initialization instead of spapr_phb_realize(). This does not make differenct now but later with more than just one DMA window, we will have to remove them all and create the default one on a system reset. No visible change in behaviour is expected except the actual table will be reallocated every reset. We might optimize this later. The other way to implement this would be dynamically create/remove the TCE table QOM objects but this would make migration impossible as the migration code expects all QOM objects to exist at the receiver so we have to have TCE table objects created when migration begins. Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2016-06-01 11:57:33 +03:00
tcet->need_vfio = false;
snprintf(tmp, sizeof(tmp), "tce-root-%x", tcet->liobn);
memory_region_init(&tcet->root, tcetobj, tmp, UINT64_MAX);
snprintf(tmp, sizeof(tmp), "tce-iommu-%x", tcet->liobn);
memory_region_init_iommu(&tcet->iommu, sizeof(tcet->iommu),
TYPE_SPAPR_IOMMU_MEMORY_REGION,
tcetobj, tmp, 0);
QLIST_INSERT_HEAD(&spapr_tce_tables, tcet, list);
vmstate_register(DEVICE(tcet), tcet->liobn, &vmstate_spapr_tce_table,
tcet);
return 0;
}
void spapr_tce_set_need_vfio(sPAPRTCETable *tcet, bool need_vfio)
{
size_t table_size = tcet->nb_table * sizeof(uint64_t);
void *newtable;
if (need_vfio == tcet->need_vfio) {
/* Nothing to do */
return;
}
if (!need_vfio) {
/* FIXME: We don't support transition back to KVM accelerated
* TCEs yet */
return;
}
tcet->need_vfio = true;
if (tcet->fd < 0) {
/* Table is already in userspace, nothing to be do */
return;
}
newtable = g_malloc(table_size);
memcpy(newtable, tcet->table, table_size);
kvmppc_remove_spapr_tce(tcet->table, tcet->fd, tcet->nb_table);
tcet->fd = -1;
tcet->table = newtable;
}
spapr_iommu: Introduce "enabled" state for TCE table Currently TCE tables are created once at start and their sizes never change. We are going to change that by introducing a Dynamic DMA windows support where DMA configuration may change during the guest execution. This changes spapr_tce_new_table() to create an empty zero-size IOMMU memory region (IOMMU MR). Only LIOBN is assigned by the time of creation. It still will be called once at the owner object (VIO or PHB) creation. This introduces an "enabled" state for TCE table objects, some helper functions are added: - spapr_tce_table_enable() receives TCE table parameters, stores in sPAPRTCETable and allocates a guest view of the TCE table (in the user space or KVM) and sets the correct size on the IOMMU MR; - spapr_tce_table_disable() disposes the table and resets the IOMMU MR size; it is made public as the following DDW code will be using it. This changes the PHB reset handler to do the default DMA initialization instead of spapr_phb_realize(). This does not make differenct now but later with more than just one DMA window, we will have to remove them all and create the default one on a system reset. No visible change in behaviour is expected except the actual table will be reallocated every reset. We might optimize this later. The other way to implement this would be dynamically create/remove the TCE table QOM objects but this would make migration impossible as the migration code expects all QOM objects to exist at the receiver so we have to have TCE table objects created when migration begins. Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2016-06-01 11:57:33 +03:00
sPAPRTCETable *spapr_tce_new_table(DeviceState *owner, uint32_t liobn)
{
sPAPRTCETable *tcet;
spapr_iommu: Introduce "enabled" state for TCE table Currently TCE tables are created once at start and their sizes never change. We are going to change that by introducing a Dynamic DMA windows support where DMA configuration may change during the guest execution. This changes spapr_tce_new_table() to create an empty zero-size IOMMU memory region (IOMMU MR). Only LIOBN is assigned by the time of creation. It still will be called once at the owner object (VIO or PHB) creation. This introduces an "enabled" state for TCE table objects, some helper functions are added: - spapr_tce_table_enable() receives TCE table parameters, stores in