Use meson benchmark() for them, adjust mtest2make.py for that.
A new target "make bench" can be used to run all benchmarks.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20200828110734.1638685-14-marcandre.lureau@redhat.com>
[Rewrite mtest2make part. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
As this makes the TAP output invalid. Use g_test_message().
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20200828110734.1638685-13-marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
It was covered already in commit d3ca592b3c ("meson: convert check-block")
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20200828110734.1638685-9-marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Since there is not minimum version specified, and it's a test-only
dependency, it's fair to depend on a version that ships with a .pc I
suppose.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20200828110734.1638685-8-marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
UEFI on ia32 never really took off. Basically the BIOS -> UEFI shift
came too late, x64 was widespread already, so vendors went from BIOS
straight to UEFI on x64.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
master snapshot, most recent commit from Aug 23th
shortlog
========
Andrew Grasso (1):
[intel] Add INTEL_NO_PHY_RST for I218LM
Ben Bleything (1):
[intel] Set INTEL_NO_PHY_RST on i218v
Bobby Lockwood (1):
[libgcc] Change __divmoddi4 from int64 [unknown] to int64_t
Bruce Rogers (1):
[build] Be explicit about -fcommon compiler directive
Christian Nilsson (1):
[intel] Add PCI ID for I219-V and -LM 6 to 9
Daniel Johnson (1):
[deflate] Fix typo in comment describing length codes
David Decotigny (2):
[pcbios] Fix "out of memory" detection when expanding bottom area
[pcbios] Take alignment into account when checking for available space
Dentcho Ludmilov Bankov (1):
[comboot] Fix stack pointer retrieval after COM32 binary returns
Ignat Korchagin (3):
[snp] Try promiscuous multicast receive filter if the regular one fails
[snp] Set EFI_SIMPLE_NETWORK_RECEIVE_MULTICAST bit as per UEFI spec
[efi] Check the status code from AllocatePool()
Joe Groocock (1):
[cmdline] Add "--timeout" parameter to "ifconf" command
Joseph Wong (1):
[bnxt] Add driver support for Broadcom NetXtreme-E Adapters
LaDerrick H (1):
[sfc] Add PCI ID for Solarflare SFC9120 and Solarflare SFC9140
Michael Brown (105):
[init] Show startup and shutdown function names in debug messages
[efi] Blacklist the Dell Ip4ConfigDxe driver
[tls] Support stateful session resumption
[tls] Fix incorrectly duplicated error number
[tls] Support stateless session resumption
[crypto] Use x509_name() in validator debug messages
[tls] Display cross-certificate and OCSP status messages
[tls] Display validator messages only while validation is in progress
[tcp] Display "connecting" status until connection is established
[ocsp] Accept response certID with missing hashAlgorithm parameters
[pci] Add support for PCI MSI-X interrupts
[vlan] Provide vlan_netdev_rx() and vlan_netdev_rx_err()
[intelxl] Use VLAN tag in receive descriptor if present
[intelxl] Allow for virtual function admin queue register maps
[intelxl] Use one admin queue buffer per admin queue descriptor
[intelxl] Allow admin queues to be reinitialised
[intelxl] Allow admin cookie to hold extended opcode and return code
[intelxl] Provide a mechanism for handling "send to VF" events
[intelxl] Use 32-byte receive descriptors
[intelxl] Allow for arbitrary placement of ring tail registers
[intelxl] Split out ring creation from context programming
[intelxl] Allow for arbitrary placement of interrupt control register
[intelxl] Expose functions required by virtual function driver
[intelxl] Choose to operate in non-PXE mode
[intelxl] Add driver for Intel 40 Gigabit Ethernet NIC virtual functions
[build] Fix use of inline assembly on GCC 8 ARM64 builds
[arm] Provide dummy implementations for {in,out}[s]{b,w,l}
[efi] Return only registered EFI devices from efidev_parent()
[fdt] Add ability to parse a MAC address from a flattened device tree
[efi] Register a device tree if provided by the platform firmware
[smsc95xx] Fetch MAC from device tree for Raspberry Pi
[build] Add named configuration for Raspberry Pi
[smscusb] Fetch MAC from device tree for Raspberry Pi Model B+
[build] Do not apply WORKAROUND_CFLAGS for host compiler
[build] Move predefined all-drivers build shortcut to Makefile
[build] Add predefined shortcut for Raspberry Pi builds
[process] Add PROC_INIT() for initialising static processes
[peerdist] Start block download timers from within opener methods
[peerdist] Limit number of concurrent raw block downloads
[tls] Add missing call to tls_tx_resume() when restarting negotiation
[crypto] Drag in configured digestInfo prefixes for any use of RSA
