Greg Kurz [Tue, 9 Nov 2021 18:35:23 +0000 (19:35 +0100)]
accel/tcg: Register a force_rcu notifier
A TCG vCPU doing a busy loop systematicaly hangs the QEMU monitor
if the user passes 'device_add' without argument. This is because
drain_cpu_all() which is called from qmp_device_add() cannot return
if readers don't exit read-side critical sections. That is typically
what busy-looping TCG vCPUs do:
int cpu_exec(CPUState *cpu)
{
[...]
rcu_read_lock();
[...]
while (!cpu_handle_exception(cpu, &ret)) {
// Busy loop keeps vCPU here
}
[...]
rcu_read_unlock();
return ret;
}
For MTTCG, have all vCPU threads register a force_rcu notifier that will
kick them out of the loop using async_run_on_cpu(). The notifier is called
with the rcu_registry_lock mutex held, using async_run_on_cpu() ensures
there are no deadlocks.
For RR, a single thread runs all vCPUs. Just register a single notifier
that kicks the current vCPU to the next one.
For MTTCG: Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
For RR: Suggested-by: Richard Henderson <richard.henderson@linaro.org> Fixes: 7bed89958bfb ("device_core: use drain_call_rcu in in qmp_device_add")
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/650 Signed-off-by: Greg Kurz <groug@kaod.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20211109183523.47726-3-groug@kaod.org> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Greg Kurz [Tue, 9 Nov 2021 18:35:22 +0000 (19:35 +0100)]
rcu: Introduce force_rcu notifier
The drain_rcu_call() function can be blocked as long as an RCU reader
stays in a read-side critical section. This is typically what happens
when a TCG vCPU is executing a busy loop. It can deadlock the QEMU
monitor as reported in https://gitlab.com/qemu-project/qemu/-/issues/650 .
This can be avoided by allowing drain_rcu_call() to enforce an RCU grace
period. Since each reader might need to do specific actions to end a
read-side critical section, do it with notifiers.
Prepare ground for this by adding a notifier list to the RCU reader
struct and use it in wait_for_readers() if drain_rcu_call() is in
progress. An API is added for readers to register their notifiers.
This is largely based on a draft from Paolo Bonzini.
Suggested-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kurz <groug@kaod.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20211109183523.47726-2-groug@kaod.org> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Richard Henderson [Mon, 8 Nov 2021 17:50:09 +0000 (18:50 +0100)]
Merge remote-tracking branch 'remotes/philmd/tags/avocado-20211108' into staging
Integration testing patches
- Rename tests/acceptance/ -> tests/avocado/
- Rename avocado_qemu.Test -> avocado_qemu.QemuSystemTest
- Introduce QemuUserTest class
- Add the first linux-user test, covering the bFLT loader
# gpg: Signature made Mon 08 Nov 2021 05:16:46 PM CET
# gpg: using RSA key FAABE75E12917221DCFD6BB2E3E32C2CDEADC0DE
# gpg: Good signature from "Philippe Mathieu-Daudé (F4BUG) <f4bug@amsat.org>" [full]
* remotes/philmd/tags/avocado-20211108:
tests/avocado: Remove p7zip binary availability check
tests/avocado: Rename avocado_qemu.Test -> QemuSystemTest
tests/avocado: Add bFLT loader linux-user test
tests/avocado: Share useful helpers from virtiofs_submounts test
tests/avocado: Introduce QemuUserTest base class
tests/avocado: Make pick_default_qemu_bin() more generic
tests/avocado: Extract QemuBaseTest from Test
tests/acceptance: rename tests acceptance to tests avocado
tests/acceptance: introduce new check-avocado target
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
The single use of the 7z binary has been removed in commit a30e114f3
("tests/acceptance: remove Armbian 19.11.3 test for orangepi-pc"),
we don't need to check for this binary availability anymore.
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com> Reviewed-by: Joaquin de Andres <me@xcancerberox.com.ar> Reviewed-by: Willian Rampazzo <willianr@redhat.com>
Message-Id: <20211106091059.465109-1-philmd@redhat.com>
To run user-mode emulation tests, we introduced the
avocado_qemu.QemuUserTest which inherits from avocado_qemu.QemuBaseTest.
System-mode emulation tests are based on the avocado_qemu.Test class,
which also inherits avocado_qemu.QemuBaseTest. To avoid confusion,
rename it as avocado_qemu.QemuSystemTest.
