When debugging QEMU it is often useful to put a breakpoint on the
error_setg_internal method impl.
Unfortunately the object_property_add / object_class_property_add
methods call object_property_find / object_class_property_find methods
to check if a property exists already before adding the new property.
As a result there are a huge number of calls to error_setg_internal
on startup of most QEMU commands, making it very painful to set a
breakpoint on this method.
Most callers of object_find_property and object_class_find_property,
however, pass in a NULL for the Error parameter. This simplifies the
methods to remove the Error parameter entirely, and then adds some
new wrapper methods that are able to raise an Error when needed.
* remotes/ehabkost/tags/machine-next-pull-request:
Use OBJECT_DECLARE_SIMPLE_TYPE when possible
Use OBJECT_DECLARE_TYPE when possible
qom: Remove module_obj_name parameter from OBJECT_DECLARE* macros
qom: Remove ParentClassType argument from OBJECT_DECLARE_SIMPLE_TYPE
scripts/codeconverter: Update to latest version
target/s390x: Set instance_align on S390CPU TypeInfo
target/riscv: Set instance_align on RISCVCPU TypeInfo
target/ppc: Set instance_align on PowerPCCPU TypeInfo
target/arm: Set instance_align on CPUARM TypeInfo
qom: Allow objects to be allocated with increased alignment
qom: Correct error values in two contracts
qom: Clean up object_property_get_enum()'s error value
qom: Correct object_class_dynamic_cast_assert() documentation
sifive: Use DECLARE_*CHECKER* macros
sifive: Move QOM typedefs and add missing includes
sifive_u: Rename memmap enum constants
sifive_e: Rename memmap enum constants
Peter Maydell [Mon, 21 Sep 2020 16:41:32 +0000 (17:41 +0100)]
Merge remote-tracking branch 'remotes/ehabkost/tags/x86-next-pull-request' into staging
x86 queue, 2020-09-18
Cleanups:
* Correct the meaning of '0xffffffff' value for hv-spinlocks (Vitaly Kuznetsov)
* vmport: Drop superfluous parenthesis (Philippe Mathieu-Daudé)
Fixes:
* Use generic APIC ID encoding code for EPYC (Babu Moger)
* remotes/ehabkost/tags/x86-next-pull-request:
i386: Simplify CPUID_8000_001E for AMD
i386: Simplify CPUID_8000_001d for AMD
hw/i386/vmport: Drop superfluous parenthesis around function typedef
i386/kvm: correct the meaning of '0xffffffff' value for hv-spinlocks
qom: Remove module_obj_name parameter from OBJECT_DECLARE* macros
One of the goals of having less boilerplate on QOM declarations
is to avoid human error. Requiring an extra argument that is
never used is an opportunity for mistakes.
Remove the unused argument from OBJECT_DECLARE_TYPE and
OBJECT_DECLARE_SIMPLE_TYPE.
Coccinelle patch used to convert all users of the macros:
qom: Remove ParentClassType argument from OBJECT_DECLARE_SIMPLE_TYPE
The requirement to specify the parent class type makes the macro
harder to use and easy to misuse (silent bugs can be introduced
if the wrong struct type is specified).
Simplify the macro by just not declaring any class struct,
allowing us to remove the class_size field from the TypeInfo
variables for those types.
I'm not documenting every single change in the codeconverter
script because most of that code will be deleted once we finish
the QOM code conversion. This patch updates the script to the
latest version that was used to perform changes in the QOM code.
analyze-migration.py: fix read_migration_debug_json() return type
Since we use result of read_migration_debug_json() as JSON formatted string,
we must provide proper type. Before Python 3.6 json.loads() method
support only str typed input.
The script 'bench_write_req.py' allows comparing performances of write
request for two qemu-img binary files.
An example with (qemu-img binary 1) and without (qemu-img binary 2) the
applied patch "qcow2: skip writing zero buffers to empty COW areas"
(git commit ID: c8bb23cbdbe32f5) has the following results:
qom: Allow objects to be allocated with increased alignment
It turns out that some hosts have a default malloc alignment less
than that required for vectors.