[crypto] Profile the various stages of modular multiplication
[coverity] Override assumptions about wcrtomb() and hmac_init()
[ethernet] Avoid false positive Coverity warning
[golan] Fix address-of-pointer bug for multicast attach/detach
[efi] Report failed control transfers as expected by the USB core
[efi] Do not attempt EFI_USB_IO_PROTOCOL transfers during shutdown
[lan78xx] Always enable automatic speed and duplex detection
[peerdist] Allow PeerDist to be globally enabled or disabled
[peerdist] Allow for the use of a hosted cache server
[build] Construct full version number automatically from git revision
[travis] Ensure that most recent tag is always available
[tftp] Eliminate unnecessary variable-length stack allocation
[infiniband] Eliminate variable-length stack allocation
[slam] Eliminate variable-length stack allocation
[slam] Allow for the possibility of IPv6 multicast addresses
[settings] Eliminate variable-length stack allocation
[iscsi] Eliminate variable-length stack allocations in CHAP handlers
[iscsi] Eliminate variable-length stack allocation in URI parsing
[bios] Define macros for constructing partition table entries
[int13con] Create log partition only when CONSOLE_INT13 is enabled
[efi] Disambiguate errors returned by ConnectController
[png] Fix potential integer overflow
[efi] Work around UEFI specification bug in LoadImage
[efi] Work around UEFI specification bug in LoadImage for SAN boot
[uri] Avoid appearing to access final byte of a potentially empty string
[efi] Attempt to connect our driver directly if ConnectController fails
[tls] Allow a minimum TLS protocol version to be specified
[tls] Default to supporting only TLSv1.1 or above
[crypto] Allow algorithms to be included without being OID-identifiable
[crypto] Ensure that test code drags in required ASN.1 object identifiers
[crypto] Disable MD5 as an OID-identifiable algorithm by default
[snp] Retry initialisation if link is reported as down
[util] Treat empty integer strings as invalid
[parseopt] Treat empty integer strings in user input as invalid
[efi] Enable stack protection where possible
[ocsp] Accept SHA1 certID responses even if SHA1 is not enabled
[intel] Avoid spurious compiler warning on GCC 10
[golan] Add explicit type casts for nodnic_queue_pair_type
[build] Disable position-independent code for ARM64 EFI builds
[efi] Detect and disable seriously broken EFI_RNG_PROTOCOL implementations
[efi] Raise TPL during driver entry point
[axge] Handle non-gigabit link speeds
[usb] Clear device endpoint halt before resetting host endpoint
[usb] Add missing usb_recycle() for completed hub interrupt transfers
[usb] Do not attempt to disable USB3 hub ports
[xhci] Set link state to RxDetect after disabling USB3 root hub port
[usb] Avoid unnecessary calls to usb_hub_set_drvdata()
[xhci] Increase link state settling delay to 100ms
[axge] Reapply USB device configuration when opening network device
[usb] Leave port enabled after a failed device registration
[crypto] Avoid temporarily setting direction flag in bigint_is_geq()
[libc] Provide an unoptimised generic_memcpy_reverse()
[efi] Avoid setting direction flag on EFI platforms
[libc] Add bit-rotation functions for unsigned long values
[efi] Distribute available entropy within stack cookie
[efi] Ensure NUL byte is at lowest address within stack cookie
[efi] Claim SNP devices early in efi_download_start()
[xfer] Remove address family from definition of a socket opener
[dns] Use all configured DNS servers
[wpa] Fix erroneous debug message in wpa_derive_ptk
[efi] Match EDK2 numbering for USB ports
[efi] Use device path to locate filesystem from which we were loaded
[efi] Attempt NII initialisation both with and without cable detection
[build] Fix building on older versions of gcc
Michael J. Bazzinotti (1):
[libc] Fix memcmp() to return proper values
Mohammed (1):
[golan] Add various new PCI device IDs
Rob Taglang (1):
[intelx] Added PCI_ROM entry for Intel x553/x557-AT and x553 (SFP+) NICs
Tobias Kortkamp (1):
[build] Fix a GNUism that FreeBSD's sed(1) cannot deal with
Valentine Barshak (2):
[build] Fix "'%s' directive argument is null" error
[build] Workaround compilation error with gcc 9.1
bit (1):
[build] Fix default target in sdsk image
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
This is required by test-qht-par unit test.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20200828110734.1638685-5-marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Rename the variable to be more explicit. A further clean-up patch will
move the actual to dependency check to meson entirely.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20200828110734.1638685-4-marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
- builds QAPI builtins types/visitor to fix a linking issue with
unresolved symbols in the static library.