Suggested-by: Wainer dos Santos Moschetta <wainersm@redhat.com> Reviewed-by: Willian Rampazzo <willianr@redhat.com> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20211105143416.148332-7-f4bug@amsat.org>
Philippe Mathieu-Daudé [Sun, 14 Feb 2021 17:59:10 +0000 (18:59 +0100)]
tests/avocado: Introduce QemuUserTest base class
Similarly to the 'System' Test base class with methods for testing
system emulation, the QemuUserTest class contains methods useful to
test user-mode emulation.
Reviewed-by: Wainer dos Santos Moschetta <wainersm@redhat.com> Reviewed-by: Willian Rampazzo <willianr@redhat.com> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20211105143416.148332-4-f4bug@amsat.org>
Philippe Mathieu-Daudé [Sun, 14 Feb 2021 17:59:08 +0000 (18:59 +0100)]
tests/avocado: Extract QemuBaseTest from Test
The Avocado Test::fetch_asset() is handy to download artifacts
before running tests. The current class is named Test but only
tests system emulation. As we want to test user emulation,
refactor the common code as QemuBaseTest.
Reviewed-by: Willian Rampazzo <willianr@redhat.com> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20211105143416.148332-2-f4bug@amsat.org>
Willian Rampazzo [Fri, 5 Nov 2021 15:53:54 +0000 (12:53 -0300)]
tests/acceptance: rename tests acceptance to tests avocado
In the discussion about renaming the `tests/acceptance` [1], the
conclusion was that the folders inside `tests` are related to the
framework running the tests and not directly related to the type of
the tests.
This changes the folder to `tests/avocado` and adjusts the MAKEFILE, the
CI related files and the documentation.
Willian Rampazzo [Fri, 5 Nov 2021 15:53:53 +0000 (12:53 -0300)]
tests/acceptance: introduce new check-avocado target
This introduces a new `make` target, `check-avocado`, and adds a
deprecation message about the `check-acceptance` target. This is
a preparation for renaming the `tests/acceptance` folder to
`tests/avocado`.
The plan is to remove the call to the `check-avocado` target one
or two months after the release and leave the warning to force
people to move to the new `check-avocado` target.
Later, the `check-acceptance` target can be removed. The intent
is to avoid a direct impact during the current soft freeze.
Suggested-by: Philippe Mathieu-Daudé <philmd@redhat.com> Signed-off-by: Willian Rampazzo <willianr@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Tested-by: Philippe Mathieu-Daudé <philmd@redhat.com> Acked-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20211105155354.154864-2-willianr@redhat.com> Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Richard Henderson [Mon, 8 Nov 2021 15:23:32 +0000 (16:23 +0100)]
Merge remote-tracking branch 'remotes/bonzini/tags/for-upstream' into staging
* Fix off-by-one in MODE SELECT commands
* extend --extra-*flags behavior to meson-based tests
* allow using snappy in static builds
* i386 TCG fixes
* fix build failure when libgbm is not available
# gpg: Signature made Mon 08 Nov 2021 12:20:24 PM CET
# gpg: using RSA key F13338574B662389866C7682BFFBD25F78C7AE83
# gpg: issuer "pbonzini@redhat.com"
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>" [full]
# gpg: aka "Paolo Bonzini <pbonzini@redhat.com>" [full]
* remotes/bonzini/tags/for-upstream:
ui/gtk-egl: Fix build failure when libgbm is not available
configure: ignore preexisting QEMU_*FLAGS envvars
configure: propagate --extra-cflags and --extra-ldflags to meson compile tests
configure: preserve CFLAGS, CXXFLAGS and LDFLAGS in config.status
configure: simplify calls to meson_quote
docs: adjust for demise of scripts/create_config
meson: perform snappy test with the C++ compiler if used
hw/scsi/scsi-disk: MODE_PAGE_ALLS not allowed in MODE SELECT commands
target-i386: mmu: fix handling of noncanonical virtual addresses
target-i386: mmu: use pg_mode instead of HF_LMA_MASK
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Richard Henderson [Mon, 8 Nov 2021 13:56:10 +0000 (14:56 +0100)]
Merge remote-tracking branch 'remotes/marcandre.lureau/tags/sphinx-pull-request' into staging
