This reverts commit a59629fcc6.
This is not needed anymore because the IOThread mutex is not
"magic" anymore (need not kick the CPU thread)and also because
fork callbacks are only enabled at the very beginning of
QEMU's execution.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Because of -daemonize, system mode QEMU sometimes needs to fork() and
keep RCU enabled in the child. However, there is a possible deadlock
with synchronize_rcu:
- the CPU thread is inside a RCU critical section and wants to take
the BQL in order to do MMIO
- the monitor thread, which is owning the BQL, calls rcu_init_lock
which tries to take the rcu_sync_lock
- the call_rcu thread has taken rcu_sync_lock in synchronize_rcu, but
synchronize_rcu needs the CPU thread to end the critical section
before returning.
This cannot happen for user-mode emulation, because it does not have
a BQL.
To fix it, assume that system mode QEMU only forks in preparation for
exec (except when daemonizing) and disable pthread_atfork as soon as
the double fork has happened.
Reported-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Tested-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Changes:
KVM T&E segment support for TCG
malta: leave space for the bootmap after the initrd
Apply CP0.PageMask before writing into TLB entry
Fix fallout from indirect branch optimisation
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (GNU/Linux)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=R2lj
-----END PGP SIGNATURE-----
Merge remote-tracking branch 'remotes/yongbok/tags/mips-20170803' into staging
MIPS patches 2017-08-03
Changes:
KVM T&E segment support for TCG
malta: leave space for the bootmap after the initrd
Apply CP0.PageMask before writing into TLB entry
Fix fallout from indirect branch optimisation
# gpg: Signature made Thu 03 Aug 2017 15:32:59 BST
# gpg: using RSA key 0x2238EB86D5F797C2
# gpg: Good signature from "Yongbok Kim <yongbok.kim@imgtec.com>"
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg: It is not certain that the signature belongs to the owner.
# Primary key fingerprint: 8600 4CF5 3415 A5D9 4CFA 2B5C 2238 EB86 D5F7 97C2
* remotes/yongbok/tags/mips-20170803:
target/mips: Fix RDHWR CC with icount
target/mips: Drop redundant gen_io_start/stop()
target/mips: Use BS_EXCP where interrupts are expected
target-mips: apply CP0.PageMask before writing into TLB entry
mips: Add KVM T&E segment support for TCG
mips: Improve segment defs for KVM T&E guests
mips/malta: leave space for the bootmap after the initrd
target-mips: Don't stop on [d]mtc0 DESAVE/KScratch
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Looks like the constant stream of additions of vhost-user devices is a
problem for some people who are concerned about external connections
from qemu. A per-device flag seems like an overkill, but a single
configure flag seems like a sane way to support that, and it looks like
we need to do it before the release.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
-----BEGIN PGP SIGNATURE-----
iQEcBAABAgAGBQJZgx3VAAoJECgfDbjSjVRpxUAH/A2ygeZWY/agnIPN0lF0au35
0sYRyQ5TfM7IGMnR4OpCZCT0q/vZ7bOc9KWfl5u95WEgnOQa2la9U43T+jR+CvN1
3dzAwtHEsUr/Y8sNVwZd0hy+EKN+saL13EBC64BGge7TmFWsOSRwE1CB12VgcNNG
Qx4LlWNA28Vy+pxcgQNGWOsH5oQpiDiEDxpfNcgLGYKcKtbjpveblWlE4iPRUOoq
mJk/tzcBiA/tqZT15VpVi9z+Zy2I06z5wzKjZjOotKjSGCajN6ZUK7lbJ327WwCM
xi3VzHrwrqmtOQF++z/YJ0CRKLiHhKSWv/nY12NaqqgmyT0GRHnKzY8oubNVlxE=
=pa4T
-----END PGP SIGNATURE-----
Merge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging
virtio: fix for rc2
Looks like the constant stream of additions of vhost-user devices is a
problem for some people who are concerned about external connections
from qemu. A per-device flag seems like an overkill, but a single
configure flag seems like a sane way to support that, and it looks like
we need to do it before the release.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# gpg: Signature made Thu 03 Aug 2017 13:57:57 BST
# gpg: using RSA key 0x281F0DB8D28D5469
# gpg: Good signature from "Michael S. Tsirkin <mst@kernel.org>"
# gpg: aka "Michael S. Tsirkin <mst@redhat.com>"
# Primary key fingerprint: 0270 606B 6F3C DF3D 0B17 0970 C350 3912 AFBE 8E67
# Subkey fingerprint: 5D09 FD08 71C8 F85B 94CA 8A0D 281F 0DB8 D28D 5469
* remotes/mst/tags/for_upstream:
build-sys: add --disable-vhost-user
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Learn to compile out vhost-user (net, scsi & upcoming users). Keep it
enabled by default on non-win32, that is assumed to be POSIX. Fail if
trying to enable it on win32.