- work around a meson limitation on generated file output directories.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20200828110734.1638685-2-marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Add genh to the sources to avoid race conditions between QAPI
file generation and libqos compilation.
Make the name_suffix .fa for consistency with other link_whole
static libraries and to work around a Meson issue where
lots of linker flags are placed between -Wl,--start-group and
-Wl,--end-group and this breaks the fork-fuzz.ld linker script.
Reported-by: Claudio Fontana <cfontana@suse.de>
Reported-by: Alexander Bulekov <alxndr@bu.edu>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Numbering files according to rules causes confusion, because
CUSTOM_COMMAND3.stamp from a previous build might represent
completely different targets after Makefile.ninja is regenerated.
As a result, the new targets are not rebuilt and compilation
fails.
Use the targets to build a SHA1 hash; the chances for collision
are one in 2^24 even with a 12-character prefix of the hash.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
On msys2, paths such as -L/e/path/to/qemu are not recognized by
the linker. Fortunately we do not need absolute paths at all in a
non-recursive build system.
Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Meson automatically adds "-undefined dynamic_lookup" to
shared_module build targets; b_lundef is only needed for
executables. Therefore, we can remove this option.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Whenever a test appears in multiple suites, the rules generated
by mtest2make are currently running it twice. Instead, after
this patch we generate a phony target for each test and we have
a generic "run-tests" target depend on all the tests that were
chosen on the command line. Tests that appear in multiple suites
will be added to the prerequisites just once.
This has other advantages: it removes the handling of -k and
it increases parallelism.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The softfloat tests are quite noisy; before the Meson conversion
they buffered the output in a file and emitted the output only
if the test failed. Tweak mtest2make.py so that the courtesy
is extended to all non-TAP tests.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Pass the working directory and test command in separate macro arguments,
so that we will be able to insert a test driver in the next patch.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Pass the environment and test command in separate macro arguments,
so that we will be able to insert a test driver in the next patch.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
check-block has its own test harness, unlike every other test. If
we capture its output, as is in general nicer to do without V=1,
there will be no sign of progress. So for lack of a better option
just move the invocation of the test back to Makefile rules.
As a side effect, this will also fix "make check" in --disable-tools
builds, as they were trying to run qemu-iotests without having
made qemu-img before.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The current implementation of h_home_node_associativity hard codes
the values of associativity domains of the vcpus. Let's make
it consider the values already initialized in spapr->numa_assoc_array,
via the spapr_numa_get_vcpu_assoc() helper.
We want to set it and forget it, and for that we also need to
assert that we don't overflow the registers of the hypercall.
>From R4 to R9 we can squeeze in 12 associativity domains for
vcpus, so let's assert that VCPU_ASSOC_SIZE -1 isn't greater
than that.
Reviewed-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20200904172422.617460-4-danielhb413@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The work to be done in h_home_node_associativity() intersects
with what is already done in spapr_numa_fixup_cpu_dt(). This
patch creates a new helper, spapr_numa_get_vcpu_assoc(), to
be used for both spapr_numa_fixup_cpu_dt() and
h_home_node_associativity().