Some Sphinx improvements
PR for 2 series:
https://patchew.org/QEMU/20211015105344.152591-1-marcandre.lureau@redhat.com/
https://patchew.org/QEMU/20211004215238.1523082-1-jsnow@redhat.com/
# gpg: Signature made Mon 08 Nov 2021 10:01:03 AM CET
# gpg: using RSA key 87A9BD933F87C606D276F62DDAE8E10975969CE5
# gpg: issuer "marcandre.lureau@redhat.com"
# gpg: Good signature from "Marc-André Lureau <marcandre.lureau@redhat.com>" [full]
# gpg: aka "Marc-André Lureau <marcandre.lureau@gmail.com>" [full]
* remotes/marcandre.lureau/tags/sphinx-pull-request:
docs/sphinx: change default role to "any"
docs: (further) remove non-reference uses of single backticks
docs: remove non-reference uses of single backticks
docs/sphinx: add 's' keyboard binding to focus search
docs/sphinx: set navigation_with_keys=True
meson: drop sphinx_template_files
meson: drop sphinx_extn_depends
tests/qapi-schema/meson: add depfile to sphinx doc
docs/sphinx: add templates files to generated depfile
docs/sphinx: add static files to generated depfile
docs/sphinx: add loaded modules to generated depfile
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Philippe Mathieu-Daudé [Mon, 8 Nov 2021 08:31:29 +0000 (09:31 +0100)]
ui/gtk-egl: Fix build failure when libgbm is not available
Since commit 4872a023a59 ("ui/gtk-egl: guest fb texture needs
to be regenerated when reinitializing egl") we get on Ubuntu
18.04.4 LTS and Debian Buster (oldstable):
$ ../configure --enable-virglrenderer
[...]
ui/gtk-egl.c: In function 'gd_egl_refresh':
ui/gtk-egl.c:159:13: error: implicit declaration of function 'egl_dmabuf_release_texture' [-Werror=implicit-function-declaration]
159 | egl_dmabuf_release_texture(vc->gfx.guest_fb.dmabuf);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~
ui/gtk-egl.c:159:13: error: this function declaration is not a prototype [-Werror,-Wstrict-prototypes]
ui/gtk-egl.c:159:13: error: nested extern declaration of 'egl_dmabuf_release_texture' [-Werror=nested-externs]
Fix by restricting the egl_dmabuf_release_texture() call to the
availability of the generic buffer management library (libgbm).
Fixes: 4872a023a593e6519b272a Cc: Dongwon Kim <dongwon.kim@intel.com> Cc: Vivek Kasireddy <vivek.kasireddy@intel.com> Reported-by: Laurent Vivier <laurent@vivier.eu> Reviewed-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Laurent Vivier <laurent@vivier.eu> Tested-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20211108083129.1262040-1-f4bug@amsat.org> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Paolo Bonzini [Mon, 8 Nov 2021 07:58:23 +0000 (08:58 +0100)]
configure: ignore preexisting QEMU_*FLAGS envvars
User flags should be passed via CFLAGS/CXXFLAGS/LDFLAGS,
or --extra-cflags/extra-cxxflags/--extra-ldflags on the
command line.
QEMU_CFLAGS, QEMU_CXXFLAGS and QEMU_LDFLAGS are reserved
for flags detected by configure, so do not add to them
and clear them at the beginning of the script.
Reviewed-by: Thomas Huth <thuth@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Paolo Bonzini [Fri, 5 Nov 2021 09:09:26 +0000 (10:09 +0100)]
configure: propagate --extra-cflags and --extra-ldflags to meson compile tests
Meson (intentionally) does not add QEMU_CFLAGS to cc.compiles/cc.links
tests, as they are supposed to be independent of the specific sets of
compilation flags used to build the programs. However, the user can
still use CFLAGS or the toolchain file's LANG_args/LANG_link_args option
to specify -I or -L options that apply to cc.compiles/cc.links as well.
This is also the intended use of configure's --extra-cflags,
--extra-cxxflags and --extra-ldflags options. For example, if
one has netmap's header in a nonstandard directory, up to commit 837b84b1c078bf3e909 it used to work fine to do:
Paolo Bonzini [Fri, 5 Nov 2021 09:08:43 +0000 (10:08 +0100)]
configure: preserve CFLAGS, CXXFLAGS and LDFLAGS in config.status
CFLAGS, CXXFLAGS and LDFLAGS influence the tests (for example if they include
-L or -I options), so they should be kept from the invocation of configure
to the subsequent reinvocations via config.status.
