once FW provides a pointer to SMBIOS entry point like it does for
RSDP it should be possible to enable this one the same way.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Message-Id: <1556808723-226478-11-git-send-email-imammedo@redhat.com>
Tested-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
For testcase to use UEFI firmware, one needs to provide and specify
firmware and varstore blob names in test_data { uefi_fl1, uefi_fl2 }
fields respectively and RAM start address plus size where to look for
test structure signature. Additionally testcase should specify
bootable cdrom image from uefi-boot-images with EFI test utility.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Message-Id: <1556808723-226478-10-git-send-email-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
introduce UEFI specific counterpart to acpi_find_rsdp_address()
that will help to find RSDP address when [OA]VMF is used as
firmware. It requires guest firmware or other guest app to place
1Mb aligned UefiTestSupport structure (defined in this patch)
in RAM with UefiTestSupport::signature_guid set to
AB87A6B1-2034-BDA0-71BD-375007757785
For test app details see commit
(09a274d82f tests: introduce "uefi-test-tools" with the BiosTablesTest UEFI app)
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <1556808723-226478-9-git-send-email-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
boot_sector_init() won't be used by arm/virt board, so move it from
global scope to x86 branch that uses it.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <1556808723-226478-8-git-send-email-imammedo@redhat.com>
Tested-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
If FADT has HW_REDUCED_ACPI flag set, do not attempt to fetch
FACS as it's not provided by the board.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <1556808723-226478-7-git-send-email-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
that way it would be possible to test a DSDT pointed by
64bit X_DSDT field in FADT.
PS:
it will allow to enable testing arm/virt board, which sets
only newer X_DSDT field.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <1556808723-226478-6-git-send-email-imammedo@redhat.com>
Reviewed-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
In case of UEFI, RSDP doesn't have to be located in lowmem,
it could be placed at any address. Make sure that test won't
break if it is placed above the first 4Gb of address space.
PS:
While at it cleanup some local variables as we don't really
need them.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <1556808723-226478-5-git-send-email-imammedo@redhat.com>
Reviewed-by: Wei Yang <richardw.yang@linux.intel.com>
Tested-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
If RSDP revision is more than 0 fetch table pointed by XSDT
and fallback to legacy RSDT table otherwise.
While at it drop unused acpi_get_xsdt_address().
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <1556808723-226478-4-git-send-email-imammedo@redhat.com>
Reviewed-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Currently acpi_fetch_table() assumes 32 bit size of table pointer
in ACPI tables. However X_foo variants are 64 bit, prepare
acpi_fetch_table() to handle both by adding an argument
for addr_ptr pointed entry size. Follow up commits will use that
to read XSDT and X_foo entries in ACPI tables.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <1556808723-226478-3-git-send-email-imammedo@redhat.com>
Reviewed-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
so name would reflect what the function does
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <1556808723-226478-2-git-send-email-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
pci_bus_is_root() currently relies on a method in the PCIBusClass.
But it's always known if a PCI bus is a root bus when we create it, so
using a dynamic method is overkill.
This replaces it with an IS_ROOT bit in a new flags field, which is set on
root buses and otherwise clear. As a bonus this removes the special
is_root logic from pci_expander_bridge, since it already creates its bus
as a root bus.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Marcel Apfelbaum <marcel@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <20190424041959.4087-3-david@gibson.dropbear.id.au>
These functions have an explicit test for accesses above the device's
config size. But pci_host_config_{read,write}_common() which they're
about to call already have checks against the config space limit and
do the right thing. So, remove the redundant tests.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Message-Id: <20190424041959.4087-2-david@gibson.dropbear.id.au>
Mark dirty as page, the step of each call is 1.
Signed-off-by: Li Feng <fengli@smartx.com>
Message-Id: <20190420091016.213160-1-fengli@smartx.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>
To build MCFG, two information is necessary:
* bus number
* base address
Abstract these two information to AcpiMcfgInfo so that build_mcfg and
build_mcfg_q35 will have the same declaration.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20190419003053.8260-5-richardw.yang@linux.intel.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
This is obvious the member in AcpiMcfgInfo describe MCFG's property.
