During migrations, after each iteration, cpu_throttle_set() is called,
which irrespective of input, re-arms the timer according to value of
new_throttle_pct. This causes cpu_throttle_thread() to be delayed in
getting scheduled and consqeuntly lets guest run for more time than what
the throttle value should allow. This leads to spikes in guest throughput
at high cpu-throttle percentage whenever cpu_throttle_set() is called.
A solution would be not to modify the timer immediately in
cpu_throttle_set(), instead, only modify throttle_percentage so that the
throttle would automatically adjust to the required percentage when
cpu_throttle_timer_tick() is invoked.
Manually tested the patch using following configuration:
Guest:
Centos7 (3.10.0-123.el7.x86_64)
Total Memory - 64GB , CPUs - 16
Tool used - stress (1.0.4)
Workload - stress --vm 32 --vm-bytes 1G --vm-keep
Migration Parameters:
Network Bandwidth - 500MBPS
cpu-throttle-initial - 99
Results:
With timer_mod(): fails to converge, continues indefinitely
Without timer_mod(): converges in 249 sec
Signed-off-by: Utkarsh Tripathi <utkarsh.tripathi@nutanix.com>
Message-Id: <1609420384-119407-1-git-send-email-utkarsh.tripathi@nutanix.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This patch enables using rng-builtin with record/replay
by making the callbacks deterministic.
Signed-off-by: Pavel Dovgalyuk <Pavel.Dovgalyuk@ispras.ru>
Message-Id: <161233201286.170686.7858208964037376305.stgit@pasha-ThinkPad-X280>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Before the change /usr/share/qemu/firmware/50-edk2-x86_64-secure.json
contained the relative path:
"filename": "share/qemu/edk2-x86_64-secure-code.fd",
"filename": "share/qemu/edk2-i386-vars.fd",
After then change the paths are absolute:
"filename": "/usr/share/qemu/edk2-x86_64-secure-code.fd",
"filename": "/usr/share/qemu/edk2-i386-vars.fd",
The regression appeared in qemu-5.2.0 (seems to be related
to meson port).
CC: Paolo Bonzini <pbonzini@redhat.com>
CC: "Marc-André Lureau" <marcandre.lureau@redhat.com>
CC: "Philippe Mathieu-Daudé" <philmd@redhat.com>
Bug: https://bugs.gentoo.org/766743
Bug: https://bugs.launchpad.net/qemu/+bug/1913012
Signed-off-by: Jannik Glückert <jannik.glueckert@gmail.com>
Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org>
Message-Id: <20210131143434.2513363-1-slyfox@gentoo.org>
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Sometimes interrupt event comes at the same time with
the virtual timers. In this case replay tries to proceed
the timers, because deadline for them is zero.
This patch allows processing interrupts and exceptions
by entering the vCPU execution loop, when deadline is zero,
but checkpoint associated with virtual timers is not ready
to be replayed.
Signed-off-by: Pavel Dovgalyuk <Pavel.Dovgalyuk@ispras.ru>
Message-Id: <161216312794.2030770.1709657858900983160.stgit@pasha-ThinkPad-X280>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The kvm_vm_ioctl() wrapper already returns -errno if the ioctl itself
returned -1, so the callers of kvm_vm_ioctl() should not check for -1
but for a value < 0 instead.
This problem has been fixed once already in commit b533f658a9
but that commit missed that the ENOENT error code is not fatal for
this ioctl, so the commit has been reverted in commit 50212d6346
since the problem occurred close to a pending release at that point
in time. The plan was to fix it properly after the release, but it
seems like this has been forgotten. So let's do it now finally instead.
Resolves: https://bugs.launchpad.net/qemu/+bug/1294227
Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20210129084354.42928-1-thuth@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20210122204441.2145197-13-philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20210122204441.2145197-12-philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20210122204441.2145197-11-philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Beside a CPU device, user-mode emulation doesn't access
anything else from qdev subsystem.
Tools don't need anything from qdev.
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20210122204441.2145197-10-philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20210122204441.2145197-9-philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20210122204441.2145197-8-philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The trace_events_subdirs array is split in two different
locations, merge it as one.
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Claudio Fontana <cfontana@suse.de>
Message-Id: <20210122204441.2145197-7-philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Avoid generating module_block.h and block-gen.c if we are
not going to use them.
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20210122204441.2145197-6-philmd@redhat.com>
[Extend to nearby files and directories. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20210122204441.2145197-4-philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20210122204441.2145197-3-philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Some new users get confused with 'TCG' and 'TCI', and enable TCI
support expecting to enable TCG.
Emit a warning when native TCG backend is available on the
host architecture, mentioning this is a suboptimal configuration.