sPAPRTCETable and allocates a guest view of the TCE table (in the user space or KVM) and sets the correct size on the IOMMU MR; - spapr_tce_table_disable() disposes the table and resets the IOMMU MR size; it is made public as the following DDW code will be using it. This changes the PHB reset handler to do the default DMA initialization instead of spapr_phb_realize(). This does not make differenct now but later with more than just one DMA window, we will have to remove them all and create the default one on a system reset. No visible change in behaviour is expected except the actual table will be reallocated every reset. We might optimize this later. The other way to implement this would be dynamically create/remove the TCE table QOM objects but this would make migration impossible as the migration code expects all QOM objects to exist at the receiver so we have to have TCE table objects created when migration begins. Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2016-06-01 11:57:33 +03:00
char tmp[32];
if (spapr_tce_find_by_liobn(liobn)) {
error_report("Attempted to create TCE table with duplicate"
" LIOBN 0x%x", liobn);
return NULL;
}
tcet = SPAPR_TCE_TABLE(object_new(TYPE_SPAPR_TCE_TABLE));
tcet->liobn = liobn;
snprintf(tmp, sizeof(tmp), "tce-table-%x", liobn);
object_property_add_child(OBJECT(owner), tmp, OBJECT(tcet), NULL);
object_property_set_bool(OBJECT(tcet), true, "realized", NULL);
return tcet;
}
spapr_iommu: Introduce "enabled" state for TCE table Currently TCE tables are created once at start and their sizes never change. We are going to change that by introducing a Dynamic DMA windows support where DMA configuration may change during the guest execution. This changes spapr_tce_new_table() to create an empty zero-size IOMMU memory region (IOMMU MR). Only LIOBN is assigned by the time of creation. It still will be called once at the owner object (VIO or PHB) creation. This introduces an "enabled" state for TCE table objects, some helper functions are added: - spapr_tce_table_enable() receives TCE table parameters, stores in sPAPRTCETable and allocates a guest view of the TCE table (in the user space or KVM) and sets the correct size on the IOMMU MR; - spapr_tce_table_disable() disposes the table and resets the IOMMU MR size; it is made public as the following DDW code will be using it. This changes the PHB reset handler to do the default DMA initialization instead of spapr_phb_realize(). This does not make differenct now but later with more than just one DMA window, we will have to remove them all and create the default one on a system reset. No visible change in behaviour is expected except the actual table will be reallocated every reset. We might optimize this later. The other way to implement this would be dynamically create/remove the TCE table QOM objects but this would make migration impossible as the migration code expects all QOM objects to exist at the receiver so we have to have TCE table objects created when migration begins. Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2016-06-01 11:57:33 +03:00
void spapr_tce_table_enable(sPAPRTCETable *tcet,
uint32_t page_shift, uint64_t bus_offset,
uint32_t nb_table)
{
if (tcet->nb_table) {
Convert error_report() to warn_report() Convert all uses of error_report("warning:"... to use warn_report() instead. This helps standardise on a single method of printing warnings to the user. All of the warnings were changed using these two commands: find ./* -type f -exec sed -i \ 's|error_report(".