Remove the mcfg_ prefix.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Suggested-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20190419003053.8260-4-richardw.yang@linux.intel.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
mcfg_start points to the start of MCFG table and is used in
build_header. While this information could be derived from mcfg.
This patch removes the unnecessary variable mcfg_start.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20190419003053.8260-3-richardw.yang@linux.intel.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Buglink: https://launchpad.net/bugs/1823458
Currently, a user CHR_EVENT_CLOSED event will cause net_vhost_user_event()
to call vhost_user_cleanup(), which calls vhost_net_cleanup() for all
its queues. However, vhost_net_cleanup() must never be called like
this for fully-initialized nets; when other code later calls
vhost_net_stop() - such as from virtio_net_vhost_status() - it will try
to access the already-cleaned-up fields and fail with assertion errors
or segfaults.
The vhost_net_cleanup() will eventually be called from
qemu_cleanup_net_client().
Signed-off-by: Dan Streetman <ddstreet@canonical.com>
Message-Id: <20190416184624.15397-3-dan.streetman@canonical.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Dummy table (with signature "QEMU") creation came from original SeaBIOS
codebase. And QEMU would have to keep it around if there were Q35 machine
that depended on keeping ACPI tables blob constant size. Luckily there
were no versioned Q35 machine types before commit:
(since 2.3) a1666142db acpi-build: make ROMs RAM blocks resizeable
which obsoleted need to keep ACPI tables blob the same size on source/destination.
Considering the 1st versioned machine is pc-q35-2.4, the dummy table
is not really necessary and it's safe to drop it without breaking
cross version migration in both directions unconditionally.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <1554822037-329838-1-git-send-email-imammedo@redhat.com>
Reviewed-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20190402161900.7374-4-armbru@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20190402161900.7374-3-armbru@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20190402161900.7374-2-armbru@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
This patch enables inflight I/O tracking for
vhost-user-blk backend so that we could restart it safely.
Signed-off-by: Xie Yongji <xieyongji@baidu.com>
Signed-off-by: Zhang Yu <zhangyu31@baidu.com>
Message-Id: <20190320112646.3712-8-xieyongji@baidu.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Since we now support the message VHOST_USER_GET_INFLIGHT_FD
and VHOST_USER_SET_INFLIGHT_FD. The backend is able to restart
safely because it can track inflight I/O in shared memory.
This patch allows qemu to reconnect the backend after
connection closed.
Signed-off-by: Xie Yongji <xieyongji@baidu.com>
Signed-off-by: Ni Xun <nixun@baidu.com>
Signed-off-by: Zhang Yu <zhangyu31@baidu.com>
Message-Id: <20190320112646.3712-7-xieyongji@baidu.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Add a return value for vhost_user_blk_start() to check whether
we start vhost-user backend successfully or not.
Signed-off-by: Xie Yongji <xieyongji@baidu.com>
Signed-off-by: Zhang Yu <zhangyu31@baidu.com>
Message-Id: <20190320112646.3712-6-xieyongji@baidu.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
We should only start vhost-user backend at the first kick for
virtio 1.0 transitional devices.
Signed-off-by: Xie Yongji <xieyongji@baidu.com>
Signed-off-by: Zhang Yu <zhangyu31@baidu.com>
Message-Id: <20190320112646.3712-5-xieyongji@baidu.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Use started flag in vhost_user_blk_set_status() to decide if
starting vhost-user backend or not.
Signed-off-by: Xie Yongji <xieyongji@baidu.com>
Signed-off-by: Zhang Yu <zhangyu31@baidu.com>
Message-Id: <20190320112646.3712-4-xieyongji@baidu.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Currently, we use DRIVER_OK status bit to check whether guest
driver has started the device in virtio_vmstate_change(). But it's
not the case for virtio 1.0 transitional devices. If migration completes
between kicking virtqueue and setting VIRTIO_CONFIG_S_DRIVER_OK, guest
may be hung. So here we use started flag to check guest state instead.