We assume that, with compiler annotation on CPUArchState, that we
can properly align the vector portion of the guest state. Fix the
alignment of the allocation by using qemu_memalloc when required.
apic_id contains all the information required to build
CPUID_8000_001E. core_id and node_id is already part of
apic_id generated by x86_topo_ids_from_apicid.
Also remove the restriction on number bits on core_id and
node_id.
Remove all the hardcoded values and replace with generalized
fields.
Refer the Processor Programming Reference (PPR) documentation
available from the bugzilla Link below.
Vitaly Kuznetsov [Fri, 15 May 2020 11:48:47 +0000 (13:48 +0200)]
i386/kvm: correct the meaning of '0xffffffff' value for hv-spinlocks
Hyper-V TLFS prior to version 6.0 had a mistake in it: special value
'0xffffffff' for CPUID 0x40000004.EBX was called 'never to retry', this
looked weird (like why it's not '0' which supposedly have the same effect?)
but nobody raised the question. In TLFS version 6.0 the mistake was
corrected to 'never notify' which sounds logical. Fix QEMU accordingly.
object_property_get_bool()'s contract claims it returns NULL on error.
Pasto; it returns false.
object_property_get_int()'s contract claims it returns "negative". It
actually returns -1. All the other object_property_get_FOO()
contracts specify the exact error value, so do the same here.
qom: Clean up object_property_get_enum()'s error value
object_property_get_enum() is the only object_property_FOO() that is
documented to return an undefined value on error. It does no such
thing, actually: it returns 0 on some errors, and -1 on others.
Needlessly complicated. Always return -1 on error, and adjust the
contract.
Eduardo Habkost [Mon, 31 Aug 2020 21:07:30 +0000 (17:07 -0400)]
sifive: Move QOM typedefs and add missing includes
Some typedefs and macros are defined after the type check macros.
This makes it difficult to automatically replace their
definitions with OBJECT_DECLARE_TYPE.
Some of the enum constant names conflict with the QOM type check
macros (SIFIVE_U_OTP, SIFIVE_U_PRCI). This needs to be addressed
to allow us to transform the QOM type check macros into functions
generated by OBJECT_DECLARE_TYPE().
Rename all the constants to SIFIVE_U_DEV_*, to avoid conflicts.
Some of the enum constant names conflict with a QOM type check
macro (SIFIVE_E_PRCI). This needs to be addressed to allow us to
transform the QOM type check macros into functions generated by
OBJECT_DECLARE_TYPE().
Rename all the constants to SIFIVE_E_DEV_*, to avoid conflicts.
tests/acceptance: Skip slow quanta-gsj U-boot+Linux test
The 'arm_quanta_gsj_initrd' test is timeouting on GitLab CI:
https://gitlab.com/philmd/qemu/-/jobs/745483978#L846
and also sometimes on my workstation, so proceed as with
the other slow tests: do not run it by default.
The test can still be run setting the AVOCADO_TIMEOUT_EXPECTED
environment variable.
tests/acceptance: Disable tests dependent of unreliable apt.armbian.com
Armbian servers are not very reliable and confused the GitLab CI
users a few times this month (path updated, archives moved, and
now the SSL: CERTIFICATE_VERIFY_FAILED "certificate has expired"
error). Time to disable these tests.
Users can still use the artifacts from the cache (or manually add
them to the cache).
Peter Maydell [Fri, 18 Sep 2020 15:34:26 +0000 (16:34 +0100)]
Merge remote-tracking branch 'remotes/philmd-gitlab/tags/fw_cfg-20200918' into staging
fw_cfg patches
Fixes a bug in the recently introduced fw_cfg_add_from_generator().
CI jobs results:
. https://gitlab.com/philmd/qemu/-/pipelines/191714875
. https://travis-ci.org/github/philmd/qemu/builds/728288784
. https://app.shippable.com/github/philmd/qemu/runs/869/1/console
# gpg: Signature made Fri 18 Sep 2020 16:20:56 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
The documentation on g_byte_array_free()
<https://developer.gnome.org/glib/stable/glib-Byte-Arrays.html#g-byte-array-free>
says:
> Returns
>
> the element data if free_segment is FALSE, otherwise NULL. The element
> data should be freed using g_free().
