This fixes:
commit e28112d007
Author: Daniel P. Berrangé <berrange@redhat.com>
Date: Thu Jun 8 17:40:16 2023 +0100
gitlab: stable staging branches publish containers in a separate tag
Due to a copy+paste mistake, that commit included "QEMU_JOB_SKIPPED"
in the final rule that was meant to be a 'catch all' for staging
branches.
As a result stable branches are still splattering dockers from the
primary development branch.
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
Tested-by: Michael Tokarev <mjt@tls.msk.ru>
Message-ID: <20240906140958.84755-1-berrange@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
(cherry picked from commit 8d5ab746b1)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
On GICv2 and later, level triggered interrupts are pending when either
the interrupt line is asserted or the interrupt was made pending by a
GICD_ISPENDRn write. Making a level triggered interrupt pending by
software persists until either the interrupt is acknowledged or cleared
by writing GICD_ICPENDRn. As long as the interrupt line is asserted,
the interrupt is pending in any case.
This logic is transparently implemented in gic_test_pending() for
GICv1 and GICv2. The function combines the "pending" irq_state flag
(used for edge triggered interrupts and software requests) and the
line status (tracked in the "level" field). However, we also
incorrectly set the pending flag on a guest write to GICD_ISENABLERn
if the line of a level triggered interrupt was asserted. This keeps
the interrupt pending even if the line is de-asserted after some
time.
This incorrect logic is a leftover of the initial 11MPCore GIC
implementation. That handles things slightly differently to the
architected GICv1 and GICv2. The 11MPCore TRM does not give a lot of
detail on the corner cases of its GIC's behaviour, and historically
we have not wanted to investigate exactly what it does in reality, so
QEMU's GIC model takes the approach of "retain our existing behaviour
for 11MPCore, and implement the architectural standard for later GIC
revisions".
On that basis, commit 8d999995e4 in 2013 is where we added the
"level-triggered interrupt with the line asserted" handling to
gic_test_pending(), and we deliberately kept the old behaviour of
gic_test_pending() for REV_11MPCORE. That commit should have added
the "only if 11MPCore" condition to the setting of the pending bit on
writes to GICD_ISENABLERn, but forgot it.
Add the missing "if REV_11MPCORE" condition, so that our behaviour
on GICv1 and GICv2 matches the GIC architecture requirements.
Cc: qemu-stable@nongnu.org
Fixes: 8d999995e4 ("arm_gic: Fix GIC pending behavior")
Signed-off-by: Jan Klötzke <jan.kloetzke@kernkonzept.com>
Message-id: 20240911114826.3558302-1-jan.kloetzke@kernkonzept.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
[PMM: expanded comment a little and converted to coding-style form;
expanded commit message with the historical backstory]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
(cherry picked from commit 110684c9a6)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Currently, the guest may write to the device configuration space,
whereas the virtio sound device specification in chapter 5.14.4
clearly states that the fields in the device configuration space
are driver-read-only.
Remove the set_config function from the virtio_snd class.
This also prevents a heap buffer overflow. See QEMU issue #2296.
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2296
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20240901130112.8242-1-vr_qemu@t-online.de>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit 7fc6611cad)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
As debian-11 transitions to LTS we are starting to have problems
building the image. While we could update to a later Debian building a
32 bit QEMU without modern floating point is niche host amongst the
few remaining 32 bit hosts we regularly build for. For now we still
have armhf-debian-cross-container which is currently built from the
more recent debian-12.
Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240910173900.4154726-2-alex.bennee@linaro.org>
(cherry picked from commit d0068b746a)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: force-remove tests/docker/dockerfiles/debian-armel-cross.docker)
Both gnutls and gcrypt can be configured to exclude support for certain
algorithms via a runtime check against system crypto policies. Thus it
is not sufficient to have a compile time test for hash support in their
pbkdf implementations.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
(cherry picked from commit e6c09ea4f9)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
As reported by Peter, we might be leaking memory when removing the
highest RAMBlock (in the weird ram_addr_t space), and adding a new one.
We will fail to realize that we already allocated bitmaps for more
dirty memory blocks, and effectively discard the pointers to them.
Fix it by getting rid of last_ram_page() and by remembering the number
of dirty memory blocks that have been allocated already.
While at it, let's use "unsigned int" for the number of blocks, which
should be sufficient until we reach ~32 exabytes.
Looks like this leak was introduced as we switched from using a single
bitmap_zero_extend() to allocating multiple bitmaps:
bitmap_zero_extend() relies on g_renew() which should have taken care of
this.
Resolves: https://lkml.kernel.org/r/CAFEAcA-k7a+VObGAfCFNygQNfCKL=AfX6A4kScq=VSSK0peqPg@mail.gmail.com
Reported-by: Peter Maydell <peter.maydell@linaro.org>
Fixes: 5b82b703b6 ("memory: RCU ram_list.dirty_memory[] for safe RAM hotplug")
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Tested-by: Peter Maydell <peter.maydell@linaro.org>
Cc: qemu-stable@nongnu.org
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: David Hildenbrand <david@redhat.com>
Link: https://lore.kernel.org/r/20240828090743.128647-1-david@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
(cherry picked from commit b84f06c2be)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: context fix due to lack of
v9.0.0-rc4-49-g15f7a80c49cb "RAMBlock: Add support of KVM private guest memfd")
In vmstate_tlbemb a cut-and-paste error meant we gave
this vmstate subsection the same "cpu/tlb6xx" name as
the vmstate_tlb6xx subsection. This breaks migration load
for any CPU using the TLB_EMB CPU type, because when we
see the "tlb6xx" name in the incoming data we try to
interpret it as a vmstate_tlb6xx subsection, which it
isn't the right format for:
$ qemu-system-ppc -drive
if=none,format=qcow2,file=/home/petmay01/test-images/virt/dummy.qcow2
-monitor stdio -M bamboo
QEMU 9.0.92 monitor - type 'help' for more information
(qemu) savevm foo
(qemu) loadvm foo
Missing section footer for cpu
Error: Error -22 while loading VM state
Correct the incorrect vmstate section name. Since migration
for these CPU types was completely broken before, we don't
need to care that this is a migration compatibility break.
This affects the PPC 405, 440, 460 and e200 CPU families.
Cc: qemu-stable@nongnu.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2522
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Arman Nabiev <nabiev.arman13@gmail.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
(cherry picked from commit 203beb6f04)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
While adding hppa64 support, the psw_v variable got extended from 32 to 64
bits. So, when packaging the PSW-V bit from the psw_v variable for interrupt
processing, check bit 31 instead the 63th (sign) bit.
This fixes a hard to find Linux kernel boot issue where the loss of the PSW-V
bit due to an ITLB interruption in the middle of a series of ds/addc
instructions (from the divU milicode library) generated the wrong division
result and thus triggered a Linux kernel crash.
Link: https://lore.kernel.org/lkml/718b8afe-222f-4b3a-96d3-93af0e4ceff1@roeck-us.net/
Reported-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Helge Deller <deller@gmx.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Fixes: 931adff314 ("target/hppa: Update cpu_hppa_get/put_psw for hppa64")
Cc: qemu-stable@nongnu.org # v8.2+
(cherry picked from commit ead5078cf1)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: context fixup in target/hppa/helper.c due to lack of
v9.0.0-688-gebc9401a4067 "target/hppa: Split PSW X and B into their own field")
Commit 9b6083465f ("virtio-snd: check for invalid param shift
operands") tries to prevent invalid parameters specified by the
guest. However, the code is not correct.
Change the code so that the parameters format and rate, which are
a bit numbers, are compared with the bit size of the data type.
Fixes: 9b6083465f ("virtio-snd: check for invalid param shift operands")
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20240802071805.7123-1-vr_qemu@t-online.de>
Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit 7d14471a12)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
This reverts commit 1f881ea4a4.
That commit causes reverse_debugging.py test failures, and does
not seem to solve the root cause of the problem x86-64 still
hangs in record/replay tests.
The problem with short-cutting the iowait that was taken during
record phase is that related events will not get consumed at the
same points (e.g., reading the clock).
A hang with zero icount always seems to be a symptom of an earlier
problem that has caused the recording to become out of synch with
the execution and consumption of events by replay.
Acked-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Message-Id: <20240813050638.446172-6-npiggin@gmail.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240813202329.1237572-14-alex.bennee@linaro.org>
(cherry picked from commit 94962ff00d)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Fixes: https://gitlab.com/qemu-project/qemu/-/issues/2524
In multifd_recv_setup() we allocate (among other things)
* a MultiFDRecvData struct to multifd_recv_state::data
* a MultiFDRecvData struct to each multfd_recv_state->params[i].data
(Then during execution we might swap these pointers around.)
But in multifd_recv_cleanup() we free multifd_recv_state->data
in multifd_recv_cleanup_state() but we don't ever free the
multifd_recv_state->params[i].data. This results in a memory
leak reported by LeakSanitizer:
(cd build/asan && \
ASAN_OPTIONS="fast_unwind_on_malloc=0:strip_path_prefix=/mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../" \
QTEST_QEMU_BINARY=./qemu-system-x86_64 \
./tests/qtest/migration-test --tap -k -p /x86_64/migration/multifd/file/mapped-ram )
[...]
Direct leak of 72 byte(s) in 3 object(s) allocated from:
#0 0x561cc0afcfd8 in __interceptor_calloc (/mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/qemu-system-x86_64+0x218efd8) (BuildId: be72e086d4e47b172b0a72779972213fd9916466)
#1 0x7f89d37acc50 in g_malloc0 debian/build/deb/../../../glib/gmem.c:161:13
#2 0x561cc1e9c83c in multifd_recv_setup migration/multifd.c:1606:19
#3 0x561cc1e68618 in migration_ioc_process_incoming migration/migration.c:972:9
#4 0x561cc1e3ac59 in migration_channel_process_incoming migration/channel.c:45:9
#5 0x561cc1e4fa0b in file_accept_incoming_migration migration/file.c:132:5
#6 0x561cc30f2c0c in qio_channel_fd_source_dispatch io/channel-watch.c:84:12
#7 0x7f89d37a3c43 in g_main_dispatch debian/build/deb/../../../glib/gmain.c:3419:28
#8 0x7f89d37a3c43 in g_main_context_dispatch debian/build/deb/../../../glib/gmain.c:4137:7
#9 0x561cc3b21659 in glib_pollfds_poll util/main-loop.c:287:9
#10 0x561cc3b1ff93 in os_host_main_loop_wait util/main-loop.c:310:5
#11 0x561cc3b1fb5c in main_loop_wait util/main-loop.c:589:11
#12 0x561cc1da2917 in qemu_main_loop system/runstate.c:801:9
#13 0x561cc3796c1c in qemu_default_main system/main.c:37:14
#14 0x561cc3796c67 in main system/main.c:48:12
#15 0x7f89d163bd8f in __libc_start_call_main csu/../sysdeps/nptl/libc_start_call_main.h:58:16
#16 0x7f89d163be3f in __libc_start_main csu/../csu/libc-start.c:392:3
#17 0x561cc0a79fa4 in _start (/mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/qemu-system-x86_64+0x210bfa4) (BuildId: be72e086d4e47b172b0a72779972213fd9916466)
Direct leak of 24 byte(s) in 1 object(s) allocated from:
#0 0x561cc0afcfd8 in __interceptor_calloc (/mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/qemu-system-x86_64+0x218efd8) (BuildId: be72e086d4e47b172b0a72779972213fd9916466)
#1 0x7f89d37acc50 in g_malloc0 debian/build/deb/../../../glib/gmem.c:161:13
#2 0x561cc1e9bed9 in multifd_recv_setup migration/multifd.c:1588:32
#3 0x561cc1e68618 in migration_ioc_process_incoming migration/migration.c:972:9
#4 0x561cc1e3ac59 in migration_channel_process_incoming migration/channel.c:45:9
#5 0x561cc1e4fa0b in file_accept_incoming_migration migration/file.c:132:5
#6 0x561cc30f2c0c in qio_channel_fd_source_dispatch io/channel-watch.c:84:12
#7 0x7f89d37a3c43 in g_main_dispatch debian/build/deb/../../../glib/gmain.c:3419:28
#8 0x7f89d37a3c43 in g_main_context_dispatch debian/build/deb/../../../glib/gmain.c:4137:7
#9 0x561cc3b21659 in glib_pollfds_poll util/main-loop.c:287:9
#10 0x561cc3b1ff93 in os_host_main_loop_wait util/main-loop.c:310:5
#11 0x561cc3b1fb5c in main_loop_wait util/main-loop.c:589:11
#12 0x561cc1da2917 in qemu_main_loop system/runstate.c:801:9
#13 0x561cc3796c1c in qemu_default_main system/main.c:37:14
#14 0x561cc3796c67 in main system/main.c:48:12
#15 0x7f89d163bd8f in __libc_start_call_main csu/../sysdeps/nptl/libc_start_call_main.h:58:16
#16 0x7f89d163be3f in __libc_start_main csu/../csu/libc-start.c:392:3
#17 0x561cc0a79fa4 in _start (/mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/qemu-system-x86_64+0x210bfa4) (BuildId: be72e086d4e47b172b0a72779972213fd9916466)
SUMMARY: AddressSanitizer: 96 byte(s) leaked in 4 allocation(s).