Reviewed-by: Thomas Huth <thuth@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Paolo Bonzini [Fri, 5 Nov 2021 09:07:37 +0000 (10:07 +0100)]
configure: simplify calls to meson_quote
meson_quote assumes a non-empty argument list, and incorrectly returns a
one-entry array if passed nothing. Move the check for an empty argument
list from the invocations to the function itself.
Reviewed-by: Thomas Huth <thuth@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Paolo Bonzini [Fri, 5 Nov 2021 12:45:40 +0000 (13:45 +0100)]
docs: adjust for demise of scripts/create_config
The config-host.h, $TARGET_NAME-config-target.h,
$TARGET_NAME-config-devices.h files are now generated by
configure_file() rather than scripts/create_config. Adjust
he relevant paragraph in docs/devel/build-system.rst, and take
the occasion to fix a preexisting confusion of *.h vs *.mak.
Reported-by: Markus Armbruster <armbru@redhat.com> Reviewed-by: Markus Armbruster <armbru@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Paolo Bonzini [Thu, 4 Nov 2021 13:35:14 +0000 (14:35 +0100)]
meson: perform snappy test with the C++ compiler if used
Snappy is written in C++ and as such needs to link against libstdc++. When
linking statically, this means that the compile test cannot succeed unless
performed with a C++ compiler. Do so if link_language is set to C++; if it
is C, the test will usually fail and snappy will be disabled.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
John Snow [Mon, 4 Oct 2021 21:52:36 +0000 (17:52 -0400)]
docs: remove non-reference uses of single backticks
The single backtick markup in ReST is the "default role". Currently,
Sphinx's default role is called "content". Sphinx suggests you can use
the "Any" role instead to turn any single-backtick enclosed item into a
cross-reference.
This is useful for things like autodoc for Python docstrings, where it's
often nicer to reference other types with `foo` instead of the more
laborious :py:meth:`foo`. It's also useful in multi-domain cases to
easily reference definitions from other Sphinx domains, such as
referencing C code definitions from outside of kerneldoc comments.
Before we do that, though, we'll need to turn all existing usages of the
"content" role to inline verbatim markup wherever it does not correctly
resolve into a cross-refernece by using double backticks instead.
Signed-off-by: John Snow <jsnow@redhat.com> Reviewed-by: Eduardo Habkost <ehabkost@redhat.com> Reviewed-by: Alexander Bulekov <alxndr@bu.edu>
Message-Id: <20211004215238.1523082-2-jsnow@redhat.com>
Marc-André Lureau [Fri, 8 Oct 2021 21:06:38 +0000 (01:06 +0400)]
docs/sphinx: set navigation_with_keys=True
Allow navigating to the previous/next page using the keyboard's left and
right arrows. I wish this would be the default, and that the themes
would provide more key navigation, but that doesn't seem on the roadmap.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: John Snow <jsnow@redhat.com> Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Paolo Bonzini [Thu, 4 Nov 2021 13:47:46 +0000 (14:47 +0100)]
target-i386: mmu: fix handling of noncanonical virtual addresses
mmu_translate is supposed to return an error code for page faults; it is
not able to handle other exceptions. The #GP case for noncanonical
virtual addresses is not handled correctly, and incorrectly raised as
a page fault with error code 1. Since it cannot happen for nested
page tables, move it directly to handle_mmu_fault, even before the
invocation of mmu_translate.
Fixes: #676 Fixes: 661ff4879e ("target/i386: extract mmu_translate", 2021-05-11) Cc: qemu-stable@nongnu.org Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Paolo Bonzini [Thu, 4 Nov 2021 13:45:20 +0000 (14:45 +0100)]
target-i386: mmu: use pg_mode instead of HF_LMA_MASK
Correctly look up the paging mode of the hypervisor when it is using 64-bit
mode but the guest is not.
Fixes: 68746930ae ("target/i386: use mmu_translate for NPT walk", 2021-05-11) Cc: qemu-stable@nongnu.org Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Hyman Huang(黄勇) [Fri, 5 Nov 2021 13:01:16 +0000 (21:01 +0800)]
docs: fix qemu incorrect tag
The patchset merged in 71864eadd9a ("migration/dirtyrate:
introduce struct and adjust DirtyRateStat") was targeting
QEMU 6.1 but got merged later, so correct the tag for 6.2.
