Peter Maydell [Fri, 29 May 2020 16:41:45 +0000 (17:41 +0100)]
Merge remote-tracking branch 'remotes/stsquad/tags/pull-testing-tcg-plugins-270520-1' into staging
Testing and one plugin fix:
- support alternates for genisoimage to test/vm
- add clang++ to clang tests
- fix record/replay smoke test
- enable more softfloat tests
- better detection of hung gdb
- upgrade aarch64 tcg test x-compile to gcc-10
- fix plugin cpu_index clash vs threads
# gpg: Signature made Wed 27 May 2020 14:29:20 BST
# gpg: using RSA key 6685AE99E75167BCAFC8DF35FBD0DB095A9E2A44
# gpg: Good signature from "Alex Bennée (Master Work Key) <[email protected]>" [full]
# Primary key fingerprint: 6685 AE99 E751 67BC AFC8 DF35 FBD0 DB09 5A9E 2A44
* remotes/stsquad/tags/pull-testing-tcg-plugins-270520-1:
tests/tcg: add new threadcount test
linux-user: properly "unrealize" vCPU object
cpus-common: ensure auto-assigned cpu_indexes don't clash
tests/docker: use a gcc-10 based image for arm64 tests
tests/docker: add debian11 base image
tests/tcg: better detect confused gdb which can't connect
tests/fp: split and audit the conversion tests
tests/fp: enable extf80_le_quite tests
tests/tcg: fix invocation of the memory record/replay tests
travis.yml: Use clang++ in the Clang tests
tests/vm: pass --genisoimage to basevm script
configure: add alternate binary for genisoimage
* remotes/kraxel/tags/vga-20200528-pull-request: (21 commits)
sm501: Remove obsolete changelog and todo comment
sm501: Optimize small overlapping blits
sm501: Replace hand written implementation with pixman where possible
sm501: Clean up local variables in sm501_2d_operation
sm501: Use BIT(x) macro to shorten constant
sm501: Shorten long variable names in sm501_2d_operation
sm501: Convert printf + abort to qemu_log_mask
hw/display/pxa2xx_lcd: Replace printf() call by qemu_log_mask()
hw/display/omap_dss: Replace fprintf() call by qemu_log_mask(LOG_UNIMP)
hw/display/exynos4210_fimd: Use qemu_log_mask(GUEST_ERROR)
hw/display/vmware_vga: Let the PCI device own its I/O MemoryRegion
hw/display/vmware_vga: Replace printf() calls by qemu_log_mask(ERROR)
hw/display/xlnx_dp: Replace disabled DPRINTF() by error_report()
hw/display/dpcd: Convert debug printf()s to trace events
hw/display/dpcd: Fix memory region size
hw/display/cirrus_vga: Convert debug printf() to trace event
hw/display/cirrus_vga: Use qemu_log_mask(ERROR) instead of debug printf
hw/display/cirrus_vga: Use qemu_log_mask(UNIMP) instead of debug printf
hw/display/cirrus_vga: Convert debug printf() to trace event
hw/display/cg3: Convert debug printf()s to trace events
...
Peter Maydell [Thu, 28 May 2020 15:18:06 +0000 (16:18 +0100)]
Merge remote-tracking branch 'remotes/huth-gitlab/tags/pull-request-2020-05-28' into staging
Fixes and improvements for the gitlab-CI:
- Build with other images instead of the broken Debian containers
- Fix building with the latest version of Clang (at least wrt. to
the gitlab-CI pipeline)
- Add Philippe, Alex and Wainer to the Gitlab-CI section in MAINTAINERS
* remotes/huth-gitlab/tags/pull-request-2020-05-28:
gitlab-ci: Determine the number of jobs dynamically
gitlab-ci: Do not use the standard container images from gitlab
gitlab-ci: Move edk2 and opensbi YAML files to .gitlab-ci.d folder
GitLab CI: avoid calling before_scripts on unintended jobs
gitlab-ci: Remove flex/bison packages
MAINTAINERS: Add Philippe, Alex and Wainer to the Gitlab-CI section
linux-user: limit check to HOST_LONG_BITS < TARGET_ABI_BITS
BALATON Zoltan [Thu, 21 May 2020 19:39:44 +0000 (21:39 +0200)]
sm501: Optimize small overlapping blits
AmigaOS tends to do a lot of small blits (even 1 pixel). Avoid malloc
overhead by keeping around a buffer for this and only alloc when
blitting larger areas.