Free the params[i].data too.
Cc: qemu-stable@nongnu.org
Fixes: d117ed0699 ("migration/multifd: Allow receiving pages without packets")
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
(cherry picked from commit 4c107870e8)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The crash was reported in MAC OS and NixOS, here is the link for this bug
https://gitlab.com/qemu-project/qemu/-/issues/2334https://gitlab.com/qemu-project/qemu/-/issues/2321
In this bug, they are using the virtio_input device. The guest notifier was
not supported for this device, The function virtio_pci_set_guest_notifiers()
was not called, and the vector_irqfd was not initialized.
So the fix is adding the check for vector_irqfd in virtio_pci_get_notifier()
The function virtio_pci_get_notifier() can be used in various devices.
It could also be called when VIRTIO_CONFIG_S_DRIVER_OK is not set. In this situation,
the vector_irqfd being NULL is acceptable. We can allow the device continue to boot
If the vector_irqfd still hasn't been initialized after VIRTIO_CONFIG_S_DRIVER_OK
is set, it means that the function set_guest_notifiers was not called before the
driver started. This indicates that the device is not using the notifier.
At this point, we will let the check fail.
This fix is verified in vyatta,MacOS,NixOS,fedora system.
The bt tree for this bug is:
Thread 6 "CPU 0/KVM" received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7c817be006c0 (LWP 1269146)]
kvm_virtio_pci_vq_vector_use () at ../qemu-9.0.0/hw/virtio/virtio-pci.c:817
817 if (irqfd->users == 0) {
(gdb) thread apply all bt
...
Thread 6 (Thread 0x7c817be006c0 (LWP 1269146) "CPU 0/KVM"):
0 kvm_virtio_pci_vq_vector_use () at ../qemu-9.0.0/hw/virtio/virtio-pci.c:817
1 kvm_virtio_pci_vector_use_one () at ../qemu-9.0.0/hw/virtio/virtio-pci.c:893
2 0x00005983657045e2 in memory_region_write_accessor () at ../qemu-9.0.0/system/memory.c:497
3 0x0000598365704ba6 in access_with_adjusted_size () at ../qemu-9.0.0/system/memory.c:573
4 0x0000598365705059 in memory_region_dispatch_write () at ../qemu-9.0.0/system/memory.c:1528
5 0x00005983659b8e1f in flatview_write_continue_step.isra.0 () at ../qemu-9.0.0/system/physmem.c:2713
6 0x000059836570ba7d in flatview_write_continue () at ../qemu-9.0.0/system/physmem.c:2743
7 flatview_write () at ../qemu-9.0.0/system/physmem.c:2774
8 0x000059836570bb76 in address_space_write () at ../qemu-9.0.0/system/physmem.c:2894
9 0x0000598365763afe in address_space_rw () at ../qemu-9.0.0/system/physmem.c:2904
10 kvm_cpu_exec () at ../qemu-9.0.0/accel/kvm/kvm-all.c:2917
11 0x000059836576656e in kvm_vcpu_thread_fn () at ../qemu-9.0.0/accel/kvm/kvm-accel-ops.c:50
12 0x0000598365926ca8 in qemu_thread_start () at ../qemu-9.0.0/util/qemu-thread-posix.c:541
13 0x00007c8185bcd1cf in ??? () at /usr/lib/libc.so.6
14 0x00007c8185c4e504 in clone () at /usr/lib/libc.so.6
Fixes: 2ce6cff94d ("virtio-pci: fix use of a released vector")
Cc: qemu-stable@nongnu.org
Signed-off-by: Cindy Lu <lulu@redhat.com>
Message-Id: <20240806093715.65105-1-lulu@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit a8e63ff289)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Yutaro Shimizu from the Cyber Defense Institute discovered a bug in the
NVMe emulation that leaks contents of an uninitialized heap buffer if
subsystem and FDP emulation are enabled.
Cc: qemu-stable@nongnu.org
Reported-by: Yutaro Shimizu <shimizu@cyberdefense.jp>
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
(cherry picked from commit 6a22121c4f)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When the creds->username property is set we allocate memory
for it in qcrypto_tls_creds_psk_prop_set_username(), but
we never free this when the QCryptoTLSCredsPSK is destroyed.
Free the memory in finalize.
This fixes a LeakSanitizer complaint in migration-test:
$ (cd build/asan; ASAN_OPTIONS="fast_unwind_on_malloc=0" QTEST_QEMU_BINARY=./qemu-system-x86_64 ./tests/qtest/migration-test --tap -k -p /x86_64/migration/precopy/unix/tls/psk)
=================================================================
==3867512==ERROR: LeakSanitizer: detected memory leaks
Direct leak of 5 byte(s) in 1 object(s) allocated from:
#0 0x5624e5c99dee in malloc (/mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/qemu-system-x86_64+0x218edee) (BuildId: a9e623fa1009a9435c0142c037cd7b8c1ad04ce3)
#1 0x7fb199ae9738 in g_malloc debian/build/deb/../../../glib/gmem.c:128:13
#2 0x7fb199afe583 in g_strdup debian/build/deb/../../../glib/gstrfuncs.c:361:17
#3 0x5624e82ea919 in qcrypto_tls_creds_psk_prop_set_username /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../crypto/tlscredspsk.c:255:23
#4 0x5624e812c6b5 in property_set_str /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../qom/object.c:2277:5
#5 0x5624e8125ce5 in object_property_set /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../qom/object.c:1463:5
#6 0x5624e8136e7c in object_set_properties_from_qdict /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../qom/object_interfaces.c:55:14
#7 0x5624e81372d2 in user_creatable_add_type /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../qom/object_interfaces.c:112:5
#8 0x5624e8137964 in user_creatable_add_qapi /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../qom/object_interfaces.c:157:11
#9 0x5624e891ba3c in qmp_object_add /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../qom/qom-qmp-cmds.c:227:5
#10 0x5624e8af9118 in qmp_marshal_object_add /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/qapi/qapi-commands-qom.c:337:5
#11 0x5624e8bd1d49 in do_qmp_dispatch_bh /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../qapi/qmp-dispatch.c:128:5
#12 0x5624e8cb2531 in aio_bh_call /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../util/async.c:171:5
#13 0x5624e8cb340c in aio_bh_poll /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../util/async.c:218:13
#14 0x5624e8c0be98 in aio_dispatch /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../util/aio-posix.c:423:5
#15 0x5624e8cba3ce in aio_ctx_dispatch /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../util/async.c:360:5
#16 0x7fb199ae0d3a in g_main_dispatch debian/build/deb/../../../glib/gmain.c:3419:28
#17 0x7fb199ae0d3a in g_main_context_dispatch debian/build/deb/../../../glib/gmain.c:4137:7
#18 0x5624e8cbe1d9 in glib_pollfds_poll /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../util/main-loop.c:287:9
#19 0x5624e8cbcb13 in os_host_main_loop_wait /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../util/main-loop.c:310:5
#20 0x5624e8cbc6dc in main_loop_wait /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../util/main-loop.c:589:11
#21 0x5624e6f3f917 in qemu_main_loop /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../system/runstate.c:801:9
#22 0x5624e893379c in qemu_default_main /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../system/main.c:37:14
#23 0x5624e89337e7 in main /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/../../system/main.c:48:12
#24 0x7fb197972d8f in __libc_start_call_main csu/../sysdeps/nptl/libc_start_call_main.h:58:16
#25 0x7fb197972e3f in __libc_start_main csu/../csu/libc-start.c:392:3
#26 0x5624e5c16fa4 in _start (/mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/qemu-system-x86_64+0x210bfa4) (BuildId: a9e623fa1009a9435c0142c037cd7b8c1ad04ce3)
SUMMARY: AddressSanitizer: 5 byte(s) leaked in 1 allocation(s).
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240819145021.38524-1-peter.maydell@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit 87e012f29f)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When cross compiling QEMU configured with --static, I've been getting
configure errors like the following:
Build-time dependency glib-2.0 found: NO
../target/hexagon/meson.build:303:15: ERROR: Dependency lookup for glib-2.0 with method 'pkgconfig' failed: Could not generate libs for glib-2.0:
Package libpcre2-8 was not found in the pkg-config search path.
Perhaps you should add the directory containing `libpcre2-8.pc'
to the PKG_CONFIG_PATH environment variable
Package 'libpcre2-8', required by 'glib-2.0', not found
This happens because --static sets the prefer_static Meson option, but
my build machine doesn't have a static libpcre2. I don't think it
makes sense to insist that native dependencies are static, just
because I want the non-native QEMU binaries to be static.
Signed-off-by: Alyssa Ross <hi@alyssa.is>
Link: https://lore.kernel.org/r/20240805104921.4035256-1-hi@alyssa.is
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit fe68cc0923)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Our current usage of MMU indexes when EL3 is AArch32 is confused.
Architecturally, when EL3 is AArch32, all Secure code runs under the
Secure PL1&0 translation regime:
* code at EL3, which might be Mon, or SVC, or any of the
other privileged modes (PL1)
* code at EL0 (Secure PL0)
This is different from when EL3 is AArch64, in which case EL3 is its
own translation regime, and EL1 and EL0 (whether AArch32 or AArch64)
have their own regime.
We claimed to be mapping Secure PL1 to our ARMMMUIdx_EL3, but didn't
do anything special about Secure PL0, which meant it used the same
ARMMMUIdx_EL10_0 that NonSecure PL0 does. This resulted in a bug
where arm_sctlr() incorrectly picked the NonSecure SCTLR as the
controlling register when in Secure PL0, which meant we were
spuriously generating alignment faults because we were looking at the
wrong SCTLR control bits.
The use of ARMMMUIdx_EL3 for Secure PL1 also resulted in the bug that
we wouldn't honour the PAN bit for Secure PL1, because there's no
equivalent _PAN mmu index for it.
We could fix this in one of two ways:
* The most straightforward is to add new MMU indexes EL30_0,
EL30_3, EL30_3_PAN to correspond to "Secure PL1&0 at PL0",
"Secure PL1&0 at PL1", and "Secure PL1&0 at PL1 with PAN".
This matches how we use indexes for the AArch64 regimes, and
preserves propirties like being able to determine the privilege
level from an MMU index without any other information. However
it would add two MMU indexes (we can share one with ARMMMUIdx_EL3),
and we are already using 14 of the 16 the core TLB code permits.
* The more complicated approach is the one we take here. We use
the same MMU indexes (E10_0, E10_1, E10_1_PAN) for Secure PL1&0
than we do for NonSecure PL1&0. This saves on MMU indexes, but
means we need to check in some places whether we're in the
Secure PL1&0 regime or not before we interpret an MMU index.
The changes in this commit were created by auditing all the places
where we use specific ARMMMUIdx_ values, and checking whether they
needed to be changed to handle the new index value usage.
Note for potential stable backports: taking also the previous
(comment-change-only) commit might make the backport easier.
Cc: qemu-stable@nongnu.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2326
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Tested-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240809160430.1144805-3-peter.maydell@linaro.org
(cherry picked from commit 4c2c047469)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
We have a long comment describing the Arm architectural translation
regimes and how we map them to QEMU MMU indexes. This comment has
got a bit out of date:
* FEAT_SEL2 allows Secure EL2 and corresponding new regimes
* FEAT_RME introduces Realm state and its translation regimes
* We now model the Cortex-R52 so that is no longer a hypothetical
* We separated Secure Stage 2 and NonSecure Stage 2 MMU indexes
* We have an MMU index per physical address spacea
Add the missing pieces so that the list of architectural translation
regimes matches the Arm ARM, and the list and count of QEMU MMU
indexes in the comment matches the enum.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Tested-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240809160430.1144805-2-peter.maydell@linaro.org
(cherry picked from commit 150c24f34e)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: pick this one for stable-9.0 so the next commit applies cleanly)
Commit 3e7ef738 plugged the use-after-free of the global nbd_server
object, but overlooked a use-after-free of nbd_server->listener.
Although this race is harder to hit, notice that our shutdown path
first drops the reference count of nbd_server->listener, then triggers
actions that can result in a pending client reaching the
nbd_blockdev_client_closed() callback, which in turn calls
qio_net_listener_set_client_func on a potentially stale object.
If we know we don't want any more clients to connect, and have already
told the listener socket to shut down, then we should not be trying to
update the listener socket's associated function.
