All callers of qio_net_listener_set_name() already add some sort of
"listen" or "listener" suffix.
For intance, we currently have "migration-socket-listener-listen" and
"vnc-listen-listen" as ioc names.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
To do so, create two paired sockets, but make them not providing real data.
Feed those fake sockets to src/dst QEMUs for recovery to let them go into
RECOVER stage without going out. Test that we can always kick it out and
recover again with the right ports.
This patch is based on Fabiano's version here:
https://lore.kernel.org/r/877cowmdu0.fsf@suse.de
Signed-off-by: Fabiano Rosas <farosas@suse.de>
[peterx: write commit message, remove case 1, fix bugs, and more]
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231017202633.296756-4-peterx@redhat.com>
Now that we can query more than one binary, the "starting QEMU..."
message can get a little noisy. Mute those messages unless we're
running with --verbose.
Only affects qtest_init() calls from within libqtest. The tests
continue to output as usual.
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-ID: <20231018192741.25885-13-farosas@suse.de>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Accept the QTEST_QEMU_MACHINE_TYPE environment variable to take a
machine type to use in the tests.
The full machine type is recognized (e.g. pc-q35-8.2). Aliases
(e.g. pc) are also allowed and resolve to the latest machine version
for that alias, or, if using two QEMU binaries, to the latest common
machine version between the two.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231018192741.25885-12-farosas@suse.de>
We have strict rules around migration compatibility between different
QEMU versions but no test to validate the migration state between
different binaries.
Add infrastructure to allow running the migration tests with two
different QEMU binaries as migration source and destination.
The code now recognizes two new environment variables
QTEST_QEMU_BINARY_SRC and QTEST_QEMU_BINARY_DST. In the absence of
either of them, the test will use the QTEST_QEMU_BINARY variable. If
both are missing then the tests are run with single binary as
previously.
The machine type is selected automatically as the latest machine type
version that works with both binaries.
Usage (only one of SRC|DST is allowed):
QTEST_QEMU_BINARY_SRC=../build-8.2.0/qemu-system-x86_64 \
QTEST_QEMU_BINARY=../build-8.1.0/qemu-system-x86_64 \
./tests/qtest/migration-test
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231018192741.25885-11-farosas@suse.de>
We're about to enable the x86_64 tests to run with the q35 machine,
but that machine does not work with the program we use to dirty the
memory for the tests.
The issue is that QEMU needs to guess the geometry of the "disk" we
give to it and the guessed geometry doesn't pass the sanity checks
done by SeaBIOS. This causes SeaBIOS to interpret the geometry as if
needing a translation from LBA to CHS and SeaBIOS ends up miscomputing
the number of cylinders and aborting due to that.
The reason things work with the "pc" machine is that is uses ATA
instead of AHCI like q35 and SeaBIOS has an exception for ATA that
ends up skipping the sanity checks and ignoring translation
altogether.
Workaround this situation by specifying a geometry in the command
line.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Acked-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231018192741.25885-9-farosas@suse.de>
Stop relying on defaults and select a machine explicitly for every
architecture.
This is a prerequisite for being able to select machine types for
migration using different QEMU binaries for source and destination.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231018192741.25885-8-farosas@suse.de>
The migration tests are being enhanced to test migration between
different QEMU versions. A requirement of migration is that the
machine type between source and destination matches, including the
version.
We cannot hardcode machine types in the tests because those change
with each release. QEMU provides a machine type alias that has a fixed
name, but points to the latest machine type at each release.
Add a helper to resolve the alias into the exact machine
type. E.g. "-machine pc" resolves to "pc-i440fx-8.2"
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231018192741.25885-6-farosas@suse.de>
We're adding support for using more than one QEMU binary in
tests. Modify qtest_get_machines() to take an environment variable
that contains the QEMU binary path.
Since the function keeps a cache of the machines list in the form of a
static variable, refresh it any time the environment variable changes.