When trying to make a vhost-user netdev, it gives the following error:
-netdev vhost-user,id=foo,chardev=chr-test: Parameter 'type' expects a netdev backend type
And similar error with the HMP/QMP monitors.
While at it, rename CONFIG_VHOST_NET_TEST CONFIG_VHOST_USER_NET_TEST
since it's a vhost-user specific variable.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
While parsing dhcp options string in 'dhcp_decode', if an options'
length 'len' appeared towards the end of 'bp_vend' array, ensuing
read could lead to an OOB memory access issue. Add check to avoid it.
This is CVE-2017-11434.
Reported-by: Reno Robert <renorobert@gmail.com>
Signed-off-by: Prasad J Pandit <pjp@fedoraproject.org>
Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
With "-netdev user,id=net0,dns=1.2.3.4"
error was:
qemu-system-i386: -netdev user,id=net0,dns=1.2.3.4: Device 'user' could not be initialized
Error is now:
qemu-system-i386: -netdev user,id=net0,dns=1.2.3.4: DNS doesn't belong to network
Signed-off-by: Hervé Poussineau <hpoussin@reactos.org>
Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
RDHWR CC reads the CPU timer like MFC0 CP0_Count, so with icount enabled
it must set can_do_io while it calls the helper to avoid the "Bad icount
read" error. It should also break out of the translation loop to ensure
that timer interrupts are immediately handled.
Fixes: 2e70f6efa8 ("Add instruction counter.")
Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Aurelien Jarno <aurelien@aurel32.net>
Cc: Yongbok Kim <yongbok.kim@imgtec.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Yongbok Kim <yongbok.kim@imgtec.com>
DMTC0 CP0_Cause does a redundant gen_io_start() and gen_io_end() pair,
even though this is done for all DMTC0 operations outside of the switch
statement. Remove these redundant calls.
Fixes: 5dc5d9f055 ("mips: more fixes to the MIPS interrupt glue logic")
Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Yongbok Kim <yongbok.kim@imgtec.com>
Cc: Aurelien Jarno <aurelien@aurel32.net>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Yongbok Kim <yongbok.kim@imgtec.com>
Commit e350d8ca3a ("target/mips: optimize indirect branches") made
indirect branches able to directly find the next TB and jump straight to
it without breaking out of translated code and going around the main
execution loop. This breaks the assumption in target/mips/translate.c
that BS_STOP is sufficient to cause pending interrupts to be handled,
since interrupts are only checked in the main loop.
Fix a few of these assumptions by using gen_save_pc to update the saved
PC and using BS_EXCP instead of BS_STOP:
- [D]MFC0 CP0_Count may trigger a timer interrupt which should be
immediately handled.
- [D]MTC0 CP0_Cause may trigger an interrupt (but in fact translation
was only even being stopped in the DMTC0 case).
- [D]MTC0 CP0_<any> when icount is used is assumed could potentially
cause interrupts.
- EI may trigger an interrupt which was pending. I specifically hit
this case when running KVM nested in mipsel-softmmu. A timer
interrupt while the 2nd guest was executing is caught by KVM which
switches back to the normal Linux exception base and re-enables
interrupts with EI. Since the above commit QEMU doesn't leave
translated code until the nested KVM has already restored the KVM
exception base and returned to the 2nd guest, at which point it is
too late to check for pending interrupts and it gets stuck in an
infinite loop of unhandled interrupts.