Because we currently call g_byte_array_free() with free_segment=TRUE, we
end up passing data=NULL to fw_cfg_add_file().
On the plus side, fw_cfg_data_read() and fw_cfg_dma_transfer() both deal
with NULL data gracefully: QEMU does not crash when the guest reads such
an item, the guest just gets a properly sized, but zero-filled blob.
However, the bug breaks UEFI HTTPS boot, as the IANA_TLS_CIPHER array,
generated otherwise correctly by the "tls-cipher-suites" object, is in
effect replaced with a zero blob.
Fix the issue by passing free_segment=FALSE to g_byte_array_free():
- the caller (fw_cfg_add_from_generator()) temporarily assumes ownership
of the generated byte array,
- then ownership of the byte array is transfered to fw_cfg, as
fw_cfg_add_file() links (not copies) "data" into fw_cfg.
Peter Maydell [Fri, 18 Sep 2020 12:36:42 +0000 (13:36 +0100)]
Merge remote-tracking branch 'remotes/legoater/tags/pull-aspeed-20200918' into staging
Aspeed patches :
* Couple of cleanups
* New machine properties to define the flash models
# gpg: Signature made Fri 18 Sep 2020 08:23:19 BST
# gpg: using RSA key A0F66548F04895EBFE6B0B6051A343C7CFFBECA1
# gpg: Good signature from "Cédric Le Goater <[email protected]>" [undefined]
# 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: A0F6 6548 F048 95EB FE6B 0B60 51A3 43C7 CFFB ECA1
* remotes/legoater/tags/pull-aspeed-20200918:
misc: aspeed_scu: Update AST2600 silicon id register
hw/arm/aspeed: Add machine properties to define the flash models
hw/arm/aspeed: Map the UART5 device unconditionally
The SCU004 (silicon id 1) value matches SCU014 for A0, but for
subsequent revisions it is hard coded to the A1 value.
Qemu effectively dropped support for the A0 in 7582591ae745 ("aspeed:
Support AST2600A1 silicon revision") as the A0 reset table was removed,
so it makes sense to only support the behaviour of A1 and onwards.
hw/arm/aspeed: Add machine properties to define the flash models
Some machines don't have much differences a part from the flash model
being used. Introduce new machine properties to change them from the
command line.
For instance, to start the ast2500-evb machine with a different FMC
chip and a 64M SPI chip, use :
* remotes/kraxel/tags/microvm-20200917-pull-request: (21 commits)
microvm: enable ramfb
tests/acpi: update expected data files for microvm
tests/acpi: add microvm test
tests/acpi: allow override blkdev
tests/acpi: allow microvm test data updates.
microvm: wire up hotplug
x86: move cpu hotplug from pc to x86
x86: move acpi_dev from pc/microvm
x86: constify x86_machine_is_*_enabled
microvm/acpi: disable virtio-mmio cmdline hack
microvm/acpi: use seabios with acpi=on
microvm/acpi: use GSI 16-23 for virtio
microvm/acpi: add acpi_dsdt_add_virtio() for x86
microvm/acpi: add minimal acpi support
microvm: make virtio irq base runtime configurable
acpi: move acpi_dsdt_add_power_button() to ged
acpi: ged: add x86 device variant.
acpi: ged: add control regs
seabios: add bios-microvm.bin binary
seabios: add microvm config, update build rules
...
I found that there are many spelling errors in the comments of qemu,
so I used the spellcheck tool to check the spelling errors
and finally found some spelling errors in the contrib folder.
I found that there are many spelling errors in the comments of qemu,
so I used the spellcheck tool to check the spelling errors
and finally found some spelling errors in the qapi folder.
I found that there are many spelling errors in the comments of qemu,
so I used the spellcheck tool to check the spelling errors
and finally found some spelling errors in the disas folder.
I found that there are many spelling errors in the comments of qemu,
so I used the spellcheck tool to check the spelling errors
and finally found some spelling errors in the linux-user folder.