Signed-off-by: Hyman Huang(黄勇) <huangy81@chinatelecom.cn> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Reviewed-by: Juan Quintela <quintela@redhat.com> Signed-off-by: Juan Quintela <quintela@redhat.com>
Richard Henderson [Fri, 5 Nov 2021 15:42:06 +0000 (11:42 -0400)]
Merge remote-tracking branch 'remotes/kraxel/tags/egl-20211105-pull-request' into staging
gtk: a collection of egl fixes.
# gpg: Signature made Fri 05 Nov 2021 07:30:21 AM EDT
# gpg: using RSA key A0328CFFB93A17A79901FE7D4CB6D8EED3E87138
# gpg: Good signature from "Gerd Hoffmann (work) <kraxel@redhat.com>" [full]
# gpg: aka "Gerd Hoffmann <gerd@kraxel.org>" [full]
# gpg: aka "Gerd Hoffmann (private) <kraxel@gmail.com>" [full]
* remotes/kraxel/tags/egl-20211105-pull-request:
ui/gtk-egl: blitting partial guest fb to the proper scanout surface
ui/gtk: gd_draw_event returns FALSE when no cairo surface is bound
ui/gtk-egl: guest fb texture needs to be regenerated when reinitializing egl
ui/gtk-egl: make sure the right context is set as the current
ui/gtk-egl: un-tab and re-tab should destroy egl surface and context
virtio-gpu: splitting one extended mode guest fb into n-scanouts
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Richard Henderson [Fri, 5 Nov 2021 13:59:19 +0000 (09:59 -0400)]
Merge remote-tracking branch 'remotes/jasonwang/tags/net-pull-request' into staging
# gpg: Signature made Thu 04 Nov 2021 11:42:00 PM EDT
# gpg: using RSA key EF04965B398D6211
# gpg: Good signature from "Jason Wang (Jason Wang on RedHat) <jasowang@redhat.com>" [marginal]
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg: It is not certain that the signature belongs to the owner.
# Primary key fingerprint: 215D 46F4 8246 689E C77F 3562 EF04 965B 398D 6211
Dongwon Kim [Thu, 4 Nov 2021 06:51:53 +0000 (23:51 -0700)]
ui/gtk-egl: blitting partial guest fb to the proper scanout surface
eb_fb_blit should be able to blit partial image of guest display (blob res)
in case multiple displays are configured for the guest and they are set as extended-
desktop mode.
v2: egl_fb includes dmabuf info then make egl_fb_blit position and size
parameters programmed in dmabuf structure (previously position/size
parameters were given to egl_fb_blit separately)
(Vivek Kasireddy)
changed the commit message as there is no interface change to egl_fb_blit
Dongwon Kim [Thu, 4 Nov 2021 06:51:50 +0000 (23:51 -0700)]
ui/gtk-egl: guest fb texture needs to be regenerated when reinitializing egl
If guest fb is backed by dmabuf (blob-resource), the texture bound to the
old context needs to be recreated in case the egl is re-initialized (e.g.
new window for vc is created in case of detaching/reattaching of the tab)
v2: call egl_dmabuf_release_texutre instead of putting 0 to dmabuf->texture
(Vivek Kasireddy)
Dongwon Kim [Thu, 4 Nov 2021 06:51:48 +0000 (23:51 -0700)]
ui/gtk-egl: un-tab and re-tab should destroy egl surface and context
An old esurface should be destroyed and set to be NULL when doing
un-tab and re-tab so that a new esurface an context can be created
for the window widget that those will be bound to.
v2: enabling opengl specific routines only when CONFIG_OPENGL is set
Dongwon Kim [Thu, 4 Nov 2021 06:51:52 +0000 (23:51 -0700)]
virtio-gpu: splitting one extended mode guest fb into n-scanouts
When guest is running Linux/X11 with extended multiple displays mode enabled,
the guest shares one scanout resource each time containing whole surface
rather than sharing individual display output separately. This extended frame
is properly splited and rendered on the corresponding scanout surfaces but
not in case of blob-resource (zero copy).