Something similar was needed for ARM in commit b29fd33db5
("target/arm: use DISAS_EXIT for eret handling").
Fixes: e350d8ca3a ("target/mips: optimize indirect branches")
Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Aurelien Jarno <aurelien@aurel32.net>
Cc: Yongbok Kim <yongbok.kim@imgtec.com>
Cc: Richard Henderson <rth@twiddle.net>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Yongbok Kim <yongbok.kim@imgtec.com>
PFN0 and PFN1 have to be masked out with PageMask_Mask.
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
Reviewed-by: Yongbok Kim <yongbok.kim@imgtec.com>
[Yongbok Kim:
Added commit message]
Signed-off-by: Yongbok Kim <yongbok.kim@imgtec.com>
MIPS KVM trap & emulate guest kernels have a different segment layout
compared with traditional MIPS kernels, to allow both the user and
kernel code to run from the user address segment without repeatedly
trapping to KVM.
QEMU currently supports this layout only for KVM, but its sometimes
useful to be able to run these kernels in QEMU on a PC, so enable it for
TCG too.
This also paves the way for MIPS KVM VZ support (which uses the normal
virtual memory layout) by abstracting whether user mode kernel segments
are in use.
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Yongbok Kim <yongbok.kim@imgtec.com>
Cc: Aurelien Jarno <aurelien@aurel32.net>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: kvm@vger.kernel.org
Reviewed-by: Richard Henderson <rth@twiddle.net>
[Yongbok Kim:
minor change]
Signed-off-by: Yongbok Kim <yongbok.kim@imgtec.com>
Improve the segment definitions used by get_physical_address() to yield
target_ulong types, e.g. 0xffffffff80000000 instead of 0x80000000. This
is in preparation for enabling emulation of MIPS KVM T&E segments in TCG
MIPS targets, which unlike KVM could potentially have 64-bit
target_ulong. In such a case the offset guest KSEG0 address ends up at
e.g. 0x000000008xxxxxxx instead of 0xffffffff8xxxxxxx.
This also allows the casts to int32_t that force sign extension to be
removed, which removes any confusion due to relational comparison of
unsigned (target_ulong) and signed (int32_t) types.
Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Yongbok Kim <yongbok.kim@imgtec.com>
Cc: Aurelien Jarno <aurelien@aurel32.net>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: kvm@vger.kernel.org
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Yongbok Kim <yongbok.kim@imgtec.com>
Since commit 9768e2abf7 the initrd is loaded at the end of the low
memory to avoid clash for the kernel relocation when kaslr is used.
However this in turn conflicts with the bootmap memory that the kernel
tries to place after initrd, but in low memory. The bootmap spans the
whole usable physical address space. The machine can have at most 2GiB
of memory, 256MiB of low memory mapped at 0x00000000, and 1792MiB of
high memory mapped at 0x90000000. The biggest bootmap therefore
corresponds to the adresses 0x00000000 -> 0xffffffff, which at 1 bit
per 4kiB page corresponds to 128kiB in memory.
Therefore reserve 128kiB after the initrd.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Tested-by: Yongbok Kim <yongbok.kim@imgtec.com>
Signed-off-by: Yongbok Kim <yongbok.kim@imgtec.com>
Writing to the MIPS DESAVE register (and now the KScratch registers)
will stop translation, supposedly due to risk of execution mode
switches. However these registers are basically RW scratch registers
with no side effects so there is no risk of them triggering execution
mode changes.
Drop the bstate = BS_STOP for these registers for both mtc0 and dmtc0.
Fixes: 7a387fffce ("Add MIPS32R2 instructions, and generally straighten out the instruction decoding. This is also the first percent towards MIPS64 support.")
Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Aurelien Jarno <aurelien@aurel32.net>
Cc: Yongbok Kim <yongbok.kim@imgtec.com>
Reviewed-by: Yongbok Kim <yongbok.kim@imgtec.com>
Signed-off-by: Yongbok Kim <yongbok.kim@imgtec.com>
testchardev2 is not a valid chardev id here. Use testchardev1
instead which has been created with chardev-add right before
the 'chardev-send-break' line.
And while we're at it, add the test-hmp.c file to the MAINTAINERS
file, too.
Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-id: 1501149097-19071-1-git-send-email-thuth@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Just minor fixes for 2.10
-----BEGIN PGP SIGNATURE-----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=IAK0
-----END PGP SIGNATURE-----
Merge remote-tracking branch 'remotes/dgilbert/tags/pull-migration-20170802a' into staging
Migration pull 2017-08-02
Just minor fixes for 2.10
# gpg: Signature made Wed 02 Aug 2017 14:55:21 BST
# gpg: using RSA key 0x0516331EBC5BFDE7
# gpg: Good signature from "Dr. David Alan Gilbert (RH2) <dgilbert@redhat.com>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg: There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 45F5 C71B 4A0C B7FB 977A 9FA9 0516 331E BC5B FDE7
* remotes/dgilbert/tags/pull-migration-20170802a:
io: fix qio_channel_socket_accept err handling
migration: fix comment disorder in RAMState
migration: fix small leaks
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
When accept failed, we should setup errp with the reason. More
importantly, the caller may assume errp be non-NULL when error happens,
and not setting the errp may crash QEMU.
At the same time, move the trace_qio_channel_socket_accept_fail() after
the if check on EINTR. Two reasons:
1. when EINTR happened, it's not really a fault (we should just try
again), so we should not log with an "accept failure".
2. trace_*() functions may overwrite errno, then the old errno will be
missing. We need to either check errno before trace_*() calls, or
reserve the errno.
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <1501666880-10159-3-git-send-email-peterx@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Comments for "migration_dirty_pages" and "bitmap_mutex" are switched.
Fix it.
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <1501666880-10159-2-git-send-email-peterx@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Spotted thanks to valgrind and tests/device-introspect-test:
==11711== 1 bytes in 1 blocks are definitely lost in loss record 6 of 14,537
==11711== at 0x4C2EB6B: malloc (vg_replace_malloc.c:299)
==11711== by 0x1E0CDBD8: g_malloc (gmem.c:94)
==11711== by 0x1E0E696E: g_strdup (gstrfuncs.c:363)
==11711== by 0x695693: migration_instance_init (migration.c:2226)
==11711== by 0x717C4B: object_init_with_type (object.c:344)
==11711== by 0x717E80: object_initialize_with_type (object.c:375)
==11711== by 0x7182EB: object_new_with_type (object.c:483)
==11711== by 0x718328: object_new (object.c:493)
==11711== by 0x4B8A29: qmp_device_list_properties (qmp.c:542)
==11711== by 0x4A9561: qmp_marshal_device_list_properties (qmp-marshal.c:1425)
==11711== by 0x819D4A: do_qmp_dispatch (qmp-dispatch.c:104)
==11711== by 0x819E82: qmp_dispatch (qmp-dispatch.c:131)
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20170801160419.14180-1-marcandre.lureau@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Some fixes all over the place. Notably vhost-user gained a new message
to set endian-ness. Borderline for 2.10 but seems to be the only way to
fix legacy guests. Also pc tests are run on kvm now. Not a fix at all
but doesn't touch qemu itself, so I merged it since I had to run these a
lot and I just got tired of waiting for these to finish.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
-----BEGIN PGP SIGNATURE-----
iQEcBAABAgAGBQJZgPRvAAoJECgfDbjSjVRpUKAH/RE1PohaH/klzEVFLPZUdISN
/1zOdFN0BgVOUxS2udLVdZBUrWBiJWt11fbzASeGtOu3ruTpVfp69C7HUH8z4MOB
cyBWgtcf7qQc9cqE0VGO1kLU+ujSyPlkZonjWVfeCrf298SElIXUNVXcihqVsglA
VAL+tIUFT2Bbc3SH5xGsqHCMAy1hNRkEuF5t7enphOoxm7CdqghDLMfUEW65igYg
XNt3DELfFDQVnwGP8DWSWjgLVAvGAGWGNHQ9QD/FO5mdVrXh1lukV8ZtXf6K2VJc
svR670n3ZbMOkEIOGJYcYUFEKMgJXLNEpQX8WRM+vjnuEpDML6bBROtA8hckRv8=
=9TSs
-----END PGP SIGNATURE-----