Reproducer:
> #!/usr/bin/python3
>
> import os
> from threading import Thread
>
> def start_stop():
> while 1:
> os.system('virsh qemu-monitor-command VM \'{"execute": "nbd-server-start",
+"arguments":{"addr":{"type":"unix","data":{"path":"/tmp/nbd-sock"}}}}\'')
> os.system('virsh qemu-monitor-command VM \'{"execute": "nbd-server-stop"}\'')
>
> def nbd_list():
> while 1:
> os.system('/path/to/build/qemu-nbd -L -k /tmp/nbd-sock')
>
> def test():
> sst = Thread(target=start_stop)
> sst.start()
> nlt = Thread(target=nbd_list)
> nlt.start()
>
> sst.join()
> nlt.join()
>
> test()
Fixes: CVE-2024-7409
Fixes: 3e7ef738c8 ("nbd/server: CVE-2024-7409: Close stray clients at server-stop")
CC: qemu-stable@nongnu.org
Reported-by: Andrey Drobyshev <andrey.drobyshev@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20240822143617.800419-2-eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
(cherry picked from commit 3874f5f73c)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
A malicious client can attempt to connect to an NBD server, and then
intentionally delay progress in the handshake, including if it does
not know the TLS secrets. Although the previous two patches reduce
this behavior by capping the default max-connections parameter and
killing slow clients, they did not eliminate the possibility of a
client waiting to close the socket until after the QMP nbd-server-stop
command is executed, at which point qemu would SEGV when trying to
dereference the NULL nbd_server global which is no longer present.
This amounts to a denial of service attack. Worse, if another NBD
server is started before the malicious client disconnects, I cannot
rule out additional adverse effects when the old client interferes
with the connection count of the new server (although the most likely
is a crash due to an assertion failure when checking
nbd_server->connections > 0).
For environments without this patch, the CVE can be mitigated by
ensuring (such as via a firewall) that only trusted clients can
connect to an NBD server. Note that using frameworks like libvirt
that ensure that TLS is used and that nbd-server-stop is not executed
while any trusted clients are still connected will only help if there
is also no possibility for an untrusted client to open a connection
but then stall on the NBD handshake.
Given the previous patches, it would be possible to guarantee that no
clients remain connected by having nbd-server-stop sleep for longer
than the default handshake deadline before finally freeing the global
nbd_server object, but that could make QMP non-responsive for a long
time. So intead, this patch fixes the problem by tracking all client
sockets opened while the server is running, and forcefully closing any
such sockets remaining without a completed handshake at the time of
nbd-server-stop, then waiting until the coroutines servicing those
sockets notice the state change. nbd-server-stop now has a second
AIO_WAIT_WHILE_UNLOCKED (the first is indirectly through the
blk_exp_close_all_type() that disconnects all clients that completed
handshakes), but forced socket shutdown is enough to progress the
coroutines and quickly tear down all clients before the server is
freed, thus finally fixing the CVE.
This patch relies heavily on the fact that nbd/server.c guarantees
that it only calls nbd_blockdev_client_closed() from the main loop
(see the assertion in nbd_client_put() and the hoops used in
nbd_client_put_nonzero() to achieve that); if we did not have that
guarantee, we would also need a mutex protecting our accesses of the
list of connections to survive re-entrancy from independent iothreads.
Although I did not actually try to test old builds, it looks like this
problem has existed since at least commit 862172f45c (v2.12.0, 2017) -
even back when that patch started using a QIONetListener to handle
listening on multiple sockets, nbd_server_free() was already unaware
that the nbd_blockdev_client_closed callback can be reached later by a
client thread that has not completed handshakes (and therefore the
client's socket never got added to the list closed in
nbd_export_close_all), despite that patch intentionally tearing down
the QIONetListener to prevent new clients.
Reported-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>
Fixes: CVE-2024-7409
CC: qemu-stable@nongnu.org
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20240807174943.771624-14-eblake@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
(cherry picked from commit 3e7ef738c8)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
A client that opens a socket but does not negotiate is merely hogging
qemu's resources (an open fd and a small amount of memory); and a
malicious client that can access the port where NBD is listening can
attempt a denial of service attack by intentionally opening and
abandoning lots of unfinished connections. The previous patch put a
default bound on the number of such ongoing connections, but once that
limit is hit, no more clients can connect (including legitimate ones).
The solution is to insist that clients complete handshake within a
reasonable time limit, defaulting to 10 seconds. A client that has
not successfully completed NBD_OPT_GO by then (including the case of
where the client didn't know TLS credentials to even reach the point
of NBD_OPT_GO) is wasting our time and does not deserve to stay
connected. Later patches will allow fine-tuning the limit away from
the default value (including disabling it for doing integration
testing of the handshake process itself).
Note that this patch in isolation actually makes it more likely to see
qemu SEGV after nbd-server-stop, as any client socket still connected
when the server shuts down will now be closed after 10 seconds rather
than at the client's whims. That will be addressed in the next patch.
For a demo of this patch in action:
$ qemu-nbd -f raw -r -t -e 10 file &
$ nbdsh --opt-mode -c '
H = list()
for i in range(20):
print(i)
H.insert(i, nbd.NBD())
H[i].set_opt_mode(True)
H[i].connect_uri("nbd://localhost")
'
$ kill $!
where later connections get to start progressing once earlier ones are
forcefully dropped for taking too long, rather than hanging.
Suggested-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20240807174943.771624-13-eblake@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
[eblake: rebase to changes earlier in series, reduce scope of timer]
Signed-off-by: Eric Blake <eblake@redhat.com>
(cherry picked from commit b9b72cb3ce)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Allowing an unlimited number of clients to any web service is a recipe
for a rudimentary denial of service attack: the client merely needs to
open lots of sockets without closing them, until qemu no longer has
any more fds available to allocate.
For qemu-nbd, we default to allowing only 1 connection unless more are
explicitly asked for (-e or --shared); this was historically picked as
a nice default (without an explicit -t, a non-persistent qemu-nbd goes
away after a client disconnects, without needing any additional
follow-up commands), and we are not going to change that interface now
(besides, someday we want to point people towards qemu-storage-daemon
instead of qemu-nbd).
But for qemu proper, and the newer qemu-storage-daemon, the QMP
nbd-server-start command has historically had a default of unlimited
number of connections, in part because unlike qemu-nbd it is
inherently persistent until nbd-server-stop. Allowing multiple client
sockets is particularly useful for clients that can take advantage of
MULTI_CONN (creating parallel sockets to increase throughput),
although known clients that do so (such as libnbd's nbdcopy) typically
use only 8 or 16 connections (the benefits of scaling diminish once
more sockets are competing for kernel attention). Picking a number
large enough for typical use cases, but not unlimited, makes it
slightly harder for a malicious client to perform a denial of service
merely by opening lots of connections withot progressing through the
handshake.
This change does not eliminate CVE-2024-7409 on its own, but reduces
the chance for fd exhaustion or unlimited memory usage as an attack
surface. On the other hand, by itself, it makes it more obvious that
with a finite limit, we have the problem of an unauthenticated client
holding 100 fds opened as a way to block out a legitimate client from
being able to connect; thus, later patches will further add timeouts
to reject clients that are not making progress.
This is an INTENTIONAL change in behavior, and will break any client
of nbd-server-start that was not passing an explicit max-connections
parameter, yet expects more than 100 simultaneous connections. We are
not aware of any such client (as stated above, most clients aware of
MULTI_CONN get by just fine on 8 or 16 connections, and probably cope
with later connections failing by relying on the earlier connections;
libvirt has not yet been passing max-connections, but generally
creates NBD servers with the intent for a single client for the sake
of live storage migration; meanwhile, the KubeSAN project anticipates
a large cluster sharing multiple clients [up to 8 per node, and up to
100 nodes in a cluster], but it currently uses qemu-nbd with an
explicit --shared=0 rather than qemu-storage-daemon with
nbd-server-start).
We considered using a deprecation period (declare that omitting
max-parameters is deprecated, and make it mandatory in 3 releases -
then we don't need to pick an arbitrary default); that has zero risk
of breaking any apps that accidentally depended on more than 100
connections, and where such breakage might not be noticed under unit
testing but only under the larger loads of production usage. But it
does not close the denial-of-service hole until far into the future,
and requires all apps to change to add the parameter even if 100 was
good enough. It also has a drawback that any app (like libvirt) that
is accidentally relying on an unlimited default should seriously
consider their own CVE now, at which point they are going to change to
pass explicit max-connections sooner than waiting for 3 qemu releases.
Finally, if our changed default breaks an app, that app can always
pass in an explicit max-parameters with a larger value.
It is also intentional that the HMP interface to nbd-server-start is
not changed to expose max-connections (any client needing to fine-tune
things should be using QMP).
Suggested-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20240807174943.771624-12-eblake@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
[ericb: Expand commit message to summarize Dan's argument for why we
break corner-case back-compat behavior without a deprecation period]
Signed-off-by: Eric Blake <eblake@redhat.com>
(cherry picked from commit c8a76dbd90)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Upcoming patches to fix a CVE need to track an opaque pointer passed
in by the owner of a client object, as well as request for a time
limit on how fast negotiation must complete. Prepare for that by
changing the signature of nbd_client_new() and adding an accessor to
get at the opaque pointer, although for now the two servers
(qemu-nbd.c and blockdev-nbd.c) do not change behavior even though
they pass in a new default timeout value.
Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20240807174943.771624-11-eblake@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
[eblake: s/LIMIT/MAX_SECS/ as suggested by Dan]
Signed-off-by: Eric Blake <eblake@redhat.com>
(cherry picked from commit fb1c2aaa98)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When reading with `read_cluster` we get the `mapping` with
`find_mapping_for_cluster` and then we call `open_file` for this
mapping.
The issue appear when its the same file, but a second cluster that is
not immediately after it, imagine clusters `500 -> 503`, this will give
us 2 mappings one has the range `500..501` and another `503..504`, both
point to the same file, but different offsets.
When we don't open the file since the path is the same, we won't assign
`s->current_mapping` and thus accessing way out of bound of the file.
From our example above, after `open_file` (that didn't open anything) we
will get the offset into the file with
`s->cluster_size*(cluster_num-s->current_mapping->begin)`, which will
give us `0x2000 * (504-500)`, which is out of bound for this mapping and
will produce some issues.
Signed-off-by: Amjad Alsharafi <amjadsharafi10@gmail.com>
Message-ID: <1f3ea115779abab62ba32c788073cdc99f9ad5dd.1721470238.git.amjadsharafi10@gmail.com>
[kwolf: Simplified the patch based on Amjad's analysis and input]
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit 5eed3db336)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
How this `abort` was intended to check for was:
- if the `mapping->first_mapping_index` is not the same as
`first_mapping_index`, which **should** happen only in one case,
when we are handling the first mapping, in that case
`mapping->first_mapping_index == -1`, in all other cases, the other
mappings after the first should have the condition `true`.
- From above, we know that this is the first mapping, so if the offset
is not `0`, then abort, since this is an invalid state.
The issue was that `first_mapping_index` is not set if we are
checking from the middle, the variable `first_mapping_index` is
only set if we passed through the check `cluster_was_modified` with the
first mapping, and in the same function call we checked the other
mappings.
One approach is to go into the loop even if `cluster_was_modified`
is not true so that we will be able to set `first_mapping_index` for the
first mapping, but since `first_mapping_index` is only used here,
another approach is to just check manually for the
`mapping->first_mapping_index != -1` since we know that this is the
value for the only entry where `offset == 0` (i.e. first mapping).
Signed-off-by: Amjad Alsharafi <amjadsharafi10@gmail.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-ID: <b0fbca3ee208c565885838f6a7deeaeb23f4f9c2.1721470238.git.amjadsharafi10@gmail.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit f60a6f7e17)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Before this commit, the behavior when calling `commit_one_file` for
example with `offset=0x2000` (second cluster), what will happen is that
we won't fetch the next cluster from the fat, and instead use the first
cluster for the read operation.
This is due to off-by-one error here, where `i=0x2000 !< offset=0x2000`,
thus not fetching the next cluster.
Signed-off-by: Amjad Alsharafi <amjadsharafi10@gmail.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Tested-by: Kevin Wolf <kwolf@redhat.com>
Message-ID: <b97c1e1f1bc2f776061ae914f95d799d124fcd73.1721470238.git.amjadsharafi10@gmail.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit b881cf00c9)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Analyzing qemu-produced core dumps of multi-threaded apps runs into:
(gdb) info threads
[...]
21 Thread 0x3ff83cc0740 (LWP 9295) warning: Couldn't find general-purpose registers in core file.
<unavailable> in ?? ()
The reason is that all pr_pid values are the same, because the same
TaskState is used for all CPUs when generating NT_PRSTATUS notes.
Fix by using TaskStates associated with individual CPUs.
Cc: qemu-stable@nongnu.org
Fixes: 243c470662 ("linux-user/elfload: Write corefile elf header in one block")
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20240801202340.21845-1-iii@linux.ibm.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
(cherry picked from commit 5b0c2742c8)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When a channel fails to create, the code currently just returns. This
is wrong for two reasons:
1) Channel n+1 will not get to initialize it's semaphores, leading to
an assert when terminate_threads tries to post to it:
qemu-system-x86_64: ../util/qemu-thread-posix.c:92:
qemu_mutex_lock_impl: Assertion `mutex->initialized' failed.