*warning[,:] |warn_report("|Ig' {} + Indentation fixed up manually afterwards. The test-qdev-global-props test case was manually updated to ensure that this patch passes make check (as the test cases are case sensitive). Signed-off-by: Alistair Francis <alistair.francis@xilinx.com> Suggested-by: Thomas Huth <thuth@redhat.com> Cc: Jeff Cody <jcody@redhat.com> Cc: Kevin Wolf <kwolf@redhat.com> Cc: Max Reitz <mreitz@redhat.com> Cc: Ronnie Sahlberg <ronniesahlberg@gmail.com> Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: Peter Lieven <pl@kamp.de> Cc: Josh Durgin <jdurgin@redhat.com> Cc: "Richard W.M. Jones" <rjones@redhat.com> Cc: Markus Armbruster <armbru@redhat.com> Cc: Peter Crosthwaite <crosthwaite.peter@gmail.com> Cc: Richard Henderson <rth@twiddle.net> Cc: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com> Cc: Greg Kurz <groug@kaod.org> Cc: Rob Herring <robh@kernel.org> Cc: Peter Maydell <peter.maydell@linaro.org> Cc: Peter Chubb <peter.chubb@nicta.com.au> Cc: Eduardo Habkost <ehabkost@redhat.com> Cc: Marcel Apfelbaum <marcel@redhat.com> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Igor Mammedov <imammedo@redhat.com> Cc: David Gibson <david@gibson.dropbear.id.au> Cc: Alexander Graf <agraf@suse.de> Cc: Gerd Hoffmann <kraxel@redhat.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Cornelia Huck <cohuck@redhat.com> Cc: Stefan Hajnoczi <stefanha@redhat.com> Acked-by: David Gibson <david@gibson.dropbear.id.au> Acked-by: Greg Kurz <groug@kaod.org> Acked-by: Cornelia Huck <cohuck@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> Reviewed by: Peter Chubb <peter.chubb@data61.csiro.au> Acked-by: Max Reitz <mreitz@redhat.com> Acked-by: Marcel Apfelbaum <marcel@redhat.com> Message-Id: <e1cfa2cd47087c248dd24caca9c33d9af0c499b0.1499866456.git.alistair.francis@xilinx.com> Reviewed-by: Markus Armbruster <armbru@redhat.com> Signed-off-by: Markus Armbruster <armbru@redhat.com>
2017-07-12 16:57:41 +03:00
warn_report("trying to enable already enabled TCE table");
spapr_iommu: Introduce "enabled" state for TCE table Currently TCE tables are created once at start and their sizes never change. We are going to change that by introducing a Dynamic DMA windows support where DMA configuration may change during the guest execution. This changes spapr_tce_new_table() to create an empty zero-size IOMMU memory region (IOMMU MR). Only LIOBN is assigned by the time of creation. It still will be called once at the owner object (VIO or PHB) creation. This introduces an "enabled" state for TCE table objects, some helper functions are added: - spapr_tce_table_enable() receives TCE table parameters, stores in sPAPRTCETable and allocates a guest view of the TCE table (in the user space or KVM) and sets the correct size on the IOMMU MR; - spapr_tce_table_disable() disposes the table and resets the IOMMU MR size; it is made public as the following DDW code will be using it. This changes the PHB reset handler to do the default DMA initialization instead of spapr_phb_realize(). This does not make differenct now but later with more than just one DMA window, we will have to remove them all and create the default one on a system reset. No visible change in behaviour is expected except the actual table will be reallocated every reset. We might optimize this later. The other way to implement this would be dynamically create/remove the TCE table QOM objects but this would make migration impossible as the migration code expects all QOM objects to exist at the receiver so we have to have TCE table objects created when migration begins. Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2016-06-01 11:57:33 +03:00
return;
}
tcet->bus_offset = bus_offset;
tcet->page_shift = page_shift;
tcet->nb_table = nb_table;
tcet->table = spapr_tce_alloc_table(tcet->liobn,
tcet->page_shift,
tcet->bus_offset,
spapr_iommu: Introduce "enabled" state for TCE table Currently TCE tables are created once at start and their sizes never change. We are going to change that by introducing a Dynamic DMA windows support where DMA configuration may change during the guest execution. This changes spapr_tce_new_table() to create an empty zero-size IOMMU memory region (IOMMU MR). Only LIOBN is assigned by the time of creation. It still will be called once at the owner object (VIO or PHB) creation. This introduces an "enabled" state for TCE table objects, some helper functions are added: - spapr_tce_table_enable() receives TCE table parameters, stores in sPAPRTCETable and allocates a guest view of the TCE table (in the user space or KVM) and sets the correct size on the IOMMU MR; - spapr_tce_table_disable() disposes the table and resets the IOMMU MR size; it is made public