This code change lets the qemu split this one large surface data into multiple
in case of blob-resource as well so that each sub frame then can be blitted
properly to each scanout.
v2: resizing qemu console in virtio_gpu_update_dmabuf to scanout's width and
height
v3: updating stub function of virtio_gpu_update_dmabuf to match the type
Eduardo Habkost [Wed, 3 Nov 2021 11:38:31 +0000 (12:38 +0100)]
Fix virtio-net-pci* "vectors" compat
hw_compat_5_2 has an issue: it affects only "virtio-net-pci"
but not "virtio-net-pci-transitional" and
"virtio-net-pci-non-transitional". The solution is to use the
"virtio-net-pci-base" type in compat_props.
Jon Maloy [Thu, 21 Oct 2021 16:10:47 +0000 (12:10 -0400)]
e1000: fix tx re-entrancy problem
The fact that the MMIO handler is not re-entrant causes an infinite
loop under certain conditions:
Guest write to TDT -> Loopback -> RX (DMA to TDT) -> TX
We now eliminate the effect of this problem locally in e1000, by adding
a boolean in struct E1000State indicating when the TX side is busy. This
will cause any entering new call to return early instead of interfering
with the ongoing work, and eliminates any risk of looping.
This is intended to address CVE-2021-20257.
Signed-off-by: Jon Maloy <jmaloy@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
Richard Henderson [Thu, 4 Nov 2021 12:33:46 +0000 (08:33 -0400)]
Merge remote-tracking branch 'remotes/stsquad/tags/pull-for-6.2-041121-2' into staging
Testing, gdbstub and plugin updates for 6.2
- add microblaze and nios2 compiler docker images
- fix test cross compiler detection for some targets
- don't try and link ebf to user targets
- add L2 tracking to cache plugin
- exit cleanly on C-a x
- clean up debug output in check-tcg
- switch to thread on break in gdbstub
- update openbsd VM to 7.0
# gpg: Signature made Thu 04 Nov 2021 08:14:35 AM EDT
# gpg: using RSA key 6685AE99E75167BCAFC8DF35FBD0DB095A9E2A44
# gpg: Good signature from "Alex Bennée (Master Work Key) <alex.bennee@linaro.org>" [full]
* remotes/stsquad/tags/pull-for-6.2-041121-2:
tests/vm/openbsd: Update to release 7.0
tests/tcg: remove debug polluting make output
gdbstub: Switch to the thread receiving a signal
tests/tcg: remove duplicate EXTRA_RUNS
plugins: try and make plugin_insn_append more ergonomic
tests/plugins: extend the insn plugin to track opcode sizes
chardev: don't exit() straight away on C-a x
docs/tcg-plugins: add L2 arguments to cache docs
plugins/cache: make L2 emulation optional through args
plugins/cache: split command line arguments into name and value
plugins/cache: implement unified L2 cache emulation
plugins/cache: freed heap-allocated mutexes
ebpf: really include it only in system emulators
tests/tcg: enable debian-nios2-cross for test building
tests/docker: split PARTIAL into PARTIAL and VIRTUAL images
tests/tcg: Fix some targets default cross compiler path
tests/tcg: Enable container_cross_cc for microblaze
tests/docker: Add debian-microblaze-cross image
tests/docker: Add debian-nios2-cross image
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Richard Henderson [Thu, 4 Nov 2021 10:34:36 +0000 (06:34 -0400)]
Merge remote-tracking branch 'remotes/quic/tags/pull-hex-20211103' into staging
This series adds support for the Hexagon Vector eXtensions (HVX)
These instructions are documented here
https://developer.qualcomm.com/downloads/qualcomm-hexagon-v66-hvx-programmer-s-reference-manual
Hexagon HVX is a wide vector engine with 128 byte vectors.
See patch 01 Hexagon HVX README for more information.
*** Changes in v2 ***
Remove HVX tests from makefile to avoid need for toolchain upgrade
# gpg: Signature made Wed 03 Nov 2021 05:14:44 PM EDT
# gpg: using RSA key 7B0244FB12DE4422
# gpg: Good signature from "Taylor Simpson (Rock on) <tsimpson@quicinc.com>" [marginal]
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg: It is not certain that the signature belongs to the owner.