I found that there are many spelling errors in the comments of qemu,
so I used the spellcheck tool to check the spelling errors
and finally found some spelling errors in the util folder.
I found that there are many spelling errors in the comments of qemu,
so I used the spellcheck tool to check the spelling errors
and finally found some spelling errors in the scripts folder.
I found that there are many spelling errors in the comments of qemu,
so I used the spellcheck tool to check the spelling errors
and finally found some spelling errors in the docs folder.
I found that there are many spelling errors in the comments of qemu,
so I used the spellcheck tool to check the spelling errors
and finally found some spelling errors in the migration folder.
I found that there are many spelling errors in the comments of qemu,
so I used the spellcheck tool to check the spelling errors
and finally found some spelling errors in the folder.
Peter Maydell [Thu, 17 Sep 2020 13:50:55 +0000 (14:50 +0100)]
Merge remote-tracking branch 'remotes/stsquad/tags/pull-configure-fixes-160920-1' into staging
configure tweaks for deprecation
- iotest fix for readlink -f
- linux-user, report rather than assert on mmap failure
- clean-up and re-factor the logic
- add tilegx-linux-user to deprecated_targets_list
- add [lm32|unicore32]-softmmu deprecated_targets_list
- add a gitlab deprecated builds test
# gpg: Signature made Wed 16 Sep 2020 10:11:41 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-configure-fixes-160920-1:
configure: add [lm32|unicore32]-softmmu to deprecation logic
gitlab: create a build-deprecated target
configure: include tilegx-linux-user in the deprecation logic
configure: clean-up the target-list-exclude logic
configure: also skip deprecated targets with target-list-exclude
configure: move deprecated feature processing to supported_target
iotests: Drop readlink -f
linux-user: test, don't assert addr != test in pgb_reserved_va
Peter Maydell [Thu, 17 Sep 2020 12:38:08 +0000 (13:38 +0100)]
Merge remote-tracking branch 'remotes/huth-gitlab/tags/pull-request-2020-09-16' into staging
* Fix "readlink -f" problem in iotests on macOS (to fix the Cirrus-CI tests)
* Some minor qtest improvements
* Fix the unit tests to work on MSYS2, too
* Enable building and testing on MSYS2 in the Cirrus-CI
* Build FreeBSD with one task again in the Cirrus-CI
* remotes/huth-gitlab/tags/pull-request-2020-09-16: (24 commits)
cirrus: Building freebsd in a single shot
ci: Enable msys2 ci in cirrus
tests: Fixes test-qdev-global-props.c
tests: fix test-util-sockets.c
tests: Fixes test-io-channel-file by mask only owner file state mask bits
tests: fixes aio-win32 about aio_remove_fd_handler, get it consistence with aio-posix.c
tests: Fixes test-io-channel-socket.c tests under msys2/mingw
vmstate: Fixes test-vmstate.c on msys2/mingw
meson: remove empty else and duplicated gio deps
meson: Use -b to ignore CR vs. CR-LF issues on Windows
osdep: file locking functions are not available on Win32
tests: test-replication disable /replication/secondary/* on msys2/mingw.
tests: Fixes test-replication.c on msys2/mingw.
meson: disable crypto tests are empty under win32
meson: Disable test-char on msys2/mingw for fixing tests stuck
rcu: fixes test-logging.c by call drain_call_rcu before rmdir_full
tests: Convert g_free to g_autofree macro in test-logging.c
rcu: Implement drain_call_rcu
qga/commands-win32: Fix problem with redundant protype declaration
Simplify the .gitignore file
...