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231018192741.25885-4-farosas@suse.de>
Add a version of qtest_init() that takes an environment variable
containing the path of the QEMU binary. This allows tests to use more
than one QEMU binary.
If no variable is provided or the environment variable does not exist,
that is not an error. Fallback to using QTEST_QEMU_BINARY.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231018192741.25885-3-farosas@suse.de>
We're adding support for testing migration using two different QEMU
binaries. We'll provide the second binary in a new environment
variable.
Allow qtest_qemu_binary() to receive the name of the new variable. If
the new environment variable is not set, that's not an error, we use
QTEST_QEMU_BINARY as a fallback.
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231018192741.25885-2-farosas@suse.de>
We don't need to check p->quit in the multifd_send_thread() because it
is shadowed by the 'exiting' flag. Ever since that flag was added
p->quit became obsolete as a way to stop the thread.
Since p->quit is set at multifd_send_terminate_threads() under the
p->mutex lock, the thread will only see it once it loops, so 'exiting'
will always be seen first.
Note that setting p->quit at multifd_send_terminate_threads() still
makes sense because we need a way to inform multifd_send_pages() that
the channel has stopped.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231012140651.13122-3-farosas@suse.de>
The function is currently called from two sites, one always gives it a
NULL Error and the other always gives it a non-NULL Error.
In the non-NULL case, all it does it trace the error and return. One
of the callers already have tracing, add a tracepoint to the other and
stop passing the error into the function.
Cc: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231012134343.23757-4-farosas@suse.de>
The preferred usage of the Error type is to always set both the return
code and the error when a failure happens. As all code called from the
send thread follows this pattern, we'll always have the return code
and the error set at the same time.
Aside from the convention, in this piece of code this must be the
case, otherwise the if (ret != 0) would be exiting the thread without
calling multifd_send_terminate_threads() which is incorrect.
Unify both paths to make it clear that both are taken when there's an
error.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231012134343.23757-3-farosas@suse.de>
It makes a bit more sense to have the zero page handling of xbzrle
right where we save the zero page.
Also invert the exit condition to remove one level of indentation
which makes the next patch easier to grasp.
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231011184604.32364-6-farosas@suse.de>
Add basic tests for file-based migration.
Note that we cannot use test_precopy_common because that routine
expects it to be possible to run the migration live. With the file
transport there is no live migration because we must wait for the
source to finish writing the migration data to the file before the
destination can start reading. Add a new migration function
specifically to handle the file migration.
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20230712190742.22294-7-farosas@suse.de>
Add a smoke test that migrates to a file and gives it to the
script. It should catch the most annoying errors such as changes in
the ram flags.
After code has been merged it becomes way harder to figure out what is
causing the script to fail, the person making the change is the most
likely to know right away what the problem is.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Acked-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231009184326.15777-7-farosas@suse.de>
The migration code uses unsigned values for 16, 32 and 64-bit
operations. Fix the script to do the same.
This was causing an issue when parsing the migration stream generated
on the ppc64 target because one of instance_ids was larger than the
32bit signed maximum:
Traceback (most recent call last):
File "/home/fabiano/kvm/qemu/build/scripts/analyze-migration.py", line 658, in <module>
dump.read(dump_memory = args.memory)
File "/home/fabiano/kvm/qemu/build/scripts/analyze-migration.py", line 592, in read
classdesc = self.section_classes[section_key]
KeyError: ('spapr_iommu', -2147483648)
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231009184326.15777-6-farosas@suse.de>
The script is currently broken when the x-ignore-shared capability is
used:
Traceback (most recent call last):
File "./scripts/analyze-migration.py", line 656, in <module>
dump.read(dump_memory = args.memory)
File "./scripts/analyze-migration.py", line 593, in read
section.read()
File "./scripts/analyze-migration.py", line 163, in read
self.name = self.file.readstr(len = namelen)
File "./scripts/analyze-migration.py", line 53, in readstr
return self.readvar(len).decode('utf-8')
UnicodeDecodeError: 'utf-8' codec can't decode byte 0x82 in position 55: invalid start byte
We're currently adding data to the middle of the ram section depending
on the presence of the capability. As a consequence, any code loading
the ram section needs to know about capabilities so it can interpret
the stream.