Merge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging
pc, acpi, virtio: fixes, test speedup for rc1
Some fixes all over the place. Notably vhost-user gained a new message
to set endian-ness. Borderline for 2.10 but seems to be the only way to
fix legacy guests. Also pc tests are run on kvm now. Not a fix at all
but doesn't touch qemu itself, so I merged it since I had to run these a
lot and I just got tired of waiting for these to finish.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# gpg: Signature made Tue 01 Aug 2017 22:36:47 BST
# gpg: using RSA key 0x281F0DB8D28D5469
# gpg: Good signature from "Michael S. Tsirkin <mst@kernel.org>"
# gpg: aka "Michael S. Tsirkin <mst@redhat.com>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg: There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 0270 606B 6F3C DF3D 0B17 0970 C350 3912 AFBE 8E67
# Subkey fingerprint: 5D09 FD08 71C8 F85B 94CA 8A0D 281F 0DB8 D28D 5469
* remotes/mst/tags/for_upstream:
pc: acpi: force FADT rev1 for 440fx based machine types
pc: make 'pc.rom' readonly when machine has PCI enabled
vhost-user: fix watcher need be removed when vhost-user hotplug
tests/bios-tables-test: Compiler warning fix
accel: cleanup error output
intel_iommu: use access_flags for iotlb
intel_iommu: fix iova for pt
vhost-user: fix legacy cross-endian configurations
vhost: fix a memory leak
tests: switch pxe and vm gen id tests to use kvm
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
w2k used to boot on QEMU until revision of FADT has
been bumped to rev3
(commit 77af8a2b hw/i386: Use Rev3 FADT (ACPI 2.0) instead of Rev1 to improve guest OS support.)
Keep PC machine at rev1 to remain compatible and Q35
at rev3 where w2k isn't supported anyway so OSX could
run as well.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Tested-by: John Arbuckle <programmingkidx@gmail.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
looking at bios ROM mapping in QEMU it seems that only isapc
(i.e. not PCI enabled machine) requires ROM being mapped as
RW in other cases BIOS is mapped as RO. Do the same for option
ROM 'pc.rom' when machine has PCI enabled.
As useful side-effect pc.rom MemoryRegion stops being
put in vhost memory map (filtered out by vhost_section()),
which reduces number of entries by 1.
Coincidentally it fixes migration failure reported in
"[PATCH V2] vhost: fix a migration failed because of vhost region merge"
where following destination CLI with /sys/module/vhost/parameters/max_mem_regions = 8
export DIMMSCOUNT=6
QEMU -enable-kvm \
-netdev type=tap,id=guest0,vhost=on,script=no,vhostforce \
-device virtio-net-pci,netdev=guest0 \
-m 256,slots=256,maxmem=2G \
`i=0; while [ $i -lt $DIMMSCOUNT ]; do echo \
"-object memory-backend-ram,id=m$i,size=128M \
-device pc-dimm,id=d$i,memdev=m$i"; i=$(($i + 1)); \
done`
will fail to startup with error:
"-device pc-dimm,id=d5,memdev=m5: a used vhost backend has no free memory slots left"
while it's possible to add the 6th DIMM during hotplug
on source.
Issue is caused by the fact that number of entries in vhost map
is bigger on 1 entry, when -device is processed, than
after guest boots up, and that offending entry belongs to
'pc.rom', it's not like vhost intends to do IO in ROM range
so making it RO hides region from vhost and makes number
of entries in vhost memory map at -device/machine_done time
match number of entries after guest boots.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reported-by: Peng Hao <peng.hao2@zte.com.cn>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
"nc" is freed after hotplug vhost-user, but the watcher is not removed.
The QEMU crash when the watcher access the "nc" when socket disconnects.
Program received signal SIGSEGV, Segmentation fault.