as the following DDW code will be using it. This changes the PHB reset handler to do the default DMA initialization instead of spapr_phb_realize(). This does not make differenct now but later with more than just one DMA window, we will have to remove them all and create the default one on a system reset. No visible change in behaviour is expected except the actual table will be reallocated every reset. We might optimize this later. The other way to implement this would be dynamically create/remove the TCE table QOM objects but this would make migration impossible as the migration code expects all QOM objects to exist at the receiver so we have to have TCE table objects created when migration begins. Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2016-06-01 11:57:33 +03:00
tcet->nb_table,
&tcet->fd,
tcet->need_vfio);
memory_region_set_size(MEMORY_REGION(&tcet->iommu),
spapr_iommu: Introduce "enabled" state for TCE table Currently TCE tables are created once at start and their sizes never change. We are going to change that by introducing a Dynamic DMA windows support where DMA configuration may change during the guest execution. This changes spapr_tce_new_table() to create an empty zero-size IOMMU memory region (IOMMU MR). Only LIOBN is assigned by the time of creation. It still will be called once at the owner object (VIO or PHB) creation. This introduces an "enabled" state for TCE table objects, some helper functions are added: - spapr_tce_table_enable() receives TCE table parameters, stores in sPAPRTCETable and allocates a guest view of the TCE table (in the user space or KVM) and sets the correct size on the IOMMU MR; - spapr_tce_table_disable() disposes the table and resets the IOMMU MR size; it is made public as the following DDW code will be using it. This changes the PHB reset handler to do the default DMA initialization instead of spapr_phb_realize(). This does not make differenct now but later with more than just one DMA window, we will have to remove them all and create the default one on a system reset. No visible change in behaviour is expected except the actual table will be reallocated every reset. We might optimize this later. The other way to implement this would be dynamically create/remove the TCE table QOM objects but this would make migration impossible as the migration code expects all QOM objects to exist at the receiver so we have to have TCE table objects created when migration begins. Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2016-06-01 11:57:33 +03:00
(uint64_t)tcet->nb_table << tcet->page_shift);
memory_region_add_subregion(&tcet->root, tcet->bus_offset,
MEMORY_REGION(&tcet->iommu));
spapr_iommu: Introduce "enabled" state for TCE table Currently TCE tables are created once at start and their sizes never change. We are going to change that by introducing a Dynamic DMA windows support where DMA configuration may change during the guest execution. This changes spapr_tce_new_table() to create an empty zero-size IOMMU memory region (IOMMU MR). Only LIOBN is assigned by the time of creation. It still will be called once at the owner object (VIO or PHB) creation. This introduces an "enabled" state for TCE table objects, some helper functions are added: - spapr_tce_table_enable() receives TCE table parameters, stores in sPAPRTCETable and allocates a guest view of the TCE table (in the user space or KVM) and sets the correct size on the IOMMU MR; - spapr_tce_table_disable() disposes the table and resets the IOMMU MR size; it is made public as the following DDW code will be using it. This changes the PHB reset handler to do the default DMA initialization instead of spapr_phb_realize(). This does not make differenct now but later with more than just one DMA window, we will have to remove them all and create the default one on a system reset. No visible change in behaviour is expected except the actual table will be reallocated every reset. We might optimize this later. The other way to implement this would be dynamically create/remove the TCE table QOM objects but this would make migration impossible as the migration code expects all QOM objects to exist at the receiver so we have to have TCE table objects created when migration begins. Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2016-06-01 11:57:33 +03:00