BALATON Zoltan [Thu, 21 May 2020 19:39:44 +0000 (21:39 +0200)]
sm501: Convert printf + abort to qemu_log_mask
Some places already use qemu_log_mask() to log unimplemented features
or errors but some others have printf() then abort(). Convert these to
qemu_log_mask() and avoid aborting to prevent guests to easily cause
denial of service.
Thomas Huth [Thu, 21 May 2020 04:50:10 +0000 (06:50 +0200)]
gitlab-ci: Determine the number of jobs dynamically
Some people might want to run the gitlab CI pipelines in an environment
where multiple CPUs are available to the runners, so let's rather get
the number for "-j" from the "nproc" program (increased by 1 to compensate
for jobs that wait for I/O) instead of hard-coding it.
Thomas Huth [Wed, 20 May 2020 08:38:37 +0000 (10:38 +0200)]
gitlab-ci: Do not use the standard container images from gitlab
Currently all pipelines of the gitlab CI are failing, except for the
"build-user" pipeline. There is an issue with the default container
image (likely Debian stable) where they imported something bad in one
of the system headers:
/usr/include/linux/swab.h: In function '__swab':
/builds/huth/qemu/include/qemu/bitops.h:20:34: error: "sizeof" is not
defined, evaluates to 0 [-Werror=undef]
#define BITS_PER_LONG (sizeof (unsigned long) * BITS_PER_BYTE)
We could maybe work-around this issue or wait for the default containers
to get fixed, but considering that we use Ubuntu (and thus Debian-style)
CI in Travis already to a very large extent, we should consider to use
some RPM-based distros in our gitlab CI instead. Thus let's change the
failing pipelines to use Fedora and CentOS (and also one Ubuntu 19.10,
since 20.04 is broken, too) now.
Cleber Rosa [Thu, 12 Mar 2020 19:36:14 +0000 (15:36 -0400)]
GitLab CI: avoid calling before_scripts on unintended jobs
At this point it seems that all jobs depend on those steps, with
maybe the EDK2 jobs as exceptions.
The jobs that will be added later will not want those scripts to be
run, so let's move these steps to the appropriate jobs, while
still trying to avoid repetition.
Thomas Huth [Mon, 10 Feb 2020 15:41:04 +0000 (16:41 +0100)]
MAINTAINERS: Add Philippe, Alex and Wainer to the Gitlab-CI section
Initially, I was the only one who was using Gitlab while most developers
had their git trees still on other systems, but that has changed nowadays.
There is now much more interest in the Gitlab-CI today, so it would be
good to have more than only one maintainer / reviewer for the gitlab-ci.yml
file. Alex, Wainer and Philippe kindly offered their help here, so let's
add them to the corresponding section in the MAINTAINERS file now.
Peter Maydell [Wed, 27 May 2020 21:23:13 +0000 (22:23 +0100)]
Merge remote-tracking branch 'remotes/dgibson/tags/ppc-for-5.1-20200527' into staging
ppc patch queue 2020-05-27
Here's the next pull request for qemu-5.1. It includes:
* Support for the scv and rfscv POWER9 instructions in TCG
* Support for the new SPAPR_LMB_FLAGS_HOTREMOVABLE flag, which
provides a way for guests to know memory which should be removable
(so the guest can avoid putting immovable allocations there).
* Some fixes for the recently added partition scope radix translation
in softmmu
* Assorted minor fixes and cleanups
It includes one patch to avoid a clash with SELinux when using NVLink
VFIO devices. That's not technically within the files under my
maintainership, but it is in a section of the VFIO quirks code that's
specific to the POWER-only NVLink devices, and has an ack from Alex
Williamson.