Skip the byte that's added when x-ignore-shared is used to fix the
script.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231009184326.15777-5-farosas@suse.de>
The 'configuration' state subsections are currently not being parsed
and the script fails when analyzing an aarch64 stream:
Traceback (most recent call last):
File "./scripts/analyze-migration.py", line 625, in <module>
dump.read(dump_memory = args.memory)
File "./scripts/analyze-migration.py", line 571, in read
raise Exception("Unknown section type: %d" % section_type)
Exception: Unknown section type: 5
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231009184326.15777-3-farosas@suse.de>
There is currently no way to write a test for errors that happened in
qmp_migrate before the migration has started.
Add a version of qmp_migrate that ensures an error happens. To make
use of it a test needs to set MigrateCommon.result as
MIG_TEST_QMP_ERROR.
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20230712190742.22294-6-farosas@suse.de>
We are sending a migration event of MIGRATION_STATUS_SETUP at
qemu_start_incoming_migration but never actually setting the state.
This creates a window between qmp_migrate_incoming and
process_incoming_migration_co where the migration status is still
MIGRATION_STATUS_NONE. Calling query-migrate during this time will
return an empty response even though the incoming migration command
has already been issued.
Commit 7cf1fe6d68 ("migration: Add migration events on target side")
has added support to the 'events' capability to the incoming part of
migration, but chose to send the SETUP event without setting the
state. I'm assuming this was a mistake.
This introduces a change in behavior, any QMP client waiting for the
SETUP event will hang, unless it has previously enabled the 'events'
capability. Having the capability enabled is sufficient to continue to
receive the event.
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20230712190742.22294-5-farosas@suse.de>
file-based migration requires the target to initiate its migration after
the source has finished writing out the data in the file. Currently
there's no easy way to initiate 'migrate-incoming', allow this by
introducing migrate_incoming_qmp helper, similarly to migrate_qmp.
Also make sure migration events are enabled and wait for the incoming
migration to start before returning. This avoid a race when querying
the migration status too soon after issuing the command.
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20230712190742.22294-3-farosas@suse.de>
Our linker script for optionroms specifies only the placement of the
.text section, leaving the linker free to place the remaining sections
at arbitrary places in the file.
Since at least binutils 2.39, the .note.gnu.build-id section is now
being placed at the start of the file, which causes label addresses to
be shifted. For linuxboot_dma.bin that means that the PnP header
(among others) will not be found when determining the type of ROM at
optionrom_setup():
(0x1c is the label _pnph, where the magic "PnP" is)
$ xxd /usr/share/qemu/linuxboot_dma.bin | grep "PnP"
00000010: 0000 0000 0000 0000 0000 1c00 2450 6e50 ............$PnP
$ xxd pc-bios/optionrom/linuxboot_dma.bin | grep "PnP"
00000010: 0000 0000 0000 0000 0000 4c00 2450 6e50 ............$PnP
^bad
Using a freshly built linuxboot_dma.bin ROM results in a broken boot:
SeaBIOS (version rel-1.16.2-0-gea1b7a073390-prebuilt.qemu.org)
Booting from Hard Disk...
Boot failed: could not read the boot disk
Booting from Floppy...
Boot failed: could not read the boot disk
No bootable device.
We're not using the build-id section, so pass the --build-id=none
option to the linker to remove it entirely.
Note: In theory, this same issue could happen with any other
section. The ideal solution would be to have all unused sections
discarded in the linker script. However that would be a larger change,
specially for the pvh rom which uses the .bss and COMMON sections so
I'm addressing only the immediate issue here.
Reported-by: Vasiliy Ulyanov <vulyanov@suse.de>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20230926192502.15986-1-farosas@suse.de>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Replace the return path retry logic with finishing and restarting the
thread. This fixes a race when resuming the migration that leads to a
segfault.