While we're at it, use memcpy() instead of loop assignment
to created the returned array.
Reviewed-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20200904172422.617460-3-danielhb413@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The implementation of this hypercall will be modified to use
spapr->numa_assoc_arrays input. Moving it to spapr_numa.c makes
make more sense.
Reviewed-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20200904172422.617460-2-danielhb413@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The NVLink2 GPUs works like a regular NUMA node with its
own associativity values, regardless of user input.
This can be handled inside spapr_numa_associativity_init(),
initializing NVGPU_MAX_NUM associativity arrays that can
be used by the GPUs.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20200903220639.563090-5-danielhb413@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
In a similar fashion as the previous patch, let's move the
handling of ibm,associativity-lookup-arrays from spapr.c to
spapr_numa.c. A spapr_numa_write_assoc_lookup_arrays() helper was
created, and spapr_dt_dynamic_reconfiguration_memory() can now
use it to advertise the lookup-arrays.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20200903220639.563090-4-danielhb413@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Vcpus have an additional paramenter to be appended, vcpu_id. This
also changes the size of the of property itself, which is being
represented in index 0 of numa_assoc_array[cpu->node_id],
and defaults to MAX_DISTANCE_REF_POINTS for all cases but
vcpus.
All this logic makes more sense in spapr_numa.c, where we handle
everything NUMA and associativity. A new helper spapr_numa_fixup_cpu_dt()
was added, and spapr.c uses it the same way as it was using the former
spapr_fixup_cpu_numa_dt().
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20200903220639.563090-3-danielhb413@gmail.com>
[dwg: Correct uint to int type, which can break windows builds]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The next step to centralize all NUMA/associativity handling in
the spapr machine is to create a 'one stop place' for all
things ibm,associativity.
This patch introduces numa_assoc_array, a 2 dimensional array
that will store all ibm,associativity arrays of all NUMA nodes.
This array is initialized in a new spapr_numa_associativity_init()
function, called in spapr_machine_init(). It is being initialized
with the same values used in other ibm,associativity properties
around spapr files (i.e. all zeros, last value is node_id).
The idea is to remove all hardcoded definitions and FDT writes
of ibm,associativity arrays, doing instead a call to the new
helper spapr_numa_write_associativity_dt() helper, that will
be able to write the DT with the correct values.
We'll start small, handling the trivial cases first. The
remaining instances of ibm,associativity will be handled
next.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20200903220639.563090-2-danielhb413@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
This function is only used inside spapr_nvdimm.c.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20200901125645.118026-3-danielhb413@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
We're going to make changes in how spapr handles all
ibm,associativity* related properties to enhance our current NUMA
support.
At this moment we have associativity code scattered all around
spapr_* files, with hardcoded values and array sizes. This
makes it harder to change any NUMA specific parameters in
the future. Having everything in the same place allows not
only for easier tuning, but also easier understanding since all
NUMA related code is on the same file.
This patch introduces a new file to gather all NUMA/associativity
handling code in spapr, spapr_numa.c. To get things started, let's
remove associativity-reference-points and max-associativity-domains
code from spapr_dt_rtas() to a new helper called spapr_numa_write_rtas_dt().
This will decouple spapr_dt_rtas() from the NUMA changes that
are going to happen in those two properties.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20200901125645.118026-2-danielhb413@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
We call pci_register_root_bus() to register 4 IRQs with the
ppc4xx_pci_set_irq() handler. As it can only be called with
values in the [0-4[ range, replace the pointless warning by
an assert().
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20200901104043.91383-5-f4bug@amsat.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Replace the magic '4' by ARRAY_SIZE(s->irq) which is more explicit.
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20200901104043.91383-4-f4bug@amsat.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Instead of setting CPUState::halted to 1 in s390_cpu_initfn(), use the
start-powered-off property which makes cpu_common_reset() initialize it
to 1 in common code.
Note that this changes behavior by setting cs->halted to 1 on reset, which
didn't happen before.