}
void spapr_tce_table_disable(sPAPRTCETable *tcet)
spapr_iommu: Introduce "enabled" state for TCE table Currently TCE tables are created once at start and their sizes never change. We are going to change that by introducing a Dynamic DMA windows support where DMA configuration may change during the guest execution. This changes spapr_tce_new_table() to create an empty zero-size IOMMU memory region (IOMMU MR). Only LIOBN is assigned by the time of creation. It still will be called once at the owner object (VIO or PHB) creation. This introduces an "enabled" state for TCE table objects, some helper functions are added: - spapr_tce_table_enable() receives TCE table parameters, stores in sPAPRTCETable and allocates a guest view of the TCE table (in the user space or KVM) and sets the correct size on the IOMMU MR; - spapr_tce_table_disable() disposes the table and resets the IOMMU MR size; it is made public as the following DDW code will be using it. This changes the PHB reset handler to do the default DMA initialization instead of spapr_phb_realize(). This does not make differenct now but later with more than just one DMA window, we will have to remove them all and create the default one on a system reset. No visible change in behaviour is expected except the actual table will be reallocated every reset. We might optimize this later. The other way to implement this would be dynamically create/remove the TCE table QOM objects but this would make migration impossible as the migration code expects all QOM objects to exist at the receiver so we have to have TCE table objects created when migration begins. Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2016-06-01 11:57:33 +03:00
{
if (!tcet->nb_table) {
return;
}
memory_region_del_subregion(&tcet->root, MEMORY_REGION(&tcet->iommu));
memory_region_set_size(MEMORY_REGION(&tcet->iommu), 0);
spapr_iommu: Introduce "enabled" state for TCE table Currently TCE tables are created once at start and their sizes never change. We are going to change that by introducing a Dynamic DMA windows support where DMA configuration may change during the guest execution. This changes spapr_tce_new_table() to create an empty zero-size IOMMU memory region (IOMMU MR). Only LIOBN is assigned by the time of creation. It still will be called once at the owner object (VIO or PHB) creation. This introduces an "enabled" state for TCE table objects, some helper functions are added: - spapr_tce_table_enable() receives TCE table parameters, stores in sPAPRTCETable and allocates a guest view of the TCE table (in the user space or KVM) and sets the correct size on the IOMMU MR; - spapr_tce_table_disable() disposes the table and resets the IOMMU MR size; it is made public as the following DDW code will be using it. This changes the PHB reset handler to do the default DMA initialization instead of spapr_phb_realize(). This does not make differenct now but later with more than just one DMA window, we will have to remove them all and create the default one on a system reset. No visible change in behaviour is expected except the actual table will be reallocated every reset. We might optimize this later. The other way to implement this would be dynamically create/remove the TCE table QOM objects but this would make migration impossible as the migration code expects all QOM objects to exist at the receiver so we have to have TCE table objects created when migration begins. Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2016-06-01 11:57:33 +03:00
spapr_tce_free_table(tcet->table, tcet->fd, tcet->nb_table);
tcet->fd = -1;
tcet->table = NULL;
tcet->bus_offset = 0;
tcet->page_shift = 0;
tcet->nb_table = 0;
}
static void spapr_tce_table_unrealize(DeviceState *dev, Error **errp)
{
sPAPRTCETable *tcet = SPAPR_TCE_TABLE(dev);
QLIST_REMOVE(tcet, list);