2) (theoretical) If channel n-1 already started creation it will
defeat the purpose of the channels_created logic which is in place
to avoid migrate_fd_cleanup() to run while channels are still being
created.
This cannot really happen today because the current failure cases
for multifd_new_send_channel_create() are all synchronous,
resulting from qio_channel_file_new_path() getting a bad
filename. This would hit all channels equally.
But I don't want to set a trap for future people, so have all
channels try to create (even if failing), and only fail after the
channels_created semaphore has been posted.
While here, remove the error_report_err call. There's one already at
migrate_fd_cleanup later on.
Cc: qemu-stable@nongnu.org
Reported-by: Jim Fehlig <jfehlig@suse.com>
Fixes: b7b03eb614 ("migration/multifd: Add outgoing QIOChannelFile support")
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
(cherry picked from commit 0bd5b9284f)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
While refactoring the NIC initialization code, I broke '-net nic,model=help'
which no longer outputs a list of available NIC models.
Fixes: 2cdeca04ad ("net: report list of available models according to platform")
Cc: qemu-stable@nongnu.org
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
Signed-off-by: Jason Wang <jasowang@redhat.com>
(cherry picked from commit 64f75f57f9)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Patch 06b1297017 ("virtio-net: fix network stall under load")
added double-check to test whether the available buffer size
can satisfy the request or not, in case the guest has added
some buffers to the avail ring simultaneously after the first
check. It will be lucky if the available buffer size becomes
okay after the double-check, then the host can send the packet
to the guest. If the buffer size still can't satisfy the request,
even if the guest has added some buffers, viritio-net would
stall at the host side forever.
The patch enables notification and checks whether the guest has
added some buffers since last check of available buffers when
the available buffers are insufficient. If no buffer is added,
return false, else recheck the available buffers in the loop.
If the available buffers are sufficient, disable notification
and return true.
Changes:
1. Change the return type of virtqueue_get_avail_bytes() from void
to int, it returns an opaque that represents the shadow_avail_idx
of the virtqueue on success, else -1 on error.
2. Add a new API: virtio_queue_enable_notification_and_check(),
it takes an opaque as input arg which is returned from
virtqueue_get_avail_bytes(). It enables notification firstly,
then checks whether the guest has added some buffers since
last check of available buffers or not by virtio_queue_poll(),
return ture if yes.
The patch also reverts patch "06b12970174".
The case below can reproduce the stall.
Guest 0
+--------+
| iperf |
---------------> | server |
Host | +--------+
+--------+ | ...
| iperf |----
| client |---- Guest n
+--------+ | +--------+
| | iperf |
---------------> | server |
+--------+
Boot many guests from qemu with virtio network:
qemu ... -netdev tap,id=net_x \
-device virtio-net-pci-non-transitional,\
iommu_platform=on,mac=xx:xx:xx:xx:xx:xx,netdev=net_x
Each guest acts as iperf server with commands below:
iperf3 -s -D -i 10 -p 8001
iperf3 -s -D -i 10 -p 8002
The host as iperf client:
iperf3 -c guest_IP -p 8001 -i 30 -w 256k -P 20 -t 40000
iperf3 -c guest_IP -p 8002 -i 30 -w 256k -P 20 -t 40000
After some time, the host loses connection to the guest,
the guest can send packet to the host, but can't receive
packet from the host.
It's more likely to happen if SWIOTLB is enabled in the guest,
allocating and freeing bounce buffer takes some CPU ticks,
copying from/to bounce buffer takes more CPU ticks, compared
with that there is no bounce buffer in the guest.
Once the rate of producing packets from the host approximates
the rate of receiveing packets in the guest, the guest would
loop in NAPI.
receive packets ---
| |
v |
free buf virtnet_poll
| |
v |
add buf to avail ring ---
|
| need kick the host?
| NAPI continues
v
receive packets ---
| |
v |
free buf virtnet_poll
| |
v |
add buf to avail ring ---
|
v
... ...
On the other hand, the host fetches free buf from avail
ring, if the buf in the avail ring is not enough, the
host notifies the guest the event by writing the avail
idx read from avail ring to the event idx of used ring,
then the host goes to sleep, waiting for the kick signal
from the guest.
Once the guest finds the host is waiting for kick singal
(in virtqueue_kick_prepare_split()), it kicks the host.
The host may stall forever at the sequences below:
Host Guest
------------ -----------
fetch buf, send packet receive packet ---
... ... |
fetch buf, send packet add buf |
... add buf virtnet_poll
buf not enough avail idx-> add buf |
read avail idx add buf |
add buf ---
receive packet ---
write event idx ... |
wait for kick add buf virtnet_poll
... |
---
no more packet, exit NAPI
In the first loop of NAPI above, indicated in the range of
virtnet_poll above, the host is sending packets while the
guest is receiving packets and adding buffers.
step 1: The buf is not enough, for example, a big packet
needs 5 buf, but the available buf count is 3.
The host read current avail idx.
step 2: The guest adds some buf, then checks whether the
host is waiting for kick signal, not at this time.
The used ring is not empty, the guest continues
the second loop of NAPI.
step 3: The host writes the avail idx read from avail
ring to used ring as event idx via
virtio_queue_set_notification(q->rx_vq, 1).
step 4: At the end of the second loop of NAPI, recheck
whether kick is needed, as the event idx in the
used ring written by the host is beyound the
range of kick condition, the guest will not
send kick signal to the host.
Fixes: 06b1297017 ("virtio-net: fix network stall under load")
Cc: qemu-stable@nongnu.org
Signed-off-by: Wencheng Yang <east.moutain.yang@gmail.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
(cherry picked from commit f937309fbd)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: context fixup in include/hw/virtio/virtio.h)
Ensure the queue index points to a valid queue when software RSS
enabled. The new calculation matches with the behavior of Linux's TAP
device with the RSS eBPF program.
Fixes: 4474e37a5b ("virtio-net: implement RX RSS processing")
Reported-by: Zhibin Hu <huzhibin5@huawei.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
(cherry picked from commit f1595ceb9a)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Fixes: CVE-2024-6505
The FMOPA (widening) SME instruction takes pairs of half-precision
floating point values, widens them to single-precision, does a
two-way dot product and accumulates the results into a
single-precision destination. We don't quite correctly handle the
FPCR bits FZ and FZ16 which control flushing of denormal inputs and
outputs. This is because at the moment we pass a single float_status
value to the helper function, which then uses that configuration for
all the fp operations it does. However, because the inputs to this
operation are float16 and the outputs are float32 we need to use the
fp_status_f16 for the float16 input widening but the normal fp_status
for everything else. Otherwise we will apply the flushing control
FPCR.FZ16 to the 32-bit output rather than the FPCR.FZ control, and
incorrectly flush a denormal output to zero when we should not (or
vice-versa).
(In commit 207d30b5fd we tried to fix the FZ handling but
didn't get it right, switching from "use FPCR.FZ for everything" to
"use FPCR.FZ16 for everything".)
(Mjt: it is commit 43929c818c in stable-9.0)
Pass the CPU env to the sme_fmopa_h helper instead of an fp_status
pointer, and have the helper pass an extra fp_status into the
f16_dotadd() function so that we can use the right status for the
right parts of this operation.
Cc: qemu-stable@nongnu.org
Fixes: 207d30b5fd ("target/arm: Use FPST_F16 for SME FMOPA (widening)")
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2373
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
(cherry picked from commit 55f9f4ee01)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
In newer versions of Sphinx the env.doc2path() API is going to change
to return a Path object rather than a str. This was originally visible
in Sphinx 8.0.0rc1, but has been rolled back for the final 8.0.0
release. However it will probably emit a deprecation warning and is
likely to change for good in 9.0:
https://github.com/sphinx-doc/sphinx/issues/12686
Our use in depfile.py assumes a str, and if it is passed a Path
it will fall over:
Handler <function write_depfile at 0x77a1775ff560> for event 'build-finished' threw an exception (exception: unsupported operand type(s) for +: 'PosixPath' and 'str')
Wrapping the env.doc2path() call in str() will coerce a Path object
to the str we expect, and have no effect in older Sphinx versions
that do return a str.
Cc: qemu-stable@nongnu.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2458
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240729120533.2486427-1-peter.maydell@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit 48e5b5f994)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When determining the current vector length, the SMCR_EL2.LEN and
SVCR_EL2.LEN settings should only be considered if EL2 is enabled
(compare the pseudocode CurrentSVL and CurrentNSVL which call
EL2Enabled()).
We were checking against ARM_FEATURE_EL2 rather than calling
arm_is_el2_enabled(), which meant that we would look at
SMCR_EL2/SVCR_EL2 when in Secure EL1 or Secure EL0 even if Secure EL2
was not enabled.
Use the correct check in sve_vqm1_for_el_sm().
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240722172957.1041231-5-peter.maydell@linaro.org
(cherry picked from commit f573ac059e)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The function tszimm_esz() returns a shift amount, or possibly -1 in
certain cases that correspond to unallocated encodings in the
instruction set. We catch these later in the trans_ functions
(generally with an "a-esz < 0" check), but before we do the
decodetree-generated code will also call tszimm_shr() or tszimm_sl(),
which will use the tszimm_esz() return value as a shift count without
checking that it is not negative, which is undefined behaviour.
Avoid the UB by checking the return value in tszimm_shr() and
tszimm_shl().
Cc: qemu-stable@nongnu.org
Resolves: Coverity CID 1547617, 1547694
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240722172957.1041231-4-peter.maydell@linaro.org
(cherry picked from commit 76916dfa89)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The UMOPA/UMOPS instructions are supposed to multiply unsigned 8 or
16 bit elements and accumulate the products into a 64-bit element.
In the Arm ARM pseudocode, this is done with the usual
infinite-precision signed arithmetic. However our implementation
doesn't quite get it right, because in the DEF_IMOP_64() macro we do:
sum += (NTYPE)(n >> 0) * (MTYPE)(m >> 0);
where NTYPE and MTYPE are uint16_t or int16_t. In the uint16_t case,
the C usual arithmetic conversions mean the values are converted to
"int" type and the multiply is done as a 32-bit multiply. This means
that if the inputs are, for example, 0xffff and 0xffff then the
result is 0xFFFE0001 as an int, which is then promoted to uint64_t
for the accumulation into sum; this promotion incorrectly sign
extends the multiply.
Avoid the incorrect sign extension by casting to int64_t before
the multiply, so we do the multiply as 64-bit signed arithmetic,
which is a type large enough that the multiply can never
overflow into the sign bit.
(The equivalent 8-bit operations in DEF_IMOP_32() are fine, because
the 8-bit multiplies can never overflow into the sign bit of a
32-bit integer.)
Cc: qemu-stable@nongnu.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2372
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240722172957.1041231-3-peter.maydell@linaro.org
(cherry picked from commit ea3f5a90f0)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
For an instruction which accesses a 128-bit element tile when
the SVL is also 128 (for example MOV z0.Q, p0/M, ZA0H.Q[w0,0]),
we will assert in get_tile_rowcol():
qemu-system-aarch64: ../../tcg/tcg-op.c:926: tcg_gen_deposit_z_i32: Assertion `len > 0' failed.
This happens because we calculate
len = ctz32(streaming_vec_reg_size(s)) - esz;$
but if the SVL and the element size are the same len is 0, and
the deposit operation asserts.
In this case the ZA storage contains exactly one 128 bit
element ZA tile, and the horizontal or vertical slice is just
that tile. This means that regardless of the index value in
the Ws register, we always access that tile. (In pseudocode terms,
we calculate (index + offset) MOD 1, which is 0.)
Special case the len == 0 case to avoid hitting the assertion
in tcg_gen_deposit_z_i32().
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240722172957.1041231-2-peter.maydell@linaro.org
(cherry picked from commit 56f1c0db92)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The documentation of the "Set palette" mailbox property at
https://github.com/raspberrypi/firmware/wiki/Mailbox-property-interface#set-palette
says it has the form:
Length: 24..1032
Value:
u32: offset: first palette index to set (0-255)
u32: length: number of palette entries to set (1-256)
u32...: RGBA palette values (offset to offset+length-1)
We get this wrong in a couple of ways:
* we aren't checking the offset and length are in range, so the guest
can make us spin for a long time by providing a large length
* the bounds check on our loop is wrong: we should iterate through
'length' palette entries, not 'length - offset' entries
Fix the loop to implement the bounds checks and get the loop
condition right. In the process, make the variables local to
this switch case, rather than function-global, so it's clearer
what type they are when reading the code.
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20240723131029.1159908-2-peter.maydell@linaro.org
(cherry picked from commit 0892fffc2a)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: context fix due to lack of
v9.0.0-1812-g5d5f1b60916a "hw/misc: Implement mailbox properties for customer OTP and device specific private keys"
also remove now-unused local `n' variable which gets removed in the next change in this file,
v9.0.0-2720-g32f1c201eedf "hw/misc/bcm2835_property: Avoid overflow in OTP access properties")
When a bare-metal application on the raspi3 board reads the
AUX_MU_STAT_REG MMIO register while the device's buffer is
at full receive FIFO capacity
(i.e. `s->read_count == BCM2835_AUX_RX_FIFO_LEN`) the
assertion `assert(s->read_count < BCM2835_AUX_RX_FIFO_LEN)`
fails.