Acked-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Thiago Jung Bauermann <bauerman@linux.ibm.com>
Message-Id: <20200826055535.951207-9-bauerman@linux.ibm.com>
[dwg: Fix from Laurent Vivier for user only case]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Instead of setting CPUState::halted to 1 in secondary_cpu_reset(), use the
start-powered-off property which makes cpu_common_reset() initialize it
to 1 in common code.
Now secondary_cpu_reset() becomes equivalent to main_cpu_reset() so rename
the function to sun4m_cpu_reset().
Also remove setting of cs->halted from cpu_devinit(), which seems out of
place when compared to similar code in other architectures (e.g.,
ppce500_init() in hw/ppc/e500.c).
Finally, change creation of CPU object from cpu_create() to object_new()
and qdev_realize_and_unref() because cpu_create() realizes the CPU and it's
not possible to set a property after the object is realized.
Suggested-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Thiago Jung Bauermann <bauerman@linux.ibm.com>
Message-Id: <20200826055535.951207-8-bauerman@linux.ibm.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
We rely on cpu_common_reset() to set cs->halted to 0, it's redundant to do
it in main_cpu_reset().
Signed-off-by: Thiago Jung Bauermann <bauerman@linux.ibm.com>
Message-Id: <20200826055535.951207-7-bauerman@linux.ibm.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Instead of setting CPUState::halted to 1 in main_cpu_reset(), use the
start-powered-off property which makes cpu_common_reset() initialize it
to 1 in common code.
Also change creation of CPU object from cpu_create() to object_new() and
qdev_realize_and_unref() because cpu_create() realizes the CPU and it's not
possible to set a property after the object is realized.
Signed-off-by: Thiago Jung Bauermann <bauerman@linux.ibm.com>
Message-Id: <20200826055535.951207-6-bauerman@linux.ibm.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Instead of setting CPUState::halted to 1 in ppce500_cpu_reset_sec(), use
the start-powered-off property which makes cpu_common_reset() initialize it
to 1 in common code.
Also change creation of CPU object from cpu_create() to object_new() and
qdev_realize_and_unref() because cpu_create() realizes the CPU and it's not
possible to set a property after the object is realized.
Signed-off-by: Thiago Jung Bauermann <bauerman@linux.ibm.com>
Message-Id: <20200826055535.951207-5-bauerman@linux.ibm.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
PowerPC sPAPR CPUs start in the halted state, and spapr_reset_vcpu()
attempts to implement this by setting CPUState::halted to 1. But that's too
late for the case of hotplugged CPUs in a machine configure with 2 or more
threads per core.
By then, other parts of QEMU have already caused the vCPU to run in an
unitialized state a couple of times. For example, ppc_cpu_reset() calls
ppc_tlb_invalidate_all(), which ends up calling async_run_on_cpu(). This
kicks the new vCPU while it has CPUState::halted = 0, causing QEMU to issue
a KVM_RUN ioctl on the new vCPU before the guest is able to make the
start-cpu RTAS call to initialize its register state.
This problem doesn't seem to cause visible issues for regular guests, but
on a secure guest running under the Ultravisor it does. The Ultravisor
relies on being able to snoop on the start-cpu RTAS call to map vCPUs to
guests, and this issue causes it to see a stray vCPU that doesn't belong to
any guest.
Fix by setting the start-powered-off CPUState property in
spapr_create_vcpu(), which makes cpu_common_reset() initialize
CPUState::halted to 1 at an earlier moment.
Suggested-by: Eduardo Habkost <ehabkost@redhat.com>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Thiago Jung Bauermann <bauerman@linux.ibm.com>
Message-Id: <20200826055535.951207-4-bauerman@linux.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
This change is in a separate patch because it's not so obvious that it
won't cause a regression.
Suggested-by: Eduardo Habkost <ehabkost@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Thiago Jung Bauermann <bauerman@linux.ibm.com>
Message-Id: <20200826055535.951207-3-bauerman@linux.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
There are other platforms which also have CPUs that start powered off, so
generalize the start-powered-off property so that it can be used by them.
Note that ARMv7MState also has a property of the same name but this patch
doesn't change it because that class isn't a subclass of CPUState so it
wouldn't be a trivial change.