Peter Maydell [Wed, 16 Sep 2020 15:25:30 +0000 (16:25 +0100)]
Merge remote-tracking branch 'remotes/berrange/tags/block-odirect-pull-request' into staging
block: improve error reporting for unsupported O_DIRECT
# gpg: Signature made Wed 16 Sep 2020 10:45:48 BST
# gpg: using RSA key DAF3A6FDB26B62912D0E8E3FBE86EBB415104FDF
# gpg: Good signature from "Daniel P. Berrange <[email protected]>" [full]
# gpg: aka "Daniel P. Berrange <[email protected]>" [full]
# Primary key fingerprint: DAF3 A6FD B26B 6291 2D0E 8E3F BE86 EBB4 1510 4FDF
* remotes/berrange/tags/block-odirect-pull-request:
block/file: switch to use qemu_open/qemu_create for improved errors
util: give a specific error message when O_DIRECT doesn't work
util: introduce qemu_open and qemu_create with error reporting
util: add Error object for qemu_open_internal error reporting
util: refactor qemu_open_old to split off variadic args handling
util: rename qemu_open() to qemu_open_old()
util: split off a helper for dealing with O_CLOEXEC flag
monitor: simplify functions for getting a dup'd fdset entry
Peter Maydell [Wed, 16 Sep 2020 13:47:50 +0000 (14:47 +0100)]
Merge remote-tracking branch 'remotes/cschoenebeck/tags/pull-9p-20200915' into staging
The intention of the following two patches is making users aware about
the negative file I/O performance impact when using a very low value
for 9P client parameter 'msize', which especially is the case if no
'msize' parameter was supplied by the user with a 9P Linux client at all.
All it does is logging a performance warning on host side (once) in
that case. By setting 'msize' on client side to any value larger than
8192 the performance warning will disappear.
See https://wiki.qemu.org/Documentation/9psetup#msize for details.
# gpg: Signature made Tue 15 Sep 2020 11:37:32 BST
# gpg: using RSA key 96D8D110CF7AF8084F88590134C2B58765A47395
# gpg: issuer "[email protected]"
# gpg: Good signature from "Christian Schoenebeck <[email protected]>" [unknown]
# 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: ECAB 1A45 4014 1413 BA38 4926 30DB 47C3 A012 D5F4
# Subkey fingerprint: 96D8 D110 CF7A F808 4F88 5901 34C2 B587 65A4 7395
* remotes/cschoenebeck/tags/pull-9p-20200915:
9pfs: disable msize warning for synth driver
9pfs: log warning if msize <= 8192
I think we going to a wrong direction, I think there is some tests a stall the test runner,
please look at
https://cirrus-ci.com/task/5110577531977728
When its running properly, the consumed time are little, but when tests running too long,
look at the cpu usage, the cpu usage are nearly zero. doesn't consuming time.
Yonggang Luo [Tue, 15 Sep 2020 12:13:16 +0000 (20:13 +0800)]
ci: Enable msys2 ci in cirrus
Install msys2 in a proper way refer to:
https://github.com/cirruslabs/cirrus-ci-docs/issues/699
The https://wiki.qemu.org/Hosts/W32#Native_builds_with_MSYS2 need to be updated.
There is no need of --cross-prefix, open mingw64.exe instead of msys2.exe then
we don't need the --cross-prefix, besides we use environment variable settings:
MSYS: winsymlinks:nativestrict
MSYSTEM: MINGW64
CHERE_INVOKING: 1
to opening mingw64 native shell.
block/file: switch to use qemu_open/qemu_create for improved errors
Currently at startup if using cache=none on a filesystem lacking
O_DIRECT such as tmpfs, at startup QEMU prints
qemu-system-x86_64: -drive file=/tmp/foo.img,cache=none: file system may not support O_DIRECT
qemu-system-x86_64: -drive file=/tmp/foo.img,cache=none: Could not open '/tmp/foo.img': Invalid argument
while at QMP level the hint is missing, so QEMU reports just
"error": {
"class": "GenericError",
"desc": "Could not open '/tmp/foo.img': Invalid argument"
}
which is close to useless for the end user trying to figure out what
they did wrong.
With this change at startup QEMU prints
qemu-system-x86_64: -drive file=/tmp/foo.img,cache=none: Unable to open '/tmp/foo.img': filesystem does not support O_DIRECT
while at the QMP level QEMU reports a massively more informative
"error": {
"class": "GenericError",
"desc": "Unable to open '/tmp/foo.img': filesystem does not support O_DIRECT"
}
util: give a specific error message when O_DIRECT doesn't work
A common error scenario is to tell QEMU to use O_DIRECT in combination
with a filesystem that doesn't support it. To aid users to diagnosing
their mistake we want to provide a clear error message when this happens.
util: introduce qemu_open and qemu_create with error reporting
qemu_open_old() works like open(): set errno and return -1 on failure.