* remotes/dgibson/tags/ppc-for-5.1-20200527:
vfio/nvlink: Remove exec permission to avoid SELinux AVCs
target/ppc: Fix argument to ppc_radix64_partition_scoped_xlate() again
hw/nvram/mac_nvram: Convert debug printf()s to trace events
hw/pci-bridge/dec: Remove dead debug code
target/ppc: Don't update radix PTE R/C bits with gdbstub
target/ppc: Fix arguments to ppc_radix64_partition_scoped_xlate()
target/ppc: Add missing braces in ppc_radix64_partition_scoped_xlate()
target/ppc: Don't initialize some local variables in ppc_radix64_xlate()
target/ppc: Pass const pointer to ppc_radix64_get_fully_qualified_addr()
target/ppc: Pass const pointer to ppc_radix64_get_prot_amr()
ppc/spapr: Add hotremovable flag on DIMM LMBs on drmem_v2
target/ppc: Add support for scv and rfscv instructions
target/ppc: Untabify excp_helper.c
ppc/spapr: add a POWER10 CPU model
ppc/pnv: Fix NMI system reset SRR1 value
hw/registerfields: Prefix local variables with underscore in macros
One can name a local variable holding a value as 'v', but it
currently clashes with the registerfields macros. To save others
to debug the same mistake, prefix the macro's local variables
with an underscore.
Alex Bennée [Wed, 20 May 2020 14:05:39 +0000 (15:05 +0100)]
linux-user: properly "unrealize" vCPU object
We shouldn't be messing around with the CPU list in linux-user save
for the very special case of do_fork(). When threads end we need to
properly follow QOM object lifetime handling and allow the eventual
cpu_common_unrealizefn to both remove the CPU and ensure any clean-up
actions are taken place, for example calling plugin exit hooks.
There is still a race condition to avoid so use the linux-user
specific clone_lock instead of the cpu_list_lock to avoid it.
Basing the cpu_index on the number of currently allocated vCPUs fails
when vCPUs aren't removed in a LIFO manner. This is especially true
when we are allocating a cpu_index for each guest thread in
linux-user where there is no ordering constraint on their allocation
and de-allocation.
[I've dropped the assert which is there to guard against out-of-order
removal as this should probably be caught higher up the stack. Maybe
we could just ifdef CONFIG_SOFTTMU it?]
Alex Bennée [Wed, 20 May 2020 14:05:36 +0000 (15:05 +0100)]
tests/docker: use a gcc-10 based image for arm64 tests
As we enable newer features that we want to test on arm64 targets we
need newer compilers. Split off a new debian-arm64-test-cross image
which we can use to build these new tests.
Alex Bennée [Wed, 20 May 2020 14:05:32 +0000 (15:05 +0100)]
tests/fp: split and audit the conversion tests
Split the float conversion tests into separate groups and audit the
tests to check what is still broken. I was able to enable a bunch of
tests that had been missed before:
all the float to float conversions
ui32_to_extF80
ui64_to_extF80
extF80_to_ui32
extF80_to_ui32_r_minMag
extF80_to_ui64
extF80_to_ui64_r_minMag
Thomas Huth [Wed, 20 May 2020 14:05:29 +0000 (15:05 +0100)]
travis.yml: Use clang++ in the Clang tests
Our configure script does not look for clang++ automatically, so we
should use --cxx=clang++ to make sure that we test our C++ code with
Clang, too. And while we're at it, also use --host-cc=clang here
to avoid that we use the normal "cc" as host C compiler.
Alex Bennée [Tue, 19 May 2020 13:22:49 +0000 (09:22 -0400)]
tests/vm: pass --genisoimage to basevm script
If we have an alternative to genisoimage we really need to tell the
script about it as well so it can use it. It will still default to
genisoimage in case it is run outside our build machinery.