Currently when doing postcopy we consider that an IO error on the
return path file could be due to a network intermittency. We then keep
the thread alive but have it do cleanup of the 'from_dst_file' and
wait on the 'postcopy_pause_rp' semaphore. When the user issues a
migrate resume, a new return path is opened and the thread is allowed
to continue.
There's a race condition in the above mechanism. It is possible for
the new return path file to be setup *before* the cleanup code in the
return path thread has had a chance to run, leading to the *new* file
being closed and the pointer set to NULL. When the thread is released
after the resume, it tries to dereference 'from_dst_file' and crashes:
Thread 7 "return path" received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7fffd1dbf700 (LWP 9611)]
0x00005555560e4893 in qemu_file_get_error_obj (f=0x0, errp=0x0) at ../migration/qemu-file.c:154
154 return f->last_error;
(gdb) bt
#0 0x00005555560e4893 in qemu_file_get_error_obj (f=0x0, errp=0x0) at ../migration/qemu-file.c:154
#1 0x00005555560e4983 in qemu_file_get_error (f=0x0) at ../migration/qemu-file.c:206
#2 0x0000555555b9a1df in source_return_path_thread (opaque=0x555556e06000) at ../migration/migration.c:1876
#3 0x000055555602e14f in qemu_thread_start (args=0x55555782e780) at ../util/qemu-thread-posix.c:541
#4 0x00007ffff38d76ea in start_thread (arg=0x7fffd1dbf700) at pthread_create.c:477
#5 0x00007ffff35efa6f in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95
Here's the race (important bit is open_return_path happening before
migration_release_dst_files):
migration | qmp | return path
--------------------------+-----------------------------+---------------------------------
qmp_migrate_pause()
shutdown(ms->to_dst_file)
f->last_error = -EIO
migrate_detect_error()
postcopy_pause()
set_state(PAUSED)
wait(postcopy_pause_sem)
qmp_migrate(resume)
migrate_fd_connect()
resume = state == PAUSED
open_return_path <-- TOO SOON!
set_state(RECOVER)
post(postcopy_pause_sem)
(incoming closes to_src_file)
res = qemu_file_get_error(rp)
migration_release_dst_files()
ms->rp_state.from_dst_file = NULL
post(postcopy_pause_rp_sem)
postcopy_pause_return_path_thread()
wait(postcopy_pause_rp_sem)
rp = ms->rp_state.from_dst_file
goto retry
qemu_file_get_error(rp)
SIGSEGV
-------------------------------------------------------------------------------------------
We can keep the retry logic without having the thread alive and
waiting. The only piece of data used by it is the 'from_dst_file' and
it is only allowed to proceed after a migrate resume is issued and the
semaphore released at migrate_fd_connect().
Move the retry logic to outside the thread by waiting for the thread
to finish before pausing the migration.
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230918172822.19052-8-farosas@suse.de>
It's not safe to call qemu_file_shutdown() on the to_dst_file without
first checking for the file's presence under the lock. The cleanup of
this file happens at postcopy_pause() and migrate_fd_cleanup() which
are not necessarily running in the same thread as migrate_fd_cancel().
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230918172822.19052-5-farosas@suse.de>
We cannot call qemu_file_shutdown() on the return path file without
taking the file lock. The return path thread could be running it's
cleanup code and have just cleared the from_dst_file pointer.
Checking ms->to_dst_file for errors could also race with
migrate_fd_cleanup() which clears the to_dst_file pointer.
Protect both accesses by taking the file lock.
This was caught by inspection, it should be rare, but the next patches
will start calling this code from other places, so let's do the
correct thing.
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230918172822.19052-4-farosas@suse.de>
We don't need to set the rp_state.error right after a shutdown because
qemu_file_shutdown() always sets the QEMUFile error, so the return
path thread would have seen it and set the rp error itself.
Setting the error outside of the thread is also racy because the
thread could clear it after we set it.