Signed-off-by: Xie Yongji <xieyongji@baidu.com>
Signed-off-by: Zhang Yu <zhangyu31@baidu.com>
Message-Id: <20190320112646.3712-3-xieyongji@baidu.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
The virtio 1.0 transitional devices support driver uses the device
before setting the DRIVER_OK status bit. So we introduce a started
flag to indicate whether driver has started the device or not.
Signed-off-by: Xie Yongji <xieyongji@baidu.com>
Signed-off-by: Zhang Yu <zhangyu31@baidu.com>
Message-Id: <20190320112646.3712-2-xieyongji@baidu.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20190315180735.13096-1-marcandre.lureau@redhat.com>
Reviewed-by: Jens Freimann <jfreimann@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
This reverts commit d7741743f4.
Relying on setting properties on parents types which may not
be relevant to certain sub-classes had unexpected side-effects
causing bugs in device config defaults. It is preferrable to
be explicit about which devices get which properties, even if
this needs repetition.
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20190215103239.28640-3-berrange@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
A number of virtio devices (gpu, crypto, mouse, keyboard, tablet) only
support the virtio-1 (aka modern) mode. Currently if the user launches
QEMU, setting those devices to enable legacy mode, QEMU will silently
create them in modern mode, ignoring the user's (mistaken) request.
This patch introduces proper data validation so that an attempt to
configure a virtio-1-only devices in legacy mode gets reported as an
error to the user.
Checking this required introduction of a new field to explicitly track
what operating model is to be used for a device, separately from the
disable_modern and disable_legacy fields that record the user's
requested configuration.
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20190215103239.28640-2-berrange@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
- block: AioContext management, part 1
- qmp: forbid qmp_cont in RUN_STATE_FINISH_MIGRATE
- nvme: fix copy direction in DMA reads going to CMB
- file-posix: Fix block status for unaligned raw images with O_DIRECT
- file-posix: Fix xfs_write_zeroes() after EOF
- Documentation and iotests improvements
-----BEGIN PGP SIGNATURE-----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=kdOA
-----END PGP SIGNATURE-----
Merge remote-tracking branch 'remotes/kevin/tags/for-upstream' into staging
Block layer patches:
- block: AioContext management, part 1
- qmp: forbid qmp_cont in RUN_STATE_FINISH_MIGRATE
- nvme: fix copy direction in DMA reads going to CMB
- file-posix: Fix block status for unaligned raw images with O_DIRECT
- file-posix: Fix xfs_write_zeroes() after EOF
- Documentation and iotests improvements
# gpg: Signature made Mon 20 May 2019 16:12:38 BST
# gpg: using RSA key 7F09B272C88F2FD6
# gpg: Good signature from "Kevin Wolf <kwolf@redhat.com>" [full]
# Primary key fingerprint: DC3D EB15 9A9A F95D 3D74 56FE 7F09 B272 C88F 2FD6
* remotes/kevin/tags/for-upstream: (24 commits)
iotests: Make 245 faster and more reliable
iotests.py: Fix VM.run_job
iotests.py: Let assert_qmp() accept an array
block: Improve "Block node is read-only" message
qemu-img.texi: Describe human-readable info output
qemu-img.texi: Be specific about JSON object types
iotests: Test unaligned raw images with O_DIRECT
block/file-posix: Unaligned O_DIRECT block-status
test-block-iothread: Test AioContext propagation for block jobs
blockjob: Remove AioContext notifiers
blockjob: Propagate AioContext change to all job nodes
block: Add blk_set_allow_aio_context_change()
block: Implement .(can_)set_aio_ctx for BlockBackend
test-block-iothread: Test AioContext propagation through the tree
block: Propagate AioContext change to parents
block: Move recursion to bdrv_set_aio_context()
block: Make bdrv_attach/detach_aio_context() static
block: Add bdrv_try_set_aio_context()
nvme: fix copy direction in DMA reads going to CMB
iotest: fix 169: do not run qmp_cont in RUN_STATE_FINISH_MIGRATE
...