# Primary key fingerprint: 3635 C788 CE62 B91F D4C5 9AB4 7B02 44FB 12DE 4422
Richard Henderson [Mon, 18 Oct 2021 20:53:13 +0000 (13:53 -0700)]
tests/vm/openbsd: Update to release 7.0
There are two minor changes required in the script for the
network configuration of the newer release.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Tested-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20211018205313.3526915-1-richard.henderson@linaro.org> Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Alex Bennée [Tue, 26 Oct 2021 17:39:14 +0000 (18:39 +0100)]
tests/tcg: remove debug polluting make output
Fixes: 5343a837cd ("tests/tcg: move some multiarch files and make conditional") Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20211026173914.79377-1-alex.bennee@linaro.org>
Pavel Labath [Tue, 26 Oct 2021 10:22:34 +0000 (11:22 +0100)]
gdbstub: Switch to the thread receiving a signal
Respond with Txxthread:yyyy; instead of a plain Sxx to indicate which
thread received the signal. Otherwise, the debugger will associate it
with the main one. Also automatically select this thread, as that is
what gdb expects.
Signed-off-by: Pavel Labath <pavel@labath.sk>
Message-Id: <20211019174953.36560-1-pavel@labath.sk> Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20211026102234.3961636-29-alex.bennee@linaro.org>
Alex Bennée [Tue, 26 Oct 2021 10:22:33 +0000 (11:22 +0100)]
tests/tcg: remove duplicate EXTRA_RUNS
We set it bellow outside the #if leg.
Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20211026102234.3961636-28-alex.bennee@linaro.org>
Alex Bennée [Tue, 26 Oct 2021 10:22:26 +0000 (11:22 +0100)]
plugins: try and make plugin_insn_append more ergonomic
Currently we make the assumption that the guest frontend loads all
op code bytes sequentially. This mostly holds up for regular fixed
encodings but some architectures like s390x like to re-read the
instruction which causes weirdness to occur. Rather than changing the
frontends make the plugin API a little more ergonomic and able to
handle the re-read case.
Stuff will still get strange if we read ahead of the opcode but so far
no front ends have done that and this patch asserts the case so we can
catch it early if they do.
Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Suggested-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20211026102234.3961636-21-alex.bennee@linaro.org>
Alex Bennée [Tue, 26 Oct 2021 10:22:25 +0000 (11:22 +0100)]
tests/plugins: extend the insn plugin to track opcode sizes
This is mostly a convenience feature for identifying frontends that do
multiple repeat loads so I can test changes to the instruction
tracking interface.
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20211026102234.3961636-20-alex.bennee@linaro.org>
Alex Bennée [Tue, 26 Oct 2021 10:22:24 +0000 (11:22 +0100)]
chardev: don't exit() straight away on C-a x
While there are a number of uses in the code-base of the exit(0)
pattern it gets in the way of clean exit which can do all of it's
house-keeping. In particular it was reported that you can crash
plugins this way because TCG can still be running on other threads
when the atexit callback is called.
Use qmp_quit() instead which takes care of some housekeeping before
triggering the shutdown.
Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reported-by: Lukas Jünger <lukas.junger@greensocs.com> Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20211026102234.3961636-19-alex.bennee@linaro.org>
Mahmoud Mandour [Tue, 26 Oct 2021 10:22:23 +0000 (11:22 +0100)]
docs/tcg-plugins: add L2 arguments to cache docs
cache plugin now allows optional L2 per-core cache emulation that can be
configured through plugin arguments, this commit adds this functionality
to the docs.
While I'm at it, I editted the bullet point for cache plugin to say:
contrib/plugins/cache.c
instead of
contrib/plugins/cache
to match other plugins.
Mahmoud Mandour [Tue, 26 Oct 2021 10:22:22 +0000 (11:22 +0100)]
plugins/cache: make L2 emulation optional through args
By default L2 is not enabled and is enabled by either using the
newly-introduced "l2" boolean argument, or by setting any of the L2
cache parameters using args. On specifying "l2=on", the default cache
configuration is used.
This adds an implementation of a simple L2 configuration, in which a
unified L2 cache (stores both blocks of instructions and data) is
maintained for each core separately, with no inter-core interaction
taken in account. The L2 cache is used as a backup for L1 and is only
accessed if the wanted block does not exist in L1.
In terms of multi-threaded user-space emulation, the same approximation
of L1 is done, a static number of caches is maintained, and each and
every memory access initiated by a thread will have to go through one of
the available caches.
An atomic increment is used to maintain the number of L2 misses per
instruction.
Paolo Bonzini [Tue, 26 Oct 2021 10:22:18 +0000 (11:22 +0100)]
ebpf: really include it only in system emulators
eBPF libraries are being included in user emulators, which is useless and
also breaks --static compilation if a shared library for libbpf is
present in the system.