#0 object_get_class (obj=obj@entry=0x2) at qom/object.c:750
#1 0x00007f9bb4180da1 in qemu_chr_fe_disconnect (be=<optimized out>) at chardev/char-fe.c:372
#2 0x00007f9bb40d1100 in net_vhost_user_watch (chan=<optimized out>, cond=<optimized out>, opaque=<optimized out>) at net/vhost-user.c:188
#3 0x00007f9baf97f99a in g_main_context_dispatch () from /usr/lib64/libglib-2.0.so.0
#4 0x00007f9bb41d7ebc in glib_pollfds_poll () at util/main-loop.c:213
#5 os_host_main_loop_wait (timeout=<optimized out>) at util/main-loop.c:261
#6 main_loop_wait (nonblocking=nonblocking@entry=0) at util/main-loop.c:515
#7 0x00007f9bb3e266a7 in main_loop () at vl.c:1917
#8 main (argc=<optimized out>, argv=<optimized out>, envp=<optimized out>) at vl.c:4786
Signed-off-by: Yunjian Wang <wangyunjian@huawei.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
gcc 7.1.1 in fedora 26 moans about the:
tables = g_new0(uint32_t, tables_nr)
because it can't convince itself that tables_nr is positive.
This is fallout from g_assert_cmpint no longer necessarily being
no-return; replace it with a plain g_assert.
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Marcel Apfelbaum <marcel@redhat.com>
Only emit "XXX accelerator not found", if there are not
further accelerators listed. eg
accel=kvm:tcg
doesn't print a "KVM accelerator not found" warning
when it falls back to tcg, but a
accel=kvm
prints a warning, since no fallback is given.
Suggested-by: Daniel P. Berrange <berrange@redhat.com>
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Tested-by: Thomas Huth <thuth@redhat.com>
It was cached by read/write separately. Let's merge them.
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
IOMMUTLBEntry.iova is returned incorrectly on one PT path (though mostly
we cannot really trigger this path, even if we do, we are mostly
disgarding this value, so it didn't break anything). Fix it by
converting the VTD_PAGE_MASK into the correct definition
VTD_PAGE_MASK_4K, then remove VTD_PAGE_MASK.
Fixes: b93130 ("intel_iommu: cleanup vtd_{do_}iommu_translate()")
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Currently, vhost-user does not implement any means for notifying the
backend about guest endianess. This commit introduces a new message
called VHOST_USER_SET_VRING_ENDIAN which is analogous to the ioctl()
called VHOST_SET_VRING_ENDIAN used for kernel vhost backends. Such
message is necessary for backends supporting legacy (pre-1.0) virtio
devices running in big-endian guests.
Signed-off-by: Felipe Franciosi <felipe@nutanix.com>
Signed-off-by: Mike Cui <cui@nutanix.com>
vhost exists a call for g_file_get_contents, but not call g_free.
Signed-off-by: Peng Hao<peng.hao2@zte.com.cn>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Speed up tests on host systems with kvm support.
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Thomas Huth <thuth@redhat.com>
Cc: Laurent Vivier <lvivier@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
When skipping implicit nodes in bdrv_block_device_info(), we know that
bs0 is always non-NULL; initially, because it's taken from a BdrvChild
and a BdrvChild never has a NULL bs, and after the first iteration
because implicit nodes always have a backing file.
Remove the NULL check and add an assertion that the implicit node does
indeed have a backing file.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
qemu-iotests 059 left a whole lot of image files behind in the scratch
directory because VMDK creates additional files for extents and cleaning
them up requires the original image intact (it parses qemu-img info
output to find all extent files), but the image overwrote it many times
like it works for all other image formats.
In addition, _use_sample_img overwrites the TEST_IMG variable, causing
new images created afterwards to reuse the name of the sample file
rather than the usual t.IMGFMT.
This patch adds an intermediate _cleanup_test_img after each subtest
that created an image file with additional extent files, and also after
each use of a sample image. _cleanup_test_img is also changed so that it
resets TEST_IMG after a sample image is cleaned up.
Note that this test was failing before this commit and continues to do
so after it. This failure was introduced in commit 9877860 ('block/vmdk:
Report failures in vmdk_read_cid()') and needs to be dealt with
separately.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
qemu-iotests 063 left t.raw.raw1 behind in the scratch directory because
it used the wrong suffix. Make sure to clean it up after completing the
test.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
qemu-iotests 162 left qemu-nbd.pid behind in the scratch directory, and
potentially a file called '42' in the current directory. Make sure to
clean it up after completing the tests.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
qemu-iotests 153 left t.qcow2.c behind in the scratch directory. Make
sure to clean it up after completing the tests.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
qemu-iotests 141 attempted to use brace expansion to remove all images
with a single command. However, for this to work, the braces shouldn't
be quoted.