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230918172822.19052-3-farosas@suse.de>
Having a name in the source helps with debugging core dumps when one
might not have access to TLS data to cross-reference AioContexts with
their addresses.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230905180359.14083-1-farosas@suse.de
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
We can fail the blk_insert_bs() at init_blk_migration(), leaving the
BlkMigDevState without a dirty_bitmap and BlockDriverState. Account
for the possibly missing elements when doing cleanup.
Fix the following crashes:
Thread 1 "qemu-system-x86" received signal SIGSEGV, Segmentation fault.
0x0000555555ec83ef in bdrv_release_dirty_bitmap (bitmap=0x0) at ../block/dirty-bitmap.c:359
359 BlockDriverState *bs = bitmap->bs;
#0 0x0000555555ec83ef in bdrv_release_dirty_bitmap (bitmap=0x0) at ../block/dirty-bitmap.c:359
#1 0x0000555555bba331 in unset_dirty_tracking () at ../migration/block.c:371
#2 0x0000555555bbad98 in block_migration_cleanup_bmds () at ../migration/block.c:681
Thread 1 "qemu-system-x86" received signal SIGSEGV, Segmentation fault.
0x0000555555e971ff in bdrv_op_unblock (bs=0x0, op=BLOCK_OP_TYPE_BACKUP_SOURCE, reason=0x0) at ../block.c:7073
7073 QLIST_FOREACH_SAFE(blocker, &bs->op_blockers[op], list, next) {
#0 0x0000555555e971ff in bdrv_op_unblock (bs=0x0, op=BLOCK_OP_TYPE_BACKUP_SOURCE, reason=0x0) at ../block.c:7073
#1 0x0000555555e9734a in bdrv_op_unblock_all (bs=0x0, reason=0x0) at ../block.c:7095
#2 0x0000555555bbae13 in block_migration_cleanup_bmds () at ../migration/block.c:690
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-id: 20230731203338.27581-1-farosas@suse.de
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This doubly linked list is common for all the multifd and migration
threads so we need to avoid concurrent access.
Add a mutex to protect the data from concurrent access. This fixes a
crash when removing two MigrationThread objects from the list at the
same time during cleanup of multifd threads.
Fixes: 671326201d ("migration: Introduce interface query-migrationthreads")
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20230607161306.31425-3-farosas@suse.de>
Signed-off-by: Juan Quintela <quintela@redhat.com>
This code is only relevant when TCG is present in the build. Building
with --disable-tcg --enable-xen on an x86 host we get:
$ ../configure --target-list=x86_64-softmmu,aarch64-softmmu --disable-tcg --enable-xen
$ make -j$(nproc)
...
libqemu-aarch64-softmmu.fa.p/target_arm_gdbstub.c.o: in function `m_sysreg_ptr':
../target/arm/gdbstub.c:358: undefined reference to `arm_v7m_get_sp_ptr'
../target/arm/gdbstub.c:361: undefined reference to `arm_v7m_get_sp_ptr'
libqemu-aarch64-softmmu.fa.p/target_arm_gdbstub.c.o: in function `arm_gdb_get_m_systemreg':
../target/arm/gdbstub.c:405: undefined reference to `arm_v7m_mrs_control'
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-id: 20230628164821.16771-1-farosas@suse.de
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
When we moved the arm default CONFIGs into Kconfig and removed them
from default.mak, we made it harder to identify which CONFIGs are
selected by default in case users want to disable them.
Bring back the default entries into default.mak, but keep them
commented out. This way users can keep their workflows of editing
default.mak to remove build options without needing to search through
Kconfig.
Reported-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-id: 20230523180525.29994-3-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
We currently need to select ARM_V7M unconditionally when TCG is
present in the build because some translate.c helpers and the whole of
m_helpers.c are not yet under CONFIG_ARM_V7M.
Suggested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230523180525.29994-2-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
We recently moved glib detection code to meson but this changes the
linker command line from -lglib-2.0 to using a path to libglib-2.0.so.