Alex Bennée [Tue, 19 May 2020 13:22:48 +0000 (09:22 -0400)]
configure: add alternate binary for genisoimage
Not all distros ship genisoimage which is a Debian fork from the
original cdrtools. As the options are pretty much the same support it
as a fallback binary.
i386: Fix x86_cpu_load_model() error API violation
The Error ** argument must be NULL, &error_abort, &error_fatal, or a
pointer to a variable containing NULL. Passing an argument of the
latter kind twice without clearing it in between is wrong: if the
first call sets an error, it no longer points to NULL for the second
call.
x86_cpu_load_model() is wrong that way. Harmless, because its @errp
is always &error_abort. To fix, cut out the @errp middleman.
The Error ** argument must be NULL, &error_abort, &error_fatal, or a
pointer to a variable containing NULL. Passing an argument of the
latter kind twice without clearing it in between is wrong: if the
first call sets an error, it no longer points to NULL for the second
call.
boston_mach_init() is wrong that way. The last calls treats an error
as fatal. Do that for the prior ones, too.
The Error ** argument must be NULL, &error_abort, &error_fatal, or a
pointer to a variable containing NULL. Passing an argument of the
latter kind twice without clearing it in between is wrong: if the
first call sets an error, it no longer points to NULL for the second
create_cps() is wrong that way. The last calls treats an error as
fatal. Do that for the prior ones, too.
migrate_get_socket_address() neglects to check
visit_type_SocketAddressList() failure. This smells like a leak, but
it actually will crash dereferencing @addrs. Pass &error_abort to
remove the code smell.
s390x/cpumodel: Fix harmless misuse of visit_check_struct()
Commit e47970f51d "s390x/cpumodel: Fix query-cpu-model-FOO error API
violations" neglected to change visit_check_struct()'s Error **
argument along with the others. If visit_check_struct() failed, we'd
take the success path. Fortunately, it can't fail here:
qobject_input_check_struct() checks we consumed the whole dictionary,
and to get here, we did. Fix it anyway.
Leonardo Bras [Fri, 1 May 2020 05:54:49 +0000 (02:54 -0300)]
vfio/nvlink: Remove exec permission to avoid SELinux AVCs
If SELinux is setup without 'execmem' permission for qemu, all mmap
with (PROT_WRITE | PROT_EXEC) will fail and print a warning in
SELinux log.
If "nvlink2-mr" memory allocation fails (fist diff), it will cause
guest NUMA nodes to not be correctly configured (V100 memory will
not be visible for guest, nor its NUMA nodes).
Not having 'execmem' permission is intesting for virtual machines to
avoid buffer-overflow based attacks, and it's adopted in distros
like RHEL.
So, removing the PROT_EXEC flag seems the right thing to do.
Browsing some other code that mmaps memory for usage with
memory_region_init_ram_device_ptr, I could notice it's usual to
not have PROT_EXEC (only PROT_READ | PROT_WRITE), so it should be
no problem around this.
Greg Kurz [Wed, 13 May 2020 22:57:19 +0000 (00:57 +0200)]
target/ppc: Don't update radix PTE R/C bits with gdbstub
gdbstub shouldn't silently change guest visible state when doing address
translation. Since the R/C bits can only be updated when handling a MMU
fault, let's reuse the cause_excp flag and rename it to guest_visible.
While here drop a not very useful comment.
This was found while reading the code. I could verify that this affects
both powernv and pseries, but I failed to observe any actual bug.
Greg Kurz [Wed, 13 May 2020 22:57:00 +0000 (00:57 +0200)]
target/ppc: Don't initialize some local variables in ppc_radix64_xlate()
It is the job of the ppc_radix64_get_fully_qualified_addr() function
which is called at the beginning of ppc_radix64_xlate() to set both
lpid *and* pid. It doesn't buy us anything to initialize them first.
Worse, a bug in ppc_radix64_get_fully_qualified_addr(), eg. failing to
set either lpid or pid, would be undetectable by static analysis tools
like coverity.