Reported-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20211012162252.263933-1-pbonzini@redhat.com> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-bt: Warner Losh <imp@bsdimp.com>
Message-Id: <20211026102234.3961636-13-alex.bennee@linaro.org>
Alex Bennée [Tue, 26 Oct 2021 10:22:17 +0000 (11:22 +0100)]
tests/tcg: enable debian-nios2-cross for test building
Now we have a nios2 test image we can start using it to build tests.
However signal handling in nios2 is still broken so we disable the
signals and linux-test tests that trigger the bug.
Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20211026102234.3961636-12-alex.bennee@linaro.org>
Philippe Mathieu-Daudé [Tue, 26 Oct 2021 10:22:14 +0000 (11:22 +0100)]
tests/tcg: Fix some targets default cross compiler path
We do not want a shell command substitution, but a parameter
substitution (with assignment). Replace $() -> ${}, otherwise
the expanded command return an empty string and the $cross_cc
variable is not set.
Fixes: 634ef789f8e ("tests/tcg: add more default compilers to configure.sh") Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
[AJB: disable sh4 linux-test] Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20211023164329.328137-1-f4bug@amsat.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20211026102234.3961636-9-alex.bennee@linaro.org>
Richard Henderson [Thu, 4 Nov 2021 04:47:37 +0000 (00:47 -0400)]
Merge remote-tracking branch 'remotes/juanquintela/tags/migration-20211102-pull-request' into staging
Migration Pull request
Hi
This are the pending migration patches on the list:
- Provide an error message for migration_cancel by Laurent
- Don't dump colo cache when a guest core is requested by Lukas
- Initialise Compression_conters for new migration by Yuxiating
On top of that I added another missing initialization
- Colo optimizations and crash improvements by Rao.
Please, apply.
# gpg: Signature made Wed 03 Nov 2021 04:45:35 AM EDT
# gpg: using RSA key 1899FF8EDEBF58CCEE034B82F487EF185872D723
# gpg: Good signature from "Juan Quintela <quintela@redhat.com>" [full]
# gpg: aka "Juan Quintela <quintela@trasno.org>" [full]
* remotes/juanquintela/tags/migration-20211102-pull-request:
Optimized the function of fill_connection_key.
colo: Don't dump colo cache if dump-guest-core=off
Changed the last-mode to none of first start COLO
Removed the qemu_fclose() in colo_process_incoming_thread
colo: fixed 'Segmentation fault' when the simplex mode PVM poweroff
Fixed SVM hang when do failover before PVM crash
Fixed qemu crash when guest power off in COLO mode
Some minor optimizations for COLO
migration: Zero migration compression counters
migration: initialise compression_counters for a new migration
migration: provide an error message to migration_cancel()
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Taylor Simpson [Wed, 3 Nov 2021 20:39:09 +0000 (15:39 -0500)]
Hexagon HVX (tests/tcg/hexagon) hvx_misc test
Tests for
packet semantics
vector loads (aligned and unaligned)
vector stores (aligned and unaligned)
vector masked stores
vector new value store
maximum HVX temps in a packet
vector operations
Acked-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Taylor Simpson [Fri, 13 Aug 2021 19:39:10 +0000 (14:39 -0500)]
Hexagon HVX (target/hexagon) import semantics
Imported from the Hexagon architecture library
imported/allext.idef Top level file for all extensions
imported/mmvec/ext.idef HVX instruction definitions
Support functions added to target/hexagon/genptr.c
Acked-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Build the infrastructure to create overrides for HVX instructions.
We create a new empty file (gen_tcg_hvx.h) that will be populated
in subsequent patches.
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Taylor Simpson [Thu, 30 Sep 2021 16:07:00 +0000 (11:07 -0500)]
Hexagon HVX (target/hexagon) helper functions
Probe and commit vector stores (masked and scatter/gather)
Log vector register writes
Add the execution counters to the debug log
Histogram instructions
Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Imported from the Hexagon architecture library
imported/allext_macros.def Top level macro include for all extensions
imported/macros.def Scalar core macros (some HVX here)
imported/mmvec/macros.def HVX macro definitions
The macro definition files specify instruction attributes that are applied
to each instruction that reverences the macro.
Acked-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
projects / qemu-nvme.git / log