Reviewed-by: Thomas Huth <thuth@redhat.com>
Suggested-by: Daniel Berrangé <berrange@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20210125144530.2837481-5-philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20210125144530.2837481-4-philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Users might want to enable all features, without realizing some
features have negative effect. Mention the TCI feature is slow
and experimental, hoping it will be selected knowingly.
Suggested-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20210125144530.2837481-3-philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Protection Keys for Supervisor-mode pages is a simple extension of
the PKU feature that QEMU already implements. For supervisor-mode
pages, protection key restrictions come from a new MSR. The MSR
has no XSAVE state associated to it.
PKS is only respected in long mode. However, in principle it is
possible to set the MSR even outside long mode, and in fact
even the XSAVE state for PKRU could be set outside long mode
using XRSTOR. So do not limit the migration subsections for
PKRU and PKRS to long mode.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This patch fixes a translation bug for a subset of x86 BMI instructions
such as the following:
c4 e2 f9 f7 c0 shlxq %rax, %rax, %rax
Currently, these incorrectly generate an undefined instruction exception
when SSE is disabled via CR4, while instructions like "shrxq" work fine.
The problem appears to be related to BMI instructions encoded using VEX
and with a mandatory prefix of "0x66" (data). Instructions with this
data prefix (such as shlxq) are currently rejected. Instructions with
other mandatory prefixes (such as shrxq) translate as expected.
This patch removes the incorrect check in "gen_sse" that causes the
exception to be generated. For the non-BMI cases, the check is
redundant: prefixes are already checked at line 3696.
Buglink: https://bugs.launchpad.net/qemu/+bug/1748296
Signed-off-by: David Greenaway <dgreenaway@google.com>
Message-Id: <20210114063958.1508050-1-dgreenaway@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Add 'initialized' field and use it to avoid touching event notifiers which are
either not initialized or if their initialization failed.
This is somewhat a hack, but it seems the less intrusive way to make
virtio code deal with event notifiers that failed initialization.
Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20201217150040.906961-4-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Count number of queues that we initialized and only deinitialize these that we
initialized successfully.
Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20201217150040.906961-3-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
No sense outputting the qemu-ga and qemu-ga-ref man pages when the guest
agent binary itself is disabled. This mirrors behaviour from before the
meson switch.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
Message-Id: <20210128145801.14384-1-s.reiter@proxmox.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Qiuhao Li <Qiuhao.Li@outlook.com>
Message-Id: <SYCPR01MB3502E9F6EB06DEDCD484F738FCBA9@SYCPR01MB3502.ausprd01.prod.outlook.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Newer AMD CPUs will add CPUID_0x8000000A_EDX[28] bit, which indicates
that SVM instructions (VMRUN/VMSAVE/VMLOAD) will trigger #VMEXIT before
CPU checking their EAX against reserved memory regions. This change will
allow the hypervisor to avoid intercepting #GP and emulating SVM
instructions. KVM turns on this CPUID bit for nested VMs. In order to
support it, let us populate this bit, along with other SVM feature bits,
in FEAT_SVM.
Signed-off-by: Wei Huang <wei.huang2@amd.com>
Message-Id: <20210126202456.589932-1-wei.huang2@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
If the link test failed, compilation proceeded with RBD disabled,
even if --enable-rbd was used on the configure command line.
Fix that.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Meson's "static" argument to cc.find_library is a tri-state. By default
Meson *prefers* a shared library, which basically means using -l to
look for it; instead, "static: false" *requires* a shared library. Of
course, "static: true" requires a static library, which is all good
for --enable-static builds.
For --disable-static, "static: false" is rarely desirable; it does not
match what the configure script used to do and the test is more complex
(and harder to debug if it fails, which was reported by Peter Lieven
for librbd).
Reported-by: Peter Lieven <pl@kamp.de>
Tested-by: Peter Lieven <pl@kamp.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Add documentation for '-machine memory-backend' CLI option and
how to use it.
And document that x-use-canonical-path-for-ramblock-id,
is considered to be stable to make sure it won't go away by accident.
x- was intended for unstable/iternal properties, and not supposed to
be stable option. However it's too late to rename (drop x-)
it as it would mean that users will have to mantain both
x-use-canonical-path-for-ramblock-id (for QEMU 5.0-5.2) versions
and prefix-less for later versions.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20210121161504.1007247-1-imammedo@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
32-bit targets by definition do not support long mode; therefore, the
bit must be masked in the features supported by the accelerator.
As a side effect, this avoids setting up the 0x80000008 CPUID leaf
for
qemu-system-i386 -cpu host
which since commit 5a140b255d ("x86/cpu: Use max host physical address
if -cpu max option is applied") would have printed this error:
qemu-system-i386: phys-bits should be between 32 and 36 (but is 48)
Reported-by: Nathan Chancellor <natechancellor@gmail.com>
Tested-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
virtio-9p devices are often used to expose a virtual-filesystem to the
guest. There have been some bugs reported in this device, such as
CVE-2018-19364, and CVE-2021-20181. We should fuzz this device
This patch adds two virtio-9p configurations:
* One with the widely used -fsdev local driver. This driver leaks some
state in the form of files/directories created in the shared dir.