Some recent versions of gcc (eg. gcc-9.3.1-2.fc30) may still think
that lpid or pid is used uninitialized though, so this also adds
default cases in the switch statements to make it clear this cannot
happen.
Leonardo Bras [Mon, 11 May 2020 20:02:02 +0000 (17:02 -0300)]
ppc/spapr: Add hotremovable flag on DIMM LMBs on drmem_v2
On reboot, all memory that was previously added using object_add and
device_add is placed in this DIMM area.
The new SPAPR_LMB_FLAGS_HOTREMOVABLE flag helps Linux to put this memory in
the correct memory zone, so no unmovable allocations are made there,
allowing the object to be easily hot-removed by device_del and
object_del.
This new flag was accepted in Power Architecture documentation.
Nicholas Piggin [Thu, 7 May 2020 11:53:28 +0000 (21:53 +1000)]
target/ppc: Add support for scv and rfscv instructions
POWER9 adds scv and rfscv instructions and the system call vectored
interrupt. Linux does not support this instruction yet but it has
been tested with a modified kernel that runs on real hardware.
Peter Maydell [Tue, 26 May 2020 19:25:06 +0000 (20:25 +0100)]
Merge remote-tracking branch 'remotes/philmd-gitlab/tags/mips-hw-next-20200526' into staging
MIPS hardware updates
- MAINTAINERS updated to welcome Huacai Chen and Jiaxun Yang,
and update Aleksandar Rikalo's email address,
- Trivial improvements in the Bonito64 North Bridge and the
Fuloong 2e machine,
- MIPS Machines names unified without 'mips_' prefix.
# gpg: Signature made Tue 26 May 2020 14:32:08 BST
# gpg: using RSA key FAABE75E12917221DCFD6BB2E3E32C2CDEADC0DE
# gpg: Good signature from "Philippe Mathieu-Daudé (F4BUG) <[email protected]>" [full]
# Primary key fingerprint: FAAB E75E 1291 7221 DCFD 6BB2 E3E3 2C2C DEAD C0DE
* remotes/philmd-gitlab/tags/mips-hw-next-20200526:
MAINTAINERS: Change Aleksandar Rikalo's email address
hw/mips/mips_int: De-duplicate KVM interrupt delivery
hw/mips/malta: Add some logging for bad register offset cases
hw/mips: Rename malta/mipssim/r4k/jazz files
hw/mips/fuloong2e: Fix typo in Fuloong machine name
hw/mips/fuloong2e: Move code and update a comment
hw/pci-host/bonito: Set the Config register reset value with FIELD_DP32
hw/pci-host/bonito: Better describe the I/O CS regions
hw/pci-host/bonito: Map the different PCI ranges more detailed
hw/pci-host/bonito: Map all the Bonito64 I/O range
hw/pci-host/bonito: Map peripheral using physical address
hw/pci-host/bonito: Fix DPRINTF() format strings
hw/pci-host: Use CONFIG_PCI_BONITO to select the Bonito North Bridge
MAINTAINERS: Add Huacai Chen as fuloong2e co-maintainer
Peter Maydell [Tue, 26 May 2020 13:05:53 +0000 (14:05 +0100)]
Merge remote-tracking branch 'remotes/gkurz/tags/9p-next-2020-05-26' into staging
- fix build with musl libc
- fix potential deadlock of QEMU main event loop (cannot be hit with linux
client)
- revert 9pfs reply truncation (LP 1877688)
- xen backend waits for client to free space on the reply ring instead of
truncating or disconnecting
# gpg: Signature made Tue 26 May 2020 10:36:23 BST
# gpg: using RSA key B4828BAF943140CEF2A3491071D4D5E5822F73D6
# gpg: Good signature from "Greg Kurz <[email protected]>" [full]
# gpg: aka "Gregory Kurz <[email protected]>" [full]
# gpg: aka "[jpeg image of size 3330]" [full]
# Primary key fingerprint: B482 8BAF 9431 40CE F2A3 4910 71D4 D5E5 822F 73D6
* remotes/gkurz/tags/9p-next-2020-05-26:
xen/9pfs: increase max ring order to 9
xen/9pfs: yield when there isn't enough room on the ring
Revert "9p: init_in_iov_from_pdu can truncate the size"
9p: Lock directory streams with a CoMutex
9pfs: include linux/limits.h for XATTR_SIZE_MAX
hw/pci-host/bonito: Map all the Bonito64 I/O range
To ease following guest accesses to the Bonito64 chipset,
map its I/O range as UnimplementedDevice.