Reported-by: Cryptjar <cryptjar@junk.studio>
Suggested-by: Cryptjar <cryptjar@junk.studio>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/459
Signed-off-by: Frederik van Hövell <frederik@fvhovell.nl>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
[PMM: commit message tweaks]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
(cherry picked from commit 546d574b11)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The unrealize functions of the various vhost-user devices are
calling the corresponding vhost_*_set_status() functions with a
status of 0 to shut down the device correctly.
Now these vhost_*_set_status() functions all follow this scheme:
bool should_start = virtio_device_should_start(vdev, status);
if (vhost_dev_is_started(&vvc->vhost_dev) == should_start) {
return;
}
if (should_start) {
/* ... do the initialization stuff ... */
} else {
/* ... do the cleanup stuff ... */
}
The problem here is virtio_device_should_start(vdev, 0) currently
always returns "true" since it internally only looks at vdev->started
instead of looking at the "status" parameter. Thus once the device
got started once, virtio_device_should_start() always returns true
and thus the vhost_*_set_status() functions return early, without
ever doing any clean-up when being called with status == 0. This
causes e.g. problems when trying to hot-plug and hot-unplug a vhost
user devices multiple times since the de-initialization step is
completely skipped during the unplug operation.
This bug has been introduced in commit 9f6bcfd99f ("hw/virtio: move
vm_running check to virtio_device_started") which replaced
should_start = status & VIRTIO_CONFIG_S_DRIVER_OK;
with
should_start = virtio_device_started(vdev, status);
which later got replaced by virtio_device_should_start(). This blocked
the possibility to set should_start to false in case the status flag
VIRTIO_CONFIG_S_DRIVER_OK was not set.
Fix it by adjusting the virtio_device_should_start() function to
only consider the status flag instead of vdev->started. Since this
function is only used in the various vhost_*_set_status() functions
for exactly the same purpose, it should be fine to fix it in this
central place there without any risk to change the behavior of other
code.
Fixes: 9f6bcfd99f ("hw/virtio: move vm_running check to virtio_device_started")
Buglink: https://issues.redhat.com/browse/RHEL-40708
Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20240618121958.88673-1-thuth@redhat.com>
Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit d72479b117)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
This reverts commit 2b316774f6.
After 038b421788 ("Revert "chardev: use a child source for qio input
source"") we've been observing the "iwp->src == NULL" assertion
triggering periodically during the initial capabilities querying by
libvirtd. One of possible backtraces:
Thread 1 (Thread 0x7f16cd4f0700 (LWP 43858)):
0 __GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:50
1 0x00007f16c6c21e65 in __GI_abort () at abort.c:79
2 0x00007f16c6c21d39 in __assert_fail_base at assert.c:92
3 0x00007f16c6c46e86 in __GI___assert_fail (assertion=assertion@entry=0x562e9bcdaadd "iwp->src == NULL", file=file@entry=0x562e9bcdaac8 "../chardev/char-io.c", line=line@entry=99, function=function@entry=0x562e9bcdab10 <__PRETTY_FUNCTION__.20549> "io_watch_poll_finalize") at assert.c:101
4 0x0000562e9ba20c2c in io_watch_poll_finalize (source=<optimized out>) at ../chardev/char-io.c:99
5 io_watch_poll_finalize (source=<optimized out>) at ../chardev/char-io.c:88
6 0x00007f16c904aae0 in g_source_unref_internal () from /lib64/libglib-2.0.so.0
7 0x00007f16c904baf9 in g_source_destroy_internal () from /lib64/libglib-2.0.so.0
8 0x0000562e9ba20db0 in io_remove_watch_poll (source=0x562e9d6720b0) at ../chardev/char-io.c:147
9 remove_fd_in_watch (chr=chr@entry=0x562e9d5f3800) at ../chardev/char-io.c:153
10 0x0000562e9ba23ffb in update_ioc_handlers (s=0x562e9d5f3800) at ../chardev/char-socket.c:592
11 0x0000562e9ba2072f in qemu_chr_fe_set_handlers_full at ../chardev/char-fe.c:279
12 0x0000562e9ba207a9 in qemu_chr_fe_set_handlers at ../chardev/char-fe.c:304
13 0x0000562e9ba2ca75 in monitor_qmp_setup_handlers_bh (opaque=0x562e9d4c2c60) at ../monitor/qmp.c:509
14 0x0000562e9bb6222e in aio_bh_poll (ctx=ctx@entry=0x562e9d4c2f20) at ../util/async.c:216
15 0x0000562e9bb4de0a in aio_poll (ctx=0x562e9d4c2f20, blocking=blocking@entry=true) at ../util/aio-posix.c:722
16 0x0000562e9b99dfaa in iothread_run (opaque=0x562e9d4c26f0) at ../iothread.c:63
17 0x0000562e9bb505a4 in qemu_thread_start (args=0x562e9d4c7ea0) at ../util/qemu-thread-posix.c:543
18 0x00007f16c70081ca in start_thread (arg=<optimized out>) at pthread_create.c:479
19 0x00007f16c6c398d3 in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95
io_remove_watch_poll(), which makes sure that iwp->src is NULL, calls
g_source_destroy() which finds that iwp->src is not NULL in the finalize
callback. This can only happen if another thread has managed to trigger
io_watch_poll_prepare() callback in the meantime.
Move iwp->src destruction back to the finalize callback to prevent the
described race, and also remove the stale comment. The deadlock glib bug
was fixed back in 2010 by b35820285668 ("gmain: move finalization of
GSource outside of context lock").
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Sergey Dyasli <sergey.dyasli@nutanix.com>
Link: https://lore.kernel.org/r/20240712092659.216206-1-sergey.dyasli@nutanix.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit e0bf95443e)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
aio_context_set_thread_pool_params() takes two int64_t arguments to
set the minimum and maximum number of threads in the pool. We do
some bounds checking on these, but we don't catch the case where the
inputs are negative. This means that later in the function when we
assign these inputs to the AioContext::thread_pool_min and
::thread_pool_max fields, which are of type int, the values might
overflow the smaller type.
A negative number of threads is meaningless, so make
aio_context_set_thread_pool_params() return an error if either min or
max are negative.
Resolves: Coverity CID 1547605
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20240723150927.1396456-1-peter.maydell@linaro.org
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
(cherry picked from commit 851495571d)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Once initialised, QOM objects can be realized and
unrealized multiple times before being finalized.
Resources allocated in REALIZE must be deallocated
in an equivalent UNREALIZE handler.
Free the CPU array in loongson_ipi_unrealize()
instead of loongson_ipi_finalize().
Cc: qemu-stable@nongnu.org
Fixes: 5e90b8db38 ("hw/loongarch: Set iocsr address space per-board rather than percpu")
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20240723111405.14208-3-philmd@linaro.org>
(cherry picked from commit 0c2086bc73)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: rename loongson back to longarch for 9.0 due to lack of
v9.0.0-582-gb4a12dfc2132 "hw/intc/loongarch_ipi: Rename as loongson_ipi")
If I use `-serial stdio` on Windows, after QEMU exits, the terminal
could not handle arrow keys and tab any more. Because stdio backend
on Windows sets console mode to virtual terminal input when starts,
but does not restore the old mode when finalize.
This small patch saves the old console mode and set it back.
Signed-off-by: Ziming Song <s.ziming@hotmail.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-ID: <ME3P282MB25488BE7C39BF0C35CD0DA5D8CA82@ME3P282MB2548.AUSP282.PROD.OUTLOOK.COM>
(cherry picked from commit 903cc9e117)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
sgx_epc_get_section assumes a PC platform is in use:
bool sgx_epc_get_section(int section_nr, uint64_t *addr, uint64_t *size)
{
PCMachineState *pcms = PC_MACHINE(qdev_get_machine());
However, sgx_epc_get_section is called by CPUID regardless of whether
SGX state has been initialized or which platform is in use. Check
whether the machine has the right QOM class and if not behave as if
there are no EPC sections.
Fixes: 1dec2e1f19 ("i386: Update SGX CPUID info according to hardware/KVM/user input", 2021-09-30)
Cc: qemu-stable@nongnu.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2142
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 13be929aff)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When setting the parameters of a PCM stream, we compute the bit flag
with the format and rate values as shift operand to check if they are
set in supported_formats and supported_rates.
If the guest provides a format/rate value which when shifting 1 results
in a value bigger than the number of bits in
supported_formats/supported_rates, we must report an error.
Previously, this ended up triggering the not reached assertions later
when converting to internal QEMU values.
Reported-by: Zheyu Ma <zheyuma97@gmail.com>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2416
Signed-off-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Message-Id: <virtio-snd-fuzz-2416-fix-v1-manos.pitsidianakis@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit 9b6083465f)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When reading input audio in the virtio-snd input callback,
virtio_snd_pcm_in_cb(), we do not check whether the iov can actually fit
the data buffer. This is because we use the buffer->size field as a
total-so-far accumulator instead of byte-size-left like in TX buffers.
This triggers an out of bounds write if the size of the virtio queue
element is equal to virtio_snd_pcm_status, which makes the available
space for audio data zero. This commit adds a check for reaching the
maximum buffer size before attempting any writes.
Reported-by: Zheyu Ma <zheyuma97@gmail.com>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2427
Signed-off-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Message-Id: <virtio-snd-fuzz-2427-fix-v1-manos.pitsidianakis@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
(cherry picked from commit 98e77e3dd8)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The allocated memory to hold LBA ranges leaks in the nvme_dsm function. This
happens because the allocated memory for iocb->range is not freed in all
error handling paths.
Fix this by adding a free to ensure that the allocated memory is properly freed.
ASAN log:
==3075137==ERROR: LeakSanitizer: detected memory leaks
Direct leak of 480 byte(s) in 6 object(s) allocated from:
#0 0x55f1f8a0eddd in malloc llvm/compiler-rt/lib/asan/asan_malloc_linux.cpp:129:3
#1 0x7f531e0f6738 in g_malloc (/lib/x86_64-linux-gnu/libglib-2.0.so.0+0x5e738)
#2 0x55f1faf1f091 in blk_aio_get block/block-backend.c:2583:12
#3 0x55f1f945c74b in nvme_dsm hw/nvme/ctrl.c:2609:30
#4 0x55f1f945831b in nvme_io_cmd hw/nvme/ctrl.c:4470:16
#5 0x55f1f94561b7 in nvme_process_sq hw/nvme/ctrl.c:7039:29
Cc: qemu-stable@nongnu.org
Fixes: d7d1474fd8 ("hw/nvme: reimplement dsm to allow cancellation")
Signed-off-by: Zheyu Ma <zheyuma97@gmail.com>
Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
(cherry picked from commit c510fe78f1)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
In commit c1a1f80518 when we added the FEAT_LSE2 relaxations to
the alignment requirements for atomic and ordered loads and stores,
we didn't quite get it right for LDAPR/LDAPRH/LDAPRB with no
immediate offset. These instructions were handled in the old decoder
as part of disas_ldst_atomic(), but unlike all the other insns that
function decoded (LDADD, LDCLR, etc) these insns are "ordered", not
"atomic", so they should be using check_ordered_align() rather than
check_atomic_align(). Commit c1a1f80518 used
check_atomic_align() regardless for everything in
disas_ldst_atomic(). We then carried that incorrect check over in
the decodetree conversion, where LDAPR/LDAPRH/LDAPRB are now handled
by trans_LDAPR().
The effect is that when FEAT_LSE2 is implemented, these instructions
don't honour the SCTLR_ELx.nAA bit and will generate alignment
faults when they should not.
(The LDAPR insns with an immediate offset were in disas_ldst_ldapr_stlr()
and then in trans_LDAPR_i() and trans_STLR_i(), and have always used
the correct check_ordered_align().)
Use check_ordered_align() in trans_LDAPR().
Cc: qemu-stable@nongnu.org
Fixes: c1a1f80518 ("target/arm: Relax ordered/atomic alignment checks for LSE2")
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240709134504.3500007-3-peter.maydell@linaro.org
(cherry picked from commit 25489b521b)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When we converted the LDAPR/STLR instructions to decodetree we
accidentally introduced a regression where the offset is negative.
The 9-bit immediate field is signed, and the old hand decoder
correctly used sextract32() to get it out of the insn word,
but the ldapr_stlr_i pattern in the decode file used "imm:9"
instead of "imm:s9", so it treated the field as unsigned.
Fix the pattern to treat the field as a signed immediate.