spapr_iommu: Introduce "enabled" state for TCE table Currently TCE tables are created once at start and their sizes never change. We are going to change that by introducing a Dynamic DMA windows support where DMA configuration may change during the guest execution. This changes spapr_tce_new_table() to create an empty zero-size IOMMU memory region (IOMMU MR). Only LIOBN is assigned by the time of creation. It still will be called once at the owner object (VIO or PHB) creation. This introduces an "enabled" state for TCE table objects, some helper functions are added: - spapr_tce_table_enable() receives TCE table parameters, stores in sPAPRTCETable and allocates a guest view of the TCE table (in the user space or KVM) and sets the correct size on the IOMMU MR; - spapr_tce_table_disable() disposes the table and resets the IOMMU MR size; it is made public as the following DDW code will be using it. This changes the PHB reset handler to do the default DMA initialization instead of spapr_phb_realize(). This does not make differenct now but later with more than just one DMA window, we will have to remove them all and create the default one on a system reset. No visible change in behaviour is expected except the actual table will be reallocated every reset. We might optimize this later. The other way to implement this would be dynamically create/remove the TCE table QOM objects but this would make migration impossible as the migration code expects all QOM objects to exist at the receiver so we have to have TCE table objects created when migration begins. Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
2016-06-01 11:57:33 +03:00
spapr_tce_table_disable(tcet);
}
MemoryRegion *spapr_tce_get_iommu(sPAPRTCETable *tcet)
{
return &tcet->root;
}
static void spapr_tce_reset(DeviceState *dev)
{
sPAPRTCETable *tcet = SPAPR_TCE_TABLE(dev);
size_t table_size = tcet->nb_table * sizeof(uint64_t);
if (tcet->nb_table) {
memset(tcet->table, 0, table_size);
}
}
static target_ulong put_tce_emu(sPAPRTCETable *tcet, target_ulong ioba,
target_ulong tce)
{
IOMMUTLBEntry entry;
hwaddr page_mask = IOMMU_PAGE_MASK(tcet->page_shift);
unsigned long index = (ioba - tcet->bus_offset) >> tcet->page_shift;
if (index >= tcet->nb_table) {
hcall_dprintf("spapr_vio_put_tce on out-of-bounds IOBA 0x"
TARGET_FMT_lx "\n", ioba);
return H_PARAMETER;
}
tcet->table[index] = tce;
entry.target_as = &address_space_memory,
entry.iova = (ioba - tcet->bus_offset) & page_mask;
entry.translated_addr = tce & page_mask;
entry.addr_mask = ~page_mask;
entry.perm = spapr_tce_iommu_access_flags(tce);
memory_region_notify_iommu(&tcet->iommu, entry);
return H_SUCCESS;
}
static target_ulong h_put_tce_indirect(PowerPCCPU *cpu,
sPAPRMachineState *spapr,
target_ulong opcode, target_ulong *args)
{
int i;
target_ulong liobn = args[0];
target_ulong ioba = args[1];
target_ulong ioba1 = ioba;
target_ulong tce_list = args[2];
target_ulong npages = args[3];
target_ulong ret = H_PARAMETER, tce = 0;
sPAPRTCETable *tcet = spapr_tce_find_by_liobn(liobn);
CPUState *cs = CPU(cpu);
hwaddr page_mask, page_size;
if (!tcet) {
return H_PARAMETER;
}
if ((npages > 512) || (tce_list & SPAPR_TCE_PAGE_MASK)) {
return H_PARAMETER;
}
page_mask = IOMMU_PAGE_MASK(tcet->page_shift);
page_size = IOMMU_PAGE_SIZE(tcet->page_shift);
ioba &= page_mask;
for (i = 0; i < npages; ++i, ioba += page_size) {
tce = ldq_be_phys(cs->as, tce_list + i * sizeof(target_ulong));
ret = put_tce_emu(tcet, ioba, tce);
if (ret) {
break;
}
}
/* Trace last successful or the first problematic entry */
i = i ? (i - 1) : 0;
if (SPAPR_IS_PCI_LIOBN(liobn)) {
trace_spapr_iommu_pci_indirect(liobn, ioba1, tce_list, i, tce, ret);
} else {
trace_spapr_iommu_indirect(liobn, ioba1, tce_list, i, tce, ret);
}
return ret;
}
static target_ulong h_stuff_tce(PowerPCCPU *cpu, sPAPRMachineState *spapr,
target_ulong opcode, target_ulong *args)
{
int i;
target_ulong liobn = args[0];
target_ulong ioba = args[1];
target_ulong tce_value = args[2];
target_ulong npages = args[3];
target_ulong ret = H_PARAMETER;
sPAPRTCETable *tcet = spapr_tce_find_by_liobn(liobn);
hwaddr page_mask, page_size;
if (!tcet) {
return H_PARAMETER;
}
if (npages > tcet->nb_table) {
return H_PARAMETER;