With this fix, the tests correctly cleans up its scratch images.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
qemu-iotests 074 and 179 left a blkdebug.conf behind in the scratch
directory. Make sure to clean up after completing the tests.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
qemu-iotests 041 left quorum_snapshot.img and target.img behind in the
scratch directory. Make sure to clean up after completing the tests.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
bdrv_open_driver() is called in two places, bdrv_new_open_driver() and
bdrv_open_common(). In the latter, failure cleanup in is in its caller,
bdrv_open_inherit(), which unrefs the bs->file of the failed driver open
if it exists.
Let's move the bs->file cleanup to bdrv_open_driver() to take care of
all callers and do not set bs->drv to NULL unless the driver's open
function failed. When bs is destroyed by removing its last reference, it
calls bdrv_close() which checks bs->drv to perform the needed cleanups
and also call the driver's close function. Since it cleans up options
and opaque we must take care not leave dangling pointers.
The error paths in bdrv_open_driver() are now two:
If open fails, drv->bdrv_close() should not be called. Unref the child
if it exists, free what we allocated and set bs->drv to NULL. Return the
error and let callers free their stuff.
If open succeeds but we fail after, return the error and let callers
unref and delete their bs, while cleaning up their allocations.
Signed-off-by: Manos Pitsidianakis <el13635@mail.ntua.gr>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
In some error paths it is possible to QDECREF a freed dangling
explicit_options, resulting in a heap overflow crash. For example
bdrv_open_inherit()'s fail unrefs it, then calls bdrv_unref which calls
bdrv_close which also unrefs it.
Signed-off-by: Manos Pitsidianakis <el13635@mail.ntua.gr>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The new test 190 ensures we don't regress back to an infinite loop when
measuring the size of a 2T+ qcow2 image. I did not append to test 178,
because that test is also designed to run with format 'raw'; also, this
gives us some coverage of the measure command under the quick group.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We had a bug for multiple releases where dirty-bitmap count was
documented in bytes but reported in sectors; enhance the testsuite
to add coverage of DirtyBitmapInfo to ensure we do not regress again.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Without redirecting qemu's stderr to stdout, _filter_qemu will not apply
to warnings. This results in $QEMU_PROG not being replaced by QEMU_PROG
which is not great if your qemu executable is not called
qemu-system-x86_64 (e.g. qemu-system-i386).
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
On one hand, the _make_test_img invocation for creating the target image
was missing a -u because its backing file is not supposed to exist at
that point.
On the other hand, nobody noticed probably because the backing file is
created later on and _cleanup failed to remove it: The quotation marks
were misplaced so bash tried to delete a file literally called
"$TEST_IMG{,.target}..." instead of performing brace expansion. Thus, the
files stayed around after the first run and qemu-img create did not
complain about a missing backing file on any run but the first.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
In some cases, the guest can observe the wrong ordering of UIP and
interrupts. This can happen if the VCPU exit is timed like this:
iothread VCPU
... wait for interrupt ...
t-100ns read register A
t wake up, take BQL
t+100ns update_in_progress
return false
return UIP=0
trigger interrupt
The interrupt is late; the VCPU expected the falling edge of UIP to
happen after the interrupt. update_in_progress is already trying to
cover this case by latching UIP if the timer is going to fire soon,
and the fix is documented in the commit message for commit 56038ef623
("RTC: Update the RTC clock only when reading it", 2012-09-10). It
cannot be tested with qtest, because its timing of interrupts vs. reads
is exact.
However, the implementation was incorrect because UIP cmos_ioport_read
cleared register A instead of leaving that to rtc_update_timer. Fixing
the implementation of cmos_ioport_read to match the commit message,
however, breaks the "uip-stuck" test case from the previous patch.
To fix it, skip update timer optimizations if UIP has been latched.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