Cc: qemu-stable@nongnu.org
Fixes: 2521b6073b ("target/arm: Convert LDAPR/STLR (imm) to decodetree")
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2419
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240709134504.3500007-2-peter.maydell@linaro.org
(cherry picked from commit 5669d26ec6)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Commit 3089637461 ("scsi: Don't ignore most usb-storage properties")
removed the call to object_property_set_int() and thus the 'set'
method for the bootindex property was also not called anymore. Here
that method is device_set_bootindex() (as configured by
scsi_dev_instance_init() -> device_add_bootindex_property()) which as
a side effect registers the device via add_boot_device_path().
As reported by a downstream user [0], the bootindex property did not
have the desired effect anymore for legacy drives. Fix the regression
by explicitly calling the add_boot_device_path() function after
checking that the bootindex is not yet used (to avoid
add_boot_device_path() calling exit()).
[0]: https://forum.proxmox.com/threads/149772/post-679433
Cc: qemu-stable@nongnu.org
Fixes: 3089637461 ("scsi: Don't ignore most usb-storage properties")
Suggested-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
Link: https://lore.kernel.org/r/20240710152529.1737407-1-f.ebner@proxmox.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 57a8a80d1a)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When a command's arguments are specified as an explicit type T,
generated documentation points to the members of T.
Example:
##
# @announce-self:
#
# Trigger generation of broadcast RARP frames to update network
[...]
##
{ 'command': 'announce-self', 'boxed': true,
'data' : 'AnnounceParameters'}
generates
"announce-self" (Command)
-------------------------
Trigger generation of broadcast RARP frames to update network
[...]
Arguments
~~~~~~~~~
The members of "AnnounceParameters"
Except when the command takes its arguments unboxed , i.e. it doesn't
have 'boxed': true, we generate *nothing*. A few commands have a
reference in their doc comment to compensate, but most don't.
Example:
##
# @blockdev-snapshot-sync:
#
# Takes a synchronous snapshot of a block device.
#
# For the arguments, see the documentation of BlockdevSnapshotSync.
[...]
##
{ 'command': 'blockdev-snapshot-sync',
'data': 'BlockdevSnapshotSync',
'allow-preconfig': true }
generates
"blockdev-snapshot-sync" (Command)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Takes a synchronous snapshot of a block device.
For the arguments, see the documentation of BlockdevSnapshotSync.
[...]
Same for event data.
Fix qapidoc.py to generate the reference regardless of boxing. Delete
now redundant references in the doc comments.
Fixes: 4078ee5469 (docs/sphinx: Add new qapi-doc Sphinx extension)
Cc: qemu-stable@nongnu.org
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240628112756.794237-1-armbru@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
(cherry picked from commit e389929d19)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When handling image filenames from legacy options such as -drive or from
tools, these filenames are parsed for protocol prefixes, including for
the json:{} pseudo-protocol.
This behaviour is intended for filenames that come directly from the
command line and for backing files, which may come from the image file
itself. Higher level management tools generally take care to verify that
untrusted images don't contain a bad (or any) backing file reference;
'qemu-img info' is a suitable tool for this.
However, for other files that can be referenced in images, such as
qcow2 data files or VMDK extents, the string from the image file is
usually not verified by management tools - and 'qemu-img info' wouldn't
be suitable because in contrast to backing files, it already opens these
other referenced files. So here the string should be interpreted as a
literal local filename. More complex configurations need to be specified
explicitly on the command line or in QMP.
This patch changes bdrv_open_inherit() so that it only parses filenames
if a new parameter parse_filename is true. It is set for the top level
in bdrv_open(), for the file child and for the backing file child. All
other callers pass false and disable filename parsing this way.
Cc: qemu-stable@nongnu.org
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
(cherry picked from commit 7ead946998)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
We want to disable filename parsing for data files because it's too easy
to abuse in malicious image files. Make the test ready for the change by
passing the data file explicitly in command line options.
Cc: qemu-stable@nongnu.org
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
(cherry picked from commit 7e1110664e)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
We want to disable filename parsing for data files because it's too easy
to abuse in malicious image files. Make the test ready for the change by
passing the data file explicitly in command line options.
Cc: qemu-stable@nongnu.org
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
(cherry picked from commit 2eb42a728d)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
One use case for 'qemu-img info' is verifying that untrusted images
don't reference an unwanted external file, be it as a backing file or an
external data file. To make sure that calling 'qemu-img info' can't
already have undesired side effects with a malicious image, just don't
open the data file at all with BDRV_O_NO_IO. If nothing ever tries to do
I/O, we don't need to have it open.
This changes the output of iotests case 061, which used 'qemu-img info'
to show that opening an image with an invalid data file fails. After
this patch, it succeeds. Replace this part of the test with a qemu-io
call, but keep the final 'qemu-img info' to show that the invalid data
file is correctly displayed in the output.
Fixes: CVE-2024-4467
Cc: qemu-stable@nongnu.org
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
(cherry picked from commit bd385a5298)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
This effectively reverts
commit 54c4ea8f3a
Author: Zhao Liu <zhao1.liu@intel.com>
Date: Sat Mar 9 00:01:37 2024 +0800
hw/core/machine-smp: Deprecate unsupported "parameter=1" SMP configurations
but is not done as a 'git revert' since the part of the changes to the
file hw/core/machine-smp.c which add 'has_XXX' checks remain desirable.
Furthermore, we have to tweak the subsequently added unit test to
account for differing warning message.
The rationale for the original deprecation was:
"Currently, it was allowed for users to specify the unsupported
topology parameter as "1". For example, x86 PC machine doesn't
support drawer/book/cluster topology levels, but user could specify
"-smp drawers=1,books=1,clusters=1".
This is meaningless and confusing, so that the support for this kind
of configurations is marked deprecated since 9.0."
There are varying POVs on the topic of 'unsupported' topology levels.
It is common to say that on a system without hyperthreading, that there
is always 1 thread. Likewise when new CPUs introduced a concept of
multiple "dies', it was reasonable to say that all historical CPUs
before that implicitly had 1 'die'. Likewise for the more recently
introduced 'modules' and 'clusters' parameter'. From this POV, it is
valid to set 'parameter=1' on the -smp command line for any machine,
only a value > 1 is strictly an error condition.
It doesn't cause any functional difficulty for QEMU, because internally
the QEMU code is itself assuming that all "unsupported" parameters
implicitly have a value of '1'.
At the libvirt level, we've allowed applications to set 'parameter=1'
when configuring a guest, and pass that through to QEMU.
Deprecating this creates extra difficulty for because there's no info
exposed from QEMU about which machine types "support" which parameters.
Thus, libvirt can't know whether it is valid to pass 'parameter=1' for
a given machine type, or whether it will trigger deprecation messages.
Since there's no apparent functional benefit to deleting this deprecated
behaviour from QEMU, and it creates problems for consumers of QEMU,
remove this deprecation.
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Message-ID: <20240513123358.612355-2-berrange@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit 9d7950edb0)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: remove hunk about modules in hw/core/machine-smp.c introduced in
v9.0.0-155-g8ec0a4634798 "hw/core/machine: Support modules in -smp")
When QEMU is started with:
-cpu host,host-cache-info=on,l3-cache=off \
-smp 2,sockets=1,dies=1,cores=1,threads=2
Guest can't acquire maximum number of addressable IDs for processor cores in
the physical package from CPUID[04H].
When creating a CPU topology of 1 core per package, host-cache-info only
uses the Host's addressable core IDs field (CPUID.04H.EAX[bits 31-26]),
resulting in a conflict (on the multicore Host) between the Guest core
topology information in this field and the Guest's actual cores number.
Fix it by removing the unnecessary condition to cover 1 core per package
case. This is safe because cores_per_pkg will not be 0 and will be at
least 1.
Fixes: d7caf13b5f ("x86: cpu: fixup number of addressable IDs for logical processors sharing cache")
Signed-off-by: Guixiong Wei <weiguixiong@bytedance.com>
Signed-off-by: Yipeng Yin <yinyipeng@bytedance.com>
Signed-off-by: Chuang Xu <xuchuangxclwt@bytedance.com>
Reviewed-by: Zhao Liu <zhao1.liu@intel.com>
Message-ID: <20240611032314.64076-1-xuchuangxclwt@bytedance.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 903916f0a0)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: fixup for 9.0 due to other changes in this area past 9.0)
When the "file:" migration support was added we missed the special
case in the qemu_open_old implementation that allows for a particular
file name format to be used to refer to a set of file descriptors that
have been previously provided to QEMU via the add-fd QMP command.
When using this fdset feature, we should not truncate the migration
file because being given an fd means that the management layer is in
control of the file and will likely already have some data written to
it. This is further indicated by the presence of the 'offset'
argument, which indicates the start of the region where QEMU is
allowed to write.
Fix the issue by replacing the O_TRUNC flag on open by an ftruncate
call, which will take the offset into consideration.
Fixes: 385f510df5 ("migration: file URI offset")
Suggested-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Prasad Pandit <pjp@fedoraproject.org>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
(cherry picked from commit 6d3279655a)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Simplify the logic for two-part, 32-bit pc-relative addresses.
Rather than assume all such fit in int32_t, do some arithmetic
and assert a result, do some arithmetic first and then check
to see if the pieces are in range.
Cc: qemu-stable@nongnu.org
Fixes: dacc51720d ("tcg/loongarch64: Implement tcg_out_mov and tcg_out_movi")
Reviewed-by: Song Gao <gaosong@loongson.cn>
Reported-by: Song Gao <gaosong@loongson.cn>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
(cherry picked from commit 521d7fb3eb)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The VIRTIO Sound Device conforms with the Virtio spec v1.2,
thus only use little endianness.
Remove the suspicious target_words_bigendian() noticed during
code review.
Cc: qemu-stable@nongnu.org
Fixes: eb9ad377bb ("virtio-sound: handle control messages and streams")
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Message-Id: <20240422211830.25606-1-philmd@linaro.org>
(cherry picked from commit a276ec8e26)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Commit 1f25c172f8 ("monitor: use aio_co_reschedule_self()") was a code
cleanup that uses aio_co_reschedule_self() instead of open coding
coroutine rescheduling.
Bug RHEL-34618 was reported and Kevin Wolf <kwolf@redhat.com> identified
the root cause. I missed that aio_co_reschedule_self() ->
qemu_get_current_aio_context() only knows about
qemu_aio_context/IOThread AioContexts and not about iohandler_ctx. It
does not function correctly when going back from the iohandler_ctx to
qemu_aio_context.
Go back to open coding the AioContext transitions to avoid this bug.
This reverts commit 1f25c172f8.
Cc: qemu-stable@nongnu.org
Buglink: https://issues.redhat.com/browse/RHEL-34618
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20240506190622.56095-2-stefanha@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit 719c6819ed)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Draw routine needs to be manually invoked in the next refresh
if there is a scanout blob from the guest. This is to prevent
a situation where there is a scheduled draw event but it won't
happen bacause the window is currently in inactive state
(minimized or tabified). If draw is not done for a long time,
gl_block timeout and/or fence timeout (on the guest) will happen
eventually.
v2: Use gd_gl_area_draw(vc) in gtk-gl-area.c
Suggested-by: Vivek Kasireddy <vivek.kasireddy@intel.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Cc: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Dongwon Kim <dongwon.kim@intel.com>
Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20240426225059.3871283-1-dongwon.kim@intel.com>
(cherry picked from commit 77bf310084)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Reproducer from https://gitlab.com/qemu-project/qemu/-/issues/1451
creates small packet (1 segment, len = 10 == n->guest_hdr_len),
then destroys queue.
"if (n->host_hdr_len != n->guest_hdr_len)" is triggered, if body creates
zero length/zero segment packet as there is nothing after guest header.
qemu_sendv_packet_async() tries to send it.
slirp discards it because it is smaller than Ethernet header,
but returns 0 because tx hooks are supposed to return total length of data.
0 is propagated upwards and is interpreted as "packet has been sent"
which is terrible because queue is being destroyed, nobody is waiting for TX
to complete and assert it triggered.
Fix is discard such empty packets instead of sending them.
Length 1 packets will go via different codepath:
virtqueue_push(q->tx_vq, elem, 0);
virtio_notify(vdev, q->tx_vq);
g_free(elem);
and aren't problematic.
Signed-off-by: Alexey Dobriyan <adobriyan@yandex-team.ru>
Signed-off-by: Jason Wang <jasowang@redhat.com>
(cherry picked from commit 2c3e4e2de6)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The calculation of FrameTemp is done using the size indicated by mo_pushpop()
before being written back to EBP, but the final writeback to EBP is done using
the size indicated by mo_stacksize().
In the case where mo_pushpop() is MO_32 and mo_stacksize() is MO_16 then the
final writeback to EBP is done using MO_16 which can leave junk in the top
16-bits of EBP after executing ENTER.
Change the writeback of EBP to use the same size indicated by mo_pushpop() to
ensure that the full value is written back.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2198
Message-ID: <20240606095319.229650-5-mark.cave-ayland@ilande.co.uk>
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 3973615e7f)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
By default, SDL disables the screen saver which prevents the host from powering
down the screen even if the screen is locked. This results in draining the
battery needlessly when the host isn't connected to a wall charger. Fix that by
enabling the screen saver.
Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-ID: <20240512095945.1879-1-shentey@gmail.com>
(cherry picked from commit 2e701e6785)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Commit dfcf74fa ("virtio-gpu: fix scanout migration post-load") broke
forward/backward version migration. Versioning of nested VMSD structures
is not straightforward, as the wire format doesn't have nested
structures versions. Introduce x-scanout-vmstate-version and a field
test to save/load appropriately according to the machine version.
Fixes: dfcf74fa ("virtio-gpu: fix scanout migration post-load")
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Fiona Ebner <f.ebner@proxmox.com>
Tested-by: Fiona Ebner <f.ebner@proxmox.com>
[fixed long lines]
Signed-off-by: Fabiano Rosas <farosas@suse.de>
(cherry picked from commit 40a23ef643)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
xsave.flat checks that "executing the XSETBV instruction causes a general-
protection fault (#GP) if ECX = 0 and EAX[2:1] has the value 10b". QEMU allows
that option, so the test fails. Add the condition.
Cc: qemu-stable@nongnu.org
Fixes: 892544317f ("target/i386: implement XSAVE and XRSTOR of AVX registers", 2022-10-18)
Reported-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 7604bbc2d8)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
In AIA spec, each hart (or each hart within a group) has a unique hart
number to locate the memory pages of interrupt files in the address
space. The number of bits required to represent any hart number is equal
to ceil(log2(hmax + 1)), where hmax is the largest hart number among
groups.
However, if the largest hart number among groups is a power of 2, QEMU
will pass an inaccurate hart-index-bit setting to Linux. For example, when
the guest OS has 4 harts, only ceil(log2(3 + 1)) = 2 bits are sufficient
to represent 4 harts, but we passes 3 to Linux. The code needs to be
updated to ensure accurate hart-index-bit settings.
Additionally, a Linux patch[1] is necessary to correctly recover the hart
index when the guest OS has only 1 hart, where the hart-index-bit is 0.
[1] https://lore.kernel.org/lkml/20240415064905.25184-1-yongxuan.wang@sifive.com/t/
Signed-off-by: Yong-Xuan Wang <yongxuan.wang@sifive.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Cc: qemu-stable <qemu-stable@nongnu.org>
Message-ID: <20240515091129.28116-1-yongxuan.wang@sifive.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
(cherry picked from commit 190b867f28)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When running the instruction
```
cbo.flush 0(x0)
```
QEMU would segfault.
The issue was in cpu_gpr[a->rs1] as QEMU does not have cpu_gpr[0]
allocated.
In order to fix this let's use the existing get_address()
helper. This also has the benefit of performing pointer mask
calculations on the address specified in rs1.
The pointer masking specificiation specifically states:
"""
Cache Management Operations: All instructions in Zicbom, Zicbop and Zicboz
"""
So this is the correct behaviour and we previously have been incorrectly
not masking the address.
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Reported-by: Fabian Thomas <fabian.thomas@cispa.de>
Fixes: e05da09b7c ("target/riscv: implement Zicbom extension")
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Cc: qemu-stable <qemu-stable@nongnu.org>
Message-ID: <20240514023910.301766-1-alistair.francis@wdc.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
(cherry picked from commit c5eb8d6336)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Previous patch fixed the PMP priority in raise_mmu_exception() but we're still
setting mtval2 incorrectly. In riscv_cpu_tlb_fill(), after pmp check in 2 stage
translation part, mtval2 will be set in case of successes 2 stage translation but
failed pmp check.
In this case we gonna set mtval2 via env->guest_phys_fault_addr in context of
riscv_cpu_tlb_fill(), as this was a guest-page-fault, but it didn't and mtval2
should be zero, according to RISCV privileged spec sect. 9.4.4: When a guest
page-fault is taken into M-mode, mtval2 is written with either zero or guest
physical address that faulted, shifted by 2 bits. *For other traps, mtval2
is set to zero...*
Signed-off-by: Alexei Filippov <alexei.filippov@syntacore.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-ID: <20240503103052.6819-1-alexei.filippov@syntacore.com>
Cc: qemu-stable <qemu-stable@nongnu.org>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
(cherry picked from commit 6c9a344247)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
raise_mmu_exception(), as is today, is prioritizing guest page faults by
checking first if virt_enabled && !first_stage, and then considering the
regular inst/load/store faults.
There's no mention in the spec about guest page fault being a higher
priority that PMP faults. In fact, privileged spec section 3.7.1 says:
"Attempting to fetch an instruction from a PMP region that does not have
execute permissions raises an instruction access-fault exception.
Attempting to execute a load or load-reserved instruction which accesses
a physical address within a PMP region without read permissions raises a
load access-fault exception. Attempting to execute a store,
store-conditional, or AMO instruction which accesses a physical address
within a PMP region without write permissions raises a store
access-fault exception."
So, in fact, we're doing it wrong - PMP faults should always be thrown,
regardless of also being a first or second stage fault.
The way riscv_cpu_tlb_fill() and get_physical_address() work is
adequate: a TRANSLATE_PMP_FAIL error is immediately reported and
reflected in the 'pmp_violation' flag. What we need is to change
raise_mmu_exception() to prioritize it.
Reported-by: Joseph Chan <jchan@ventanamicro.com>
Fixes: 82d53adfbb ("target/riscv/cpu_helper.c: Invalid exception on MMU translation stage")
Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-ID: <20240413105929.7030-1-alexei.filippov@syntacore.com>
Cc: qemu-stable <qemu-stable@nongnu.org>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
(cherry picked from commit 68e7c86927)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The require_scale_rvf function only checks the double width operator for
the vector floating point widen instructions, so most of the widen
checking functions need to add require_rvf for single width operator.
The vfwcvt.f.x.v and vfwcvt.f.xu.v instructions convert single width
integer to double width float, so the opfxv_widen_check function doesn’t
need require_rvf for the single width operator(integer).
Signed-off-by: Max Chou <max.chou@sifive.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Cc: qemu-stable <qemu-stable@nongnu.org>
Message-ID: <20240322092600.1198921-3-max.chou@sifive.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
(cherry picked from commit 7a999d4dd7)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Running a KVM guest using a 6.9-rc3 kernel, in a 6.8 host that has zkr
enabled, will fail with a kernel oops SIGILL right at the start. The
reason is that we can't expose zkr without implementing the SEED CSR.
Disabling zkr in the guest would be a workaround, but if the KVM doesn't
allow it we'll error out and never boot.
In hindsight this is too strict. If we keep proceeding, despite not
disabling the extension in the KVM vcpu, we'll not add the extension in
the riscv,isa. The guest kernel will be unaware of the extension, i.e.
it doesn't matter if the KVM vcpu has it enabled underneath or not. So
it's ok to keep booting in this case.
Change our current logic to not error out if we fail to disable an
extension in kvm_set_one_reg(), but show a warning and keep booting. It
is important to throw a warning because we must make the user aware that
the extension is still available in the vcpu, meaning that an
ill-behaved guest can ignore the riscv,isa settings and use the
extension.
The case we're handling happens with an EINVAL error code. If we fail to
disable the extension in KVM for any other reason, error out.
We'll also keep erroring out when we fail to enable an extension in KVM,
since adding the extension in riscv,isa at this point will cause a guest
malfunction because the extension isn't enabled in the vcpu.
Suggested-by: Andrew Jones <ajones@ventanamicro.com>
Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Cc: qemu-stable <qemu-stable@nongnu.org>
Message-ID: <20240422171425.333037-2-dbarboza@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
(cherry picked from commit 1215d45b2a)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The Zkr extension may only be exposed to KVM guests if the VMM
implements the SEED CSR. Use the same implementation as TCG.
Without this patch, running with a KVM which does not forward the
SEED CSR access to QEMU will result in an ILL exception being
injected into the guest (this results in Linux guests crashing on
boot). And, when running with a KVM which does forward the access,
QEMU will crash, since QEMU doesn't know what to do with the exit.
Fixes: 3108e2f1c6 ("target/riscv/kvm: update KVM exts to Linux 6.8")
Signed-off-by: Andrew Jones <ajones@ventanamicro.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Cc: qemu-stable <qemu-stable@nongnu.org>
Message-ID: <20240422134605.534207-2-ajones@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
(cherry picked from commit 86997772fa)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Prevent regressions when using NBD with TLS in the presence of
iothreads, adding coverage the fix to qio channels made in the
previous patch.
The shell function pick_unused_port() was copied from
nbdkit.git/tests/functions.sh.in, where it had all authors from Red
Hat, agreeing to the resulting relicensing from 2-clause BSD to GPLv2.
CC: qemu-stable@nongnu.org
CC: "Richard W.M. Jones" <rjones@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20240531180639.1392905-6-eblake@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
(cherry picked from commit a73c993780)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Since qemu 8.2, the combination of NBD + TLS + iothread crashes on an
assertion failure:
qemu-kvm: ../io/channel.c:534: void qio_channel_restart_read(void *): Assertion `qemu_get_current_aio_context() == qemu_coroutine_get_aio_context(co)' failed.
It turns out that when we removed AioContext locking, we did so by
having NBD tell its qio channels that it wanted to opt in to
qio_channel_set_follow_coroutine_ctx(); but while we opted in on the
main channel, we did not opt in on the TLS wrapper channel.
qemu-iotests has coverage of NBD+iothread and NBD+TLS, but apparently
no coverage of NBD+TLS+iothread, or we would have noticed this
regression sooner. (I'll add that in the next patch)
But while we could manually opt in to the TLS channel in nbd/server.c
(a one-line change), it is more generic if all qio channels that wrap
other channels inherit the follow status, in the same way that they
inherit feature bits.
CC: Stefan Hajnoczi <stefanha@redhat.com>
CC: Daniel P. Berrangé <berrange@redhat.com>
CC: qemu-stable@nongnu.org
Fixes: https://issues.redhat.com/browse/RHEL-34786
Fixes: 06e0f098 ("io: follow coroutine AioContext in qio_channel_yield()", v8.2.0)
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240518025246.791593-5-eblake@redhat.com>
(cherry picked from commit 199e84de1c)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
In gic_cpu_read() and gic_cpu_write(), we delegate the handling of
reading and writing the Non-Secure view of the GICC_APR<n> registers
to functions gic_apr_ns_view() and gic_apr_write_ns_view().
Unfortunately we got the order of the arguments wrong, swapping the
CPU number and the register number (which the compiler doesn't catch
because they're both integers).
Most guests probably didn't notice this bug because directly
accessing the APR registers is typically something only done by
firmware when it is doing state save for going into a sleep mode.
Correct the mismatched call arguments.
Found by Linux Verification Center (linuxtesting.org) with SVACE.
Cc: qemu-stable@nongnu.org
Fixes: 51fd06e0ee ("hw/intc/arm_gic: Fix handling of GICC_APR<n>, GICC_NSAPR<n> registers")
Signed-off-by: Andrey Shumilin <shum.sdl@nppct.ru>
[PMM: Rewrote commit message]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Alex Bennée<alex.bennee@linaro.org>
(cherry picked from commit daafa78b29)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
We wrongly encoded ID_AA64PFR1_EL1 using {3,0,0,4,2} in hvf_sreg_match[] so
we fail to get the expected ARMCPRegInfo from cp_regs hash table with the
wrong key.
Fix it with the correct encoding {3,0,0,4,1}. With that fixed, the Linux
guest can properly detect FEAT_SSBS2 on my M1 HW.
All DBG{B,W}{V,C}R_EL1 registers are also wrongly encoded with op0 == 14.
It happens to work because HVF_SYSREG(CRn, CRm, 14, op1, op2) equals to
HVF_SYSREG(CRn, CRm, 2, op1, op2), by definition. But we shouldn't rely on
it.
Cc: qemu-stable@nongnu.org
Fixes: a1477da3dd ("hvf: Add Apple Silicon support")
Signed-off-by: Zenghui Yu <zenghui.yu@linux.dev>
Reviewed-by: Alexander Graf <agraf@csgraf.de>
Message-id: 20240503153453.54389-1-zenghui.yu@linux.dev
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
(cherry picked from commit 19ed42e8ad)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The TSAN job started failing when gitlab rolled out their latest
release. The root cause is a change in the Google COS version used
on shared runners. This brings a kernel running with
vm.mmap_rnd_bits = 31
which is incompatible with TSAN in LLVM < 18, which only supports
upto '28'. LLVM 18 can support upto '30', and failing that will
re-exec itself to turn off VA randomization.
Our LLVM is too old for now, but we can run with 'setarch -R make ..'
to turn off VA randomization ourselves.
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20240513111551.488088-4-berrange@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
(cherry picked from commit b563959b90)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Intel SDM 18.3.1.4 "If an occurrence of the MOV or POP instruction
loads the SS register executes with EFLAGS.TF = 1, no single-step debug
exception occurs following the MOV or POP instruction."