}
page_mask = IOMMU_PAGE_MASK(tcet->page_shift);
page_size = IOMMU_PAGE_SIZE(tcet->page_shift);
ioba &= page_mask;
for (i = 0; i < npages; ++i, ioba += page_size) {
ret = put_tce_emu(tcet, ioba, tce_value);
if (ret) {
break;
}
}
if (SPAPR_IS_PCI_LIOBN(liobn)) {
trace_spapr_iommu_pci_stuff(liobn, ioba, tce_value, npages, ret);
} else {
trace_spapr_iommu_stuff(liobn, ioba, tce_value, npages, ret);
}
return ret;
}
static target_ulong h_put_tce(PowerPCCPU *cpu, sPAPRMachineState *spapr,
target_ulong opcode, target_ulong *args)
{
target_ulong liobn = args[0];
target_ulong ioba = args[1];
target_ulong tce = args[2];
target_ulong ret = H_PARAMETER;
sPAPRTCETable *tcet = spapr_tce_find_by_liobn(liobn);
if (tcet) {
hwaddr page_mask = IOMMU_PAGE_MASK(tcet->page_shift);
ioba &= page_mask;
ret = put_tce_emu(tcet, ioba, tce);
}
if (SPAPR_IS_PCI_LIOBN(liobn)) {
trace_spapr_iommu_pci_put(liobn, ioba, tce, ret);
} else {
trace_spapr_iommu_put(liobn, ioba, tce, ret);
}
return ret;
}
static target_ulong get_tce_emu(sPAPRTCETable *tcet, target_ulong ioba,
target_ulong *tce)
{
unsigned long index = (ioba - tcet->bus_offset) >> tcet->page_shift;
if (index >= tcet->nb_table) {
hcall_dprintf("spapr_iommu_get_tce on out-of-bounds IOBA 0x"
TARGET_FMT_lx "\n", ioba);
return H_PARAMETER;
}
*tce = tcet->table[index];
return H_SUCCESS;
}
static target_ulong h_get_tce(PowerPCCPU *cpu, sPAPRMachineState *spapr,
target_ulong opcode, target_ulong *args)
{
target_ulong liobn = args[0];
target_ulong ioba = args[1];
target_ulong tce = 0;
target_ulong ret = H_PARAMETER;
sPAPRTCETable *tcet = spapr_tce_find_by_liobn(liobn);
if (tcet) {
hwaddr page_mask = IOMMU_PAGE_MASK(tcet->page_shift);
ioba &= page_mask;
ret = get_tce_emu(tcet, ioba, &tce);
if (!ret) {
args[0] = tce;
}
}
if (SPAPR_IS_PCI_LIOBN(liobn)) {
trace_spapr_iommu_pci_get(liobn, ioba, ret, tce);
} else {
trace_spapr_iommu_get(liobn, ioba, ret, tce);
}
return ret;
}
int spapr_dma_dt(void *fdt, int node_off, const char *propname,
uint32_t liobn, uint64_t window, uint32_t size)
{
uint32_t dma_prop[5];
int ret;
dma_prop[0] = cpu_to_be32(liobn);
dma_prop[1] = cpu_to_be32(window >> 32);
dma_prop[2] = cpu_to_be32(window & 0xFFFFFFFF);
dma_prop[3] = 0; /* window size is 32 bits */
dma_prop[4] = cpu_to_be32(size);
ret = fdt_setprop_cell(fdt, node_off, "ibm,#dma-address-cells", 2);
if (ret < 0) {
return ret;
}
ret = fdt_setprop_cell(fdt, node_off, "ibm,#dma-size-cells", 2);
if (ret < 0) {
return ret;
}
ret = fdt_setprop(fdt, node_off, propname, dma_prop, sizeof(dma_prop));
if (ret < 0) {
return ret;
}
return 0;
}
int spapr_tcet_dma_dt(void *fdt, int node_off, const char *propname,
sPAPRTCETable *tcet)
{
if (!tcet) {
return 0;
}
return spapr_dma_dt(fdt, node_off, propname,
tcet->liobn, 0, tcet->nb_table << tcet->page_shift);
}
static void spapr_tce_table_class_init(ObjectClass *klass, void *data)
{
DeviceClass *dc = DEVICE_CLASS(klass);
dc->init = spapr_tce_table_realize;
dc->reset = spapr_tce_reset;
dc->unrealize = spapr_tce_table_unrealize;
QLIST_INIT(&spapr_tce_tables);
/* hcall-tce */
spapr_register_hypercall(H_PUT_TCE, h_put_tce);
spapr_register_hypercall(H_GET_TCE, h_get_tce);
spapr_register_hypercall(H_PUT_TCE_INDIRECT, h_put_tce_indirect);
spapr_register_hypercall(H_STUFF_TCE, h_stuff_tce);
}
static TypeInfo spapr_tce_table_info = {
.name = TYPE_SPAPR_TCE_TABLE,
.parent = TYPE_DEVICE,
.instance_size = sizeof(sPAPRTCETable),
.class_init = spapr_tce_table_class_init,
};
static void spapr_iommu_memory_region_class_init(ObjectClass *klass, void *data)
{
IOMMUMemoryRegionClass *imrc = IOMMU_MEMORY_REGION_CLASS(klass);
imrc->translate = spapr_tce_translate_iommu;
imrc->get_min_page_size = spapr_tce_get_min_page_size;
imrc->notify_flag_changed = spapr_tce_notify_flag_changed;
}
static const TypeInfo spapr_iommu_memory_region_info = {
.parent = TYPE_IOMMU_MEMORY_REGION,
.name = TYPE_SPAPR_IOMMU_MEMORY_REGION,
.class_init = spapr_iommu_memory_region_class_init,
};
static void register_types(void)
{
type_register_static(&spapr_tce_table_info);
type_register_static(&spapr_iommu_memory_region_info);
}
type_init(register_types);