We can now see the accesses to unimplemented peripheral
using the '-d unimp' command line option.
MAINTAINERS: Add Huacai Chen as fuloong2e co-maintainer
I submitted the MIPS/fuloong2e support about ten years ago, and
after that I became a MIPS kernel developer. Last year, Philippe
Mathieu- Daudé asked me that whether I can be a reviewer of
MIPS/fuloong2e, and I promised that I will do some QEMU work in
the next year (i.e., 2020 and later). I think now (and also in
future) I can have some spare time, so I can finally do some real
work on QEMU/MIPS. And if possible, I hope I can be a co-maintainer
of MIPS/fuloong2e.
* remotes/kraxel/tags/audio-20200526-pull-request:
hw/mips/mips_fulong2e: Remove unused 'audio/audio.h' include
audio: Let capture_callback handler use const buffer argument
audio: Let audio_sample_to_uint64() use const samples argument
audio: fix wavcapture segfault
audio/mixeng: fix clang 10+ warning
audio/jack: add JACK client audiodev
hw/audio/gus: Use AUDIO_HOST_ENDIANNESS definition from 'audio/audio.h'
es1370: check total frame count against current frame
Bruce Rogers [Thu, 21 May 2020 17:29:31 +0000 (11:29 -0600)]
audio: fix wavcapture segfault
Commit 571a8c522e caused the HMP wavcapture command to segfault when
processing audio data in audio_pcm_sw_write(), where a NULL
sw->hw->pcm_ops is dereferenced. This fix checks that the pointer is
valid before dereferincing it. A similar fix is also made in the
parallel function audio_pcm_sw_read().
Volker Rümelin [Sat, 23 May 2020 20:17:12 +0000 (22:17 +0200)]
audio/mixeng: fix clang 10+ warning
The code in CONV_NATURAL_FLOAT() and CLIP_NATURAL_FLOAT()
seems to use the constant 2^31-0.5 to convert float to integer
and back. But the float type lacks the required precision and
the constant used for the conversion is 2^31. This is equiva-
lent to a [-1.f, 1.f] <-> [INT32_MIN, INT32_MAX + 1] mapping.
This patch explicitly writes down the used constant. The
compiler generated code doesn't change.
The constant 2^31 has an exact float representation and the
clang 10 compiler stops complaining about an implicit int to
float conversion with a changed value.
A few notes:
- The conversion of 1.f to INT32_MAX + 1 doesn't overflow. The
type of the destination variable is int64_t.
- At a later stage one of the clip_* functions in
audio/mixeng_template.h limits INT32_MAX + 1 to the integer
range.
- The clip_natural_float_* functions in audio/mixeng.c convert
INT32_MAX and INT32_MAX + 1 to 1.f.
xen/9pfs: yield when there isn't enough room on the ring
Instead of truncating replies, which is problematic, wait until the
client reads more data and frees bytes on the reply ring.
Do that by calling qemu_coroutine_yield(). The corresponding
qemu_coroutine_enter_if_inactive() is called from xen_9pfs_bh upon
receiving the next notification from the client.
We need to be careful to avoid races in case xen_9pfs_bh and the
coroutine are both active at the same time. In xen_9pfs_bh, wait until
either the critical section is over (ring->co == NULL) or until the
coroutine becomes inactive (qemu_coroutine_yield() was called) before
continuing. Then, simply wake up the coroutine if it is inactive.
projects / qemu.git / log