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit f0f0136abb)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Migration from an 8.2 or 9.0 binary to an 8.1 binary with machine
version 8.1 can fail with:
> kvm: Features 0x1c0010130afffa7 unsupported. Allowed features: 0x10179bfffe7
> kvm: Failed to load virtio-net:virtio
> kvm: error while loading state for instance 0x0 of device '0000:00:12.0/virtio-net'
> kvm: load of migration failed: Operation not permitted
The series
53da8b5a99 virtio-net: Add support for USO features
9da1684954 virtio-net: Add USO flags to vhost support.
f03e0cf63b tap: Add check for USO features
2ab0ec3121 tap: Add USO support to tap device.
only landed in QEMU 8.2, so the compatibility flags should be part of
machine version 8.1.
Moving the flags unfortunately breaks forward migration with machine
version 8.1 from a binary without this patch to a binary with this
patch.
Fixes: 53da8b5a99 ("virtio-net: Add support for USO features")
Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Acked-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
(cherry picked from commit 9710401276)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
This bug fix addresses the incorrect return value of kvm_hv_handle_exit for
KVM_EXIT_HYPERV_SYNIC, which should be EXCP_INTERRUPT.
Handling of KVM_EXIT_HYPERV_SYNIC in QEMU needs to be synchronous.
This means that async_synic_update should run in the current QEMU vCPU
thread before returning to KVM, returning EXCP_INTERRUPT to guarantee this.
Returning 0 can cause async_synic_update to run asynchronously.
One problem (kvm-unit-tests's hyperv_synic test fails with timeout error)
caused by this bug:
When a guest VM writes to the HV_X64_MSR_SCONTROL MSR to enable Hyper-V SynIC,
a VM exit is triggered and processed by the kvm_hv_handle_exit function of the
QEMU vCPU. This function then calls the async_synic_update function to set
synic->sctl_enabled to true. A true value of synic->sctl_enabled is required
before creating SINT routes using the hyperv_sint_route_new() function.
If kvm_hv_handle_exit returns 0 for KVM_EXIT_HYPERV_SYNIC, the current QEMU
vCPU thread may return to KVM and enter the guest VM before running
async_synic_update. In such case, the hyperv_synic test’s subsequent call to
synic_ctl(HV_TEST_DEV_SINT_ROUTE_CREATE, ...) immediately after writing to
HV_X64_MSR_SCONTROL can cause QEMU’s hyperv_sint_route_new() function to return
prematurely (because synic->sctl_enabled is false).
If the SINT route is not created successfully, the SINT interrupt will not be
fired, resulting in a timeout error in the hyperv_synic test.
Fixes: 267e071bd6 (“hyperv: make overlay pages for SynIC”)
Suggested-by: Chao Gao <chao.gao@intel.com>
Signed-off-by: Dongsheng Zhang <dongsheng.x.zhang@intel.com>
Message-ID: <20240521200114.11588-1-dongsheng.x.zhang@intel.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 84d4b72854)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Remove gtk_widget_get_scale_factor() usage from the calculation of
the motion events in the GTK backend to make it work correctly on
environments that have `gtk_widget_get_scale_factor() != 1`.
This scale factor usage had been introduced in the commit f14aab420c and
at that time the window size was used for calculating the things and it
was working correctly. However, in the commit 2f31663ed4 the logic
switched to use the widget size instead of window size and because of
the change the usage of scale factor becomes invalid (since widgets use
`vc->gfx.scale_{x, y}` for scaling).
Tested on Crostini on ChromeOS (15823.51.0) with an external display.
Fixes: 2f31663ed4 ("ui/gtk: use widget size for cursor motion event")
Fixes: f14aab420c ("ui: fix incorrect pointer position on highdpi with
gtk")
Signed-off-by: hikalium <hikalium@hikalium.com>
Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20240512111435.30121-3-hikalium@hikalium.com>
(cherry picked from commit 37e9141501)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
If you try to run the configure script on a system without a working
C compiler, you get a very misleading error message:
ERROR: Unrecognized host OS (uname -s reports 'Linux')
Some people already opened bug tickets because of this problem:
https://gitlab.com/qemu-project/qemu/-/issues/2057https://gitlab.com/qemu-project/qemu/-/issues/2288
We should rather tell the user that we were not able to use the C
compiler instead, otherwise they will have a hard time to figure
out what was going wrong.
While we're at it, let's also suppress the "unrecognized host CPU"
message in this case since it is rather misleading than helpful.
Fixes: 264b803721 ("configure: remove compiler sanity check")
Message-ID: <20240513114010.51608-1-thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>
(cherry picked from commit 371d60dfdb)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Ensure that they go through unmodified, instead of removing one layer
of quoting.
-D is a pretty specialized option and most options that can have spaces
do not need it (for example, c_args is covered by --extra-cflags).
Therefore it's unlikely that this causes actual trouble. However,
a somewhat realistic failure case would be with -Dpkg_config_path
and a pkg-config directory that contains spaces.
Cc: qemu-stable@nongnu.org
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 23b1f53c2c)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The VMX feature bit depends on general availability of WAITPKG,
not the other way round.
Fixes: 33cc88261c ("target/i386: add support for VMX_SECONDARY_EXEC_ENABLE_USER_WAIT_PAUSE", 2023-08-28)
Cc: qemu-stable@nongnu.org
Reviewed-by: Zhao Liu <zhao1.liu@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit fe01af5d47)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
According to the manual, 32-bit vs 64-bit is governed by REX.W
and REX ignores the 0x66 prefix. This can be confirmed with this
program:
#include <stdio.h>
int main()
{
int x = 0x12340000;
int y;
asm("popcntl %1, %0" : "=r" (y) : "r" (x)); printf("%x\n", y);
asm("mov $-1, %0; .byte 0x66; popcntl %1, %0" : "+r" (y) : "r" (x)); printf("%x\n", y);
asm("mov $-1, %0; .byte 0x66; popcntq %q1, %q0" : "+r" (y) : "r" (x)); printf("%x\n", y);
}
which prints 5/ffff0000/5 on real hardware and 5/ffff0000/ffff0000
on QEMU.
Cc: qemu-stable@nongnu.org
Reviewed-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 41c685dc59)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: drop removal of mo_64_32() helper function in target/i386/tcg/translate.c
due to missing-in-9.0 v9.0.0-542-gaef4f4affde2
"target/i386: remove now-converted opcodes from old decoder"
which removed other user of it)
The char pointer 'ramName' point to a block of memory,
but never free it. Use 'g_autofree' to automatically free it.
Resolves: Coverity CID 1544773
Fixes: 0cf1478d6 ("hw/loongarch: Add numa support")
Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240507022239.3113987-1-gaosong@loongson.cn>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit 54c52ec719)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(Mjt: context fixup in hw/loongarch/virt.c due to missing-in-9.0
v9.0.0-266-gd771ca1c10 "hw/loongarch: Move boot functions to boot.c")
Match the extra inserts of INDEX_op_insn_start, fixing
the db->num_insns != 1 assert in translator_loop.
Fixes: dcd092a063 ("accel/tcg: Improve can_do_io management")
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
(cherry picked from commit ca51921158)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When emulated with QEMU, interrupts will never come in the following
loop. However, if the NOP instruction is uncommented, interrupts will
fire as normal.
loop:
cli
call do_sti
jmp loop
do_sti:
sti
# nop
ret
This behavior is different from that of a real processor. For example,
if KVM is enabled, interrupts will always fire regardless of whether the
NOP instruction is commented or not. Also, the Intel Software Developer
Manual states that after the STI instruction is executed, the interrupt
inhibit should end as soon as the next instruction (e.g., the RET
instruction if the NOP instruction is commented) is executed.
This problem is caused because the previous code may choose not to end
the TB even if the HF_INHIBIT_IRQ_MASK has just been reset (e.g., in the
case where the STI instruction is immediately followed by the RET
instruction), so that IRQs may not have a change to trigger. This commit
fixes the problem by always terminating the current TB to give IRQs a
chance to trigger when HF_INHIBIT_IRQ_MASK is reset.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Ruihan Li <lrh2000@pku.edu.cn>
Message-ID: <20240415064518.4951-4-lrh2000@pku.edu.cn>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 6a5a63f74b)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
The DMA descriptor structures for this device have
a set of "address extension" fields which extend the 32
bit source addresses with an extra 16 bits to give a
48 bit address:
https://docs.amd.com/r/en-US/ug1085-zynq-ultrascale-trm/ADDR_EXT-Field
However, we misimplemented this address extension in several ways:
* we only extracted 12 bits of the extension fields, not 16
* we didn't shift the extension field up far enough
* we accidentally did the shift as 32-bit arithmetic, which
meant that we would have an overflow instead of setting
bits [47:32] of the resulting 64-bit address
Add a type cast and use extract64() instead of extract32()
to avoid integer overflow on addition. Fix bit fields
extraction according to documentation.
Found by Linux Verification Center (linuxtesting.org) with SVACE.
Cc: qemu-stable@nongnu.org
Fixes: d3c6369a96 ("introduce xlnx-dpdma")
Signed-off-by: Alexandra Diupina <adiupina@astralinux.ru>
Message-id: 20240428181131.23801-1-adiupina@astralinux.ru
[PMM: adjusted commit message]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
(cherry picked from commit 4b00855f0e)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Cirrus-CI introduced limitations to the free CI minutes. To avoid that
we are consuming them too fast, let's drop the usual targets that are
not that important since they are either a subset of another target
(like i386 or ppc being a subset of x86_64 or ppc64 respectively), or
since there is still a similar target with the opposite endianness
(like xtensa/xtensael, microblaze/microblazeel etc.).
Message-ID: <20240429100113.53357-1-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
(cherry picked from commit a88a04906b)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Commit d424db2354 excluded some strerrorname_np() instances because they
break musl libc builds. Another instance happened to slip by via commit
d4ff3da8f4.
Remove it before it causes trouble again.
Fixes: d4ff3da8f4 (target/riscv/kvm: initialize 'vlenb' via get-reg-list)
Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(cherry picked from commit e442635317)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
nbd_negotiate() is already marked coroutine_fn. And given the fix in
the previous patch to have nbd_negotiate_handle_starttls not create
and wait on a g_main_loop (as that would violate coroutine
constraints), it is worth marking the rest of the related static
functions reachable only during option negotiation as also being
coroutine_fn.
Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20240408160214.1200629-6-eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
[eblake: drop one spurious coroutine_fn marking]
Signed-off-by: Eric Blake <eblake@redhat.com>
(cherry picked from commit 4fa333e08d)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Coroutines are not supposed to block. Instead, they should yield.
The client performs TLS upgrade outside of an AIOContext, during
synchronous handshake; this still requires g_main_loop. But the
server responds to TLS upgrade inside a coroutine, so a nested
g_main_loop is wrong. Since the two callbacks no longer share more
than the setting of data.complete and data.error, it's just as easy to
use static helpers instead of trying to share a common code path. It
is also possible to add assertions that no other code is interfering
with the eventual path to qio reaching the callback, whether or not it
required a yield or main loop.
Fixes: f95910f ("nbd: implement TLS support in the protocol negotiation")
Signed-off-by: Zhu Yangyang <zhuyangyang14@huawei.com>
[eblake: move callbacks to their use point, add assertions]
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20240408160214.1200629-5-eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
(cherry picked from commit ae6d91a7e9)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
bdrv_activate_all() should not be called from the coroutine context, move
it to the QEMU thread colo_process_incoming_thread() with the bql_lock
protected.
The backtrace is as follows:
#4 0x0000561af7948362 in bdrv_graph_rdlock_main_loop () at ../block/graph-lock.c:260
#5 0x0000561af7907a68 in graph_lockable_auto_lock_mainloop (x=0x7fd29810be7b) at /patch/to/qemu/include/block/graph-lock.h:259
#6 0x0000561af79167d1 in bdrv_activate_all (errp=0x7fd29810bed0) at ../block.c:6906
#7 0x0000561af762b4af in colo_incoming_co () at ../migration/colo.c:935
#8 0x0000561af7607e57 in process_incoming_migration_co (opaque=0x0) at ../migration/migration.c:793
#9 0x0000561af7adbeeb in coroutine_trampoline (i0=-106876144, i1=22042) at ../util/coroutine-ucontext.c:175
#10 0x00007fd2a5cf21c0 in () at /lib64/libc.so.6
Cc: qemu-stable@nongnu.org
Cc: Fabiano Rosas <farosas@suse.de>
Closes: https://gitlab.com/qemu-project/qemu/-/issues/2277
Fixes: 2b3912f135 ("block: Mark bdrv_first_blk() and bdrv_is_root_node() GRAPH_RDLOCK")
Signed-off-by: Li Zhijian <lizhijian@fujitsu.com>
Reviewed-by: Zhang Chen <chen.zhang@intel.com>
Tested-by: Zhang Chen <chen.zhang@intel.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20240417025634.1014582-1-lizhijian@fujitsu.com
Signed-off-by: Peter Xu <peterx@redhat.com>
(cherry picked from commit 2cc637f1ea)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
2024-04-25 22:17:37 +03:00
182 changed files with 3413 additions and 1281 deletions
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Michael Tokarev