This does not work for static linking, which is used by stress.c:
$ make V=1 tests/migration/initrd-stress.img
cc -m64 -mcx16 -o tests/migration/stress ... -static -Wl,--start-group
/usr/lib64/libglib-2.0.so -Wl,--end-group
...
bin/ld: attempted static link of dynamic object `/usr/lib64/libglib-2.0.so'
Add a specific dependency for stress.c, which is linked statically.
The compiler command line is now:
cc -m64 -mcx16 -o tests/migration/stress ... -static -pthread
-Wl,--start-group -lm /usr/lib64/libpcre.a -lglib-2.0 -Wl,--end-group
Fixes: fc9a809e0d ("build: move glib detection and workarounds to meson")
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-Id: <20230525212044.30222-3-farosas@suse.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Save a bit of build time by passing the number of jobs option to
sphinx.
We cannot use the -j option from make because meson does not support
setting build time parameters for custom targets. Use nproc instead or
the equivalent sphinx option "-j auto", if that is available (version
>=1.7.0).
Also make sure our plugins support parallelism and report it properly
to sphinx. Particularly, implement the merge_domaindata method in
DBusDomain that is used to merge in data from other subprocesses.
Tested-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-Id: <20230503203947.3417-2-farosas@suse.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
On a build configured with: --disable-tcg --enable-xen it is possible
to produce a QEMU binary with no TCG nor KVM support. Skip the cdrom
boot tests if that's the case.
Fixes: 0c1ae3ff9d ("tests/qtest: Fix tests when no KVM or TCG are present")
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-id: 20230508181611.2621-4-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
We cannot allow this config to be disabled at the moment as not all of
the relevant code is protected by it.
Commit 29d9efca16 ("arm/Kconfig: Do not build TCG-only boards on a
KVM-only build") moved the CONFIGs of several boards to Kconfig, so it
is now possible that nothing selects ARM_V7M (e.g. when doing a
--without-default-devices build).
Return the CONFIG_ARM_V7M entry to a state where it is always selected
whenever TCG is available.
Fixes: 29d9efca16 ("arm/Kconfig: Do not build TCG-only boards on a KVM-only build")
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230508181611.2621-3-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Semihosting has been made a 'default y' entry in Kconfig, which does
not work because when building --without-default-devices, the
semihosting code would not be available.
Make semihosting unconditional when TCG is present.
Fixes: 29d9efca16 ("arm/Kconfig: Do not build TCG-only boards on a KVM-only build")
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230508181611.2621-2-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This machine hardcodes initialization of the USB device, so select the
corresponding Kconfig. It is not enough to have it as "default y if
XLNX_VERSAL" at usb/Kconfig because building --without-default-devices
disables the default selection resulting in:
$ ./qemu-system-aarch64 -M xlnx-zcu102
qemu-system-aarch64: missing object type 'usb_dwc3'
Aborted (core dumped)
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230208192654.8854-8-farosas@suse.de>
Message-Id: <20230503091244.1450613-8-alex.bennee@linaro.org>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Move all the CONFIG_FOO=y from default.mak into "default y if TCG"
statements in Kconfig. That way they won't be selected when
CONFIG_TCG=n.
I'm leaving CONFIG_ARM_VIRT in default.mak because it allows us to
keep the two default.mak files not empty and keep aarch64-default.mak
including arm-default.mak. That way we don't surprise anyone that's
used to altering these files.
With this change we can start building with --disable-tcg.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230426180013.14814-12-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
We are about to enable the build without TCG, so CONFIG_SEMIHOSTING
and CONFIG_ARM_COMPATIBLE_SEMIHOSTING cannot be unconditionally set in
default.mak anymore. So reflect the change in a Kconfig.
Instead of using semihosting/Kconfig, use a target-specific file, so
that the change doesn't affect other architectures which might
implement semihosting in a way compatible with KVM.
The selection from ARM_v7M needs to be removed to avoid a cycle during
parsing.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230426180013.14814-11-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
The migration tests are currently broken for an aarch64 host because
the tests pass no 'machine' and 'cpu' options on the QEMU command
line.