It has even more failure modes, though. Reporting the error clearly
to users is basically impossible for many of them.
Our standard cure for "errno is too coarse" is the Error object.
Introduce two new helper methods:
int qemu_open(const char *name, int flags, Error **errp);
int qemu_create(const char *name, int flags, mode_t mode, Error **errp);
Note that with this design we no longer require or even accept the
O_CREAT flag. Avoiding overloading the two distinct operations
means we can avoid variable arguments which would prevent 'errp' from
being the last argument. It also gives us a guarantee that the 'mode' is
given when creating files, avoiding a latent security bug.
We want to introduce a new version of qemu_open() that uses an Error
object for reporting problems and make this it the preferred interface.
Rename the existing method to release the namespace for the new impl.
monitor: simplify functions for getting a dup'd fdset entry
Currently code has to call monitor_fdset_get_fd, then dup
the return fd, and then add the duplicate FD back into the
fdset. This dance is overly verbose for the caller and
introduces extra failure modes which can be avoided by
folding all the logic into monitor_fdset_dup_fd_add and
removing monitor_fdset_get_fd entirely.
scripts/git.orderfile: Display meson files along with buildsys ones
Since commit a56650518f5 ("configure: integrate Meson in the build
system") we replaced many Makefile by Meson files. Adapt the
git.orderfile script to display the new file at the same position.
Fix compiler error about defined but not used functions when compiling
with -DHPET_DEBUG by deleting the unused debug functions hpet_ram_readb
and hpet_ram_readw.
Alex Bennée [Tue, 15 Sep 2020 13:43:16 +0000 (14:43 +0100)]
gitlab: create a build-deprecated target
These targets might be deprecated but we should keep them building
before the final axe comes down. Lets keep them all in one place and
don't hold up the CI if they do fail. They are either poorly tested or
already flaky anyway.
Alex Bennée [Tue, 15 Sep 2020 13:43:13 +0000 (14:43 +0100)]
configure: also skip deprecated targets with target-list-exclude
Now the user has to make an even more deliberate decision to
enable a deprecated target rather than getting it as a side effect of
using --target-exclude-list.
Alex Bennée [Tue, 15 Sep 2020 13:43:12 +0000 (14:43 +0100)]
configure: move deprecated feature processing to supported_target
This is the common point at which we validate targets so it makes
sense to add_to deprecated_features here. It will make future target
deprecation easier as we only need to tweak one list.
Max Reitz [Tue, 15 Sep 2020 13:43:11 +0000 (14:43 +0100)]
iotests: Drop readlink -f
On macOS, (out of the box) readlink does not have -f. We do not really
need readlink here, though, it was just a replacement for realpath
(which is not available on our BSD test systems), which we needed to
make the $(dirname) into an absolute path.
Instead of using either, just use "cd; pwd" like is done for
$source_iotests.
("iotests: Allow running from different directory")
Alex Bennée [Tue, 15 Sep 2020 13:43:10 +0000 (14:43 +0100)]
linux-user: test, don't assert addr != test in pgb_reserved_va
On older kernels which don't implement MAP_FIXED_NOREPLACE the kernel
may still fail to give us the address we asked for despite having
already probed the map for a valid hole. Asserting isn't particularly
useful to the user so let us move the check up and expand the
error_report a little to give them a fighting chance of working around
the problem.
manual: escape backslashes in "parsed-literal" blocks
According to
<https://docutils.sourceforge.io/docs/ref/rst/directives.html#parsed-literal>,
"inline markup is recognized and there is no protection from parsing.
Backslash-escapes may be necessary to prevent unintended parsing".
The qemu(1) manual page (formatted with Sphinx 2.2.2) has several overlong
lines on my system. A stand-alone backslash at EOL serves as line
continuation in a "parsed-literal" block. Therefore, escape the
backslashes that we want to appear as such in the formatted documentation.
projects / qemu.git / log