* One with the synth driver. While it is not used in the real world, this
driver won't leak leak state between fuzz inputs.
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Message-Id: <20210117230924.449676-4-alxndr@bu.edu>
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Message-Id: <20210117230924.449676-3-alxndr@bu.edu>
For some device configurations, it is useful to configure some
resources, and adjust QEMU arguments at runtime, prior to fuzzing. This
patch adds an "argfunc" to generic the generic_fuzz_config. When
specified, it is responsible for configuring the resources and returning
a string containing the corresponding QEMU arguments. This can be useful
for targets that rely on e.g.:
* a temporary qcow2 image
* a temporary directory
* an unused TCP port used to bind the VNC server
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20210117230924.449676-2-alxndr@bu.edu>
This is useful for building reproducers. Instead checking the code or
the QEMU_FUZZ_ARGS, the arguments are at the top of the crash log.
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20210117201014.271610-3-alxndr@bu.edu>
Disks work differently depending on the x86 machine type (SATA vs PATA).
Additionally, we should fuzz the atapi code paths, which might contain
vulnerabilities such as CVE-2020-29443. This patch adds hard-disk and
cdrom generic-fuzzer configs for both the pc (PATA) and q35 (SATA)
machine types.
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Acked-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Message-Id: <20210120152211.109782-1-alxndr@bu.edu>
We passed an is_write flag to the fuzz_dma_read_cb function to
differentiate between the mapped DMA regions that need to be populated
with fuzzed data, and those that don't. We simply passed through the
address_space_map is_write parameter. The goal was to cut down on
unnecessarily populating mapped DMA regions, when they are not read
from.
Unfortunately, nothing precludes code from reading from regions mapped
with is_write=true. For example, see:
https://lists.gnu.org/archive/html/qemu-devel/2021-01/msg04729.html
This patch removes the is_write parameter to fuzz_dma_read_cb. As a
result, we will fill all mapped DMA regions with fuzzed data, ignoring
the specified transfer direction.
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Message-Id: <20210120060255.558535-1-alxndr@bu.edu>
In the modules that we are checking so far, we can be stricter about the
difference between Optional[T] and T types. Enable that check.
Enabling it now will assist review on further typing and cleanup work.
Signed-off-by: John Snow <jsnow@redhat.com>
Message-Id: <20210201193747.2169670-17-jsnow@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
For everything typed so far, type this parameter as
Optional[QAPISourceInfo].
In the most generic case, QAPISchemaEntity's info field may be None to
represent types that come from built-in definitions. Although some
Entity types may not currently have any built-in definitions, it is not
easily possible to constrain the type except on an ad-hoc basis using
assertions.
It's easier and simpler, then, to just say it's always an Optional type.
Signed-off-by: John Snow <jsnow@redhat.com>
Message-Id: <20210201193747.2169670-16-jsnow@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
The previous commit removed the only user of QAPIGen(None). Tighten
the type hint.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: John Snow <jsnow@redhat.com>
Message-Id: <20210201193747.2169670-15-jsnow@redhat.com>
QAPISchemaGenCommandVisitor.visit_command() needs to generate the
marshalling function into the current module, and also generate its
registration into the ./init system module. The latter is done
somewhat awkwardly: .__init__() creates a QAPIGenCCode that will not
be written out, each .visit_command() adds its registration to it, and
.visit_end() copies its contents into the ./init module it creates.
Instead provide the means to temporarily switch to another module.
Create the ./init module in .visit_begin(), and generate its initial
part. Add registrations to it in .visit_command(). Finish it in
.visit_end().
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: John Snow <jsnow@redhat.com>
Message-Id: <20210201193747.2169670-14-jsnow@redhat.com>
Many places assume they can access these fields without checking them
first to ensure they are defined. Eliminating the _genc and _genh fields
and replacing them with functional properties that check for correct
state can ease the typing overhead by eliminating the Optional[T] return
type.
Signed-off-by: John Snow <jsnow@redhat.com>
Message-Id: <20210201193747.2169670-12-jsnow@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Use a constant to make it obvious we're referring to a very specific thing.
Signed-off-by: John Snow <jsnow@redhat.com>
Message-Id: <20210201193747.2169670-11-jsnow@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
With callers to _add_system_module now explicitly using the './' prefix
to indicate a system module, there is no longer any reason to have
separate interfaces for adding system vs user modules; use a unified
interface that differentiates based on the name.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: John Snow <jsnow@redhat.com>
Message-Id: <20210201193747.2169670-10-jsnow@redhat.com>