Add a separate class to each architecture so that we can specify
'machine' and 'cpu' options instead of relying on defaults.
Add a skip decorator to keep the current behavior of only running
migration tests when the qemu target matches the host architecture.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-id: 20230426180013.14814-10-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
It is possible to have a build with both TCG and KVM disabled due to
Xen requiring the i386 and x86_64 binaries to be present in an aarch64
host.
If we build with --disable-tcg on the aarch64 host, we will end-up
with a QEMU binary (x86) that does not support TCG nor KVM.
Skip tests that crash or hang in the above scenario. Do not include
any test cases if TCG and KVM are missing.
Make sure that calls to qtest_has_accel are placed after g_test_init
in similar fashion to commit ae4b01b349 ("tests: Ensure TAP version is
printed before other messages") to avoid TAP parsing errors.
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230426180013.14814-9-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
We're about to move the 32-bit CPUs under CONFIG_TCG, so adjust the
query-cpu-model-expansion test to check against the cortex-a7, which
is already under CONFIG_TCG. That allows the next patch to contain
only code movement. (All the test cares about is that the CPU type
it's checking is one which definitely doesn't work under KVM.)
While here add comments clarifying what we're testing.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Acked-by: Thomas Huth <thuth@redhat.com>
Message-id: 20230426180013.14814-7-farosas@suse.de
Suggested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
We're about to move the TCG-only -cpu max configuration code under
CONFIG_TCG. To be able to do that we need to make sure the qtests
still have some cpu configured even when no other accelerator is
available.
Delineate now what is used with TCG-only and what is also used with
qtests to make the subsequent patches cleaner.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230426180013.14814-5-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Our dockerfiles no longer reference layers from other qemu images so
we can now use 'docker build' on them.
Also reinstate the caching that was disabled due to bad interactions
with certain runners. See commit 6ddc3dc7a8 ("tests/docker: don't use
BUILDKIT in GitLab either"). We now believe those issues to be fixed.
The COMMON_TAG needed to be fixed for the caching to work. The
docker.py script was not using the variable, but constructing the
correct URL directly.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Tested-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20230227151110.31455-2-farosas@suse.de>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230228190653.1602033-16-alex.bennee@linaro.org>
This test currently fails when run on a host for which the QEMU target
has no default machine set:
ERROR| Output: qemu-system-aarch64: No machine specified, and there is
no default
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This struct has no dependencies on TCG code and it is being used in
target/arm/ptw.c to simplify the passing around of page table walk
results. Those routines can be reached by KVM code via the gdbstub
breakpoint code, so take the structure out of CONFIG_TCG to make it
visible when building with --disable-tcg.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This function is needed by common code (ptw.c), so move it along with
the other regime_* functions in internal.h. When we enable the build
without TCG, the tlb_helper.c file will not be present.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
The next few patches will move helpers under CONFIG_TCG. We'd prefer
to keep the debug helpers and debug registers close together, so
rearrange the file a bit to be able to wrap the helpers with a TCG
ifdef.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Now that the cortex-a15 is under CONFIG_TCG, use as default CPU for a
KVM-only build the 'max' cpu.
Note that we cannot use 'host' here because the qtests can run without
any other accelerator (than qtest) and 'host' depends on KVM being
enabled.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Acked-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
If a test was tagged with the "accel" tag and the specified
accelerator it not present in the qemu binary, cancel the test.
We can now write tests without explicit calls to require_accelerator,
just the tag is enough.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Move this earlier to make the next patch diff cleaner. While here
update the comment slightly to not give the impression that the
misalignment affects only TCG.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
If we build with --without-default-devices, CONFIG_HPET and
CONFIG_PARALLEL are set to N, which makes the respective devices go
missing from acpi tables.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230208194700.11035-13-farosas@suse.de>
Signed-off-by: Thomas Huth <thuth@redhat.com>
Fabiano Rosas