This change should not cause any change in behavior.
Suggested-by: Eduardo Habkost <ehabkost@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Thiago Jung Bauermann <bauerman@linux.ibm.com>
Message-Id: <20200826055535.951207-2-bauerman@linux.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The NVDIMM support for pSeries was introduced in 5.1, but it
didn't contemplate the 'nvdimm' machine option that other
archs uses. For every other arch, if no '-machine nvdimm(=on)'
is present, it is assumed that the NVDIMM support is disabled.
The user must explictly inform that the machine supports
NVDIMM. For pseries-5.1 the 'nvdimm' option is completely
ignored, and support is always assumed to exist. This
leads to situations where the user is able to set 'nvdimm=off'
but the guest boots up with the NVDIMMs anyway.
Fixing this now, after 5.1 launch, can put the overall NVDIMM
support for pseries in a strange place regarding this 'nvdimm'
machine option. If we force everything to be like other archs,
existing pseries-5.1 guests that didn't use 'nvdimm' to use NVDIMM
devices will break. If we attempt to make the newer pseries
machines (5.2+) behave like everyone else, but keep pseries-5.1
untouched, we'll have consistency problems on machine upgrade
(5.1 will have different default values for NVDIMM support than
5.2).
The common ground here is, if the user sets 'nvdimm=off', we
must comply regardless of being 5.1 or 5.2+. This patch
changes spapr_nvdimm_validate() to verify if the user set
NVDIMM support off in the machine options and, in that
case, error out if we have a NVDIMM device. The default
value for 5.2+ pseries machines will still be 'nvdimm=on'
when there is no 'nvdimm' option declared, just like it is today
with pseries-5.1. In the end we'll have different default
semantics from everyone else in the absence of the 'nvdimm'
machine option, but this boat has sailed.
Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1848887
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20200825215749.213536-4-danielhb413@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
NVDIMM has different contraints and conditions than the regular
DIMM and we'll need to add at least one more.
Instead of relying on 'if (nvdimm)' conditionals in the body of
spapr_memory_pre_plug(), use the existing spapr_nvdimm_validate_opts()
and put all NVDIMM handling code there. Rename it to
spapr_nvdimm_validate() to reflect that the function is now checking
more than the nvdimm device options. This makes spapr_memory_pre_plug()
a bit easier to follow, and we can tune in NVDIMM parameters
and validation in the same place.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20200825215749.213536-3-danielhb413@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Since we're using the string just once, just use g_autofree and
avoid leaking it without calling g_free().
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20200825215749.213536-2-danielhb413@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
When QEMU switches to the XIVE interrupt mode, it creates all the
guest interrupts at the level of the KVM device. These interrupts are
backed by real HW interrupts from the IPI interrupt pool of the XIVE
controller.
Currently, this is done from the QEMU main thread, which results in
allocating all interrupts from the chip on which QEMU is running. IPIs
are not distributed across the system and the load is not well
balanced across the interrupt controllers.
Change the vCPU IPI allocation to run from the vCPU context. The
associated XIVE IPI interrupt will be allocated on the chip on which
the vCPU is running and improve distribution of the IPIs in the system.
When the vCPUs are pinned, this will make the IPI local to the chip of
the vCPU. It will reduce rerouting between interrupt controllers and
gives better performance.
Device interrupts are still treated the same. To improve placement, we
would need some information on the chip owning the virtual source or
the HW source in case of a passthrough device but this reuires
changes in PAPR.
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20200820134547.2355743-5-clg@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The vCPU IPIs are now allocated in kvmppc_xive_cpu_connect() when the
vCPU connects to the KVM device and not when all the sources are reset
in kvmppc_xive_source_reset()
This requires extra care for hotplug vCPUs and VM restore.
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20200820134547.2355743-4-clg@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
This is doing an extra loop but should be equivalent.
It also differentiate the reset of the sources from the restore of the
sources configuration. This will help in allocating the vCPU IPIs
independently.
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20200820134547.2355743-3-clg@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
We will use to check if a vCPU IPI has been created.
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20200820134547.2355743-2-clg@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