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Sometimes, 245 fails for me because some stream job has already finished
while the test expects it to still be active. (With -c none, it fails
basically every time.) The most reliable way to fix this is to simply
set auto_finalize=false so the job will remain in the block graph as
long as we need it. This allows us to drop the rate limiting, too,
which makes the test faster.
The only problem with this is that there is a single place that yields a
different error message depending on whether the stream job is still
copying data (so COR is enabled) or not (COR has been disabled, but the
job still has the WRITE_UNCHANGED permission on the target node). We
can easily address that by expecting either error message.
Note that we do not need auto_finalize=false (or rate limiting) for the
active commit job, because It never completes without an explicit
block-job-complete anyway.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
log() is in the current module, there is no need to prefix it. In fact,
doing so may make VM.run_job() unusable in tests that never use
iotests.log() themselves.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Sometimes we cannot tell which error message qemu will emit, and we do
not care. With this change, we can then just pass an array of all
possible messages to assert_qmp() and it will choose the right one.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This message does not make any sense when it appears as the response to
making an R/W node read-only. We should detect that case and emit a
different message, then.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Ideally, it should be self-explanatory. However, keys like "disk size"
arguably really are not self-explanatory. In any case, there is no harm
in going into a some more detail here.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Just writing that --output=json outputs JSON information does not really
help; we should also make a note of what QAPI type the result object
has. (The map subcommand does not emit a QAPI-typed object, but its
section already describes the object structure well enough.)
Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We already have 221 for accesses through the page cache, but it is
better to create a new file for O_DIRECT instead of integrating those
test cases into 221. This way, we can make use of
_supported_cache_modes (and _default_cache_mode) so the test is
automatically skipped on filesystems that do not support O_DIRECT.
As part of the split, add _supported_cache_modes to 221. With that, it
no longer fails when run with -c none or -c directsync.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Currently, qemu crashes whenever someone queries the block status of an
unaligned image tail of an O_DIRECT image:
$ echo > foo
$ qemu-img map --image-opts driver=file,filename=foo,cache.direct=on
Offset Length Mapped to File
qemu-img: block/io.c:2093: bdrv_co_block_status: Assertion `*pnum &&
QEMU_IS_ALIGNED(*pnum, align) && align > offset - aligned_offset'
failed.
This is because bdrv_co_block_status() checks that the result returned
by the driver's implementation is aligned to the request_alignment, but
file-posix can fail to do so, which is actually mentioned in a comment
there: "[...] possibly including a partial sector at EOF".
Fix this by rounding up those partial sectors.
There are two possible alternative fixes:
(1) We could refuse to open unaligned image files with O_DIRECT
altogether. That sounds reasonable until you realize that qcow2
does necessarily not fill up its metadata clusters, and that nobody
runs qemu-img create with O_DIRECT. Therefore, unpreallocated qcow2
files usually have an unaligned image tail.
(2) bdrv_co_block_status() could ignore unaligned tails. It actually
throws away everything past the EOF already, so that sounds
reasonable.
Unfortunately, the block layer knows file lengths only with a
granularity of BDRV_SECTOR_SIZE, so bdrv_co_block_status() usually
would have to guess whether its file length information is inexact
or whether the driver is broken.
Fixing what raw_co_block_status() returns is the safest thing to do.
There seems to be no other block driver that sets request_alignment and
does not make sure that it always returns aligned values.
Cc: qemu-stable@nongnu.org
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The notifiers made sure that the job is quiesced and that the
job->aio_context field is updated. The first part is unnecessary today
since bdrv_set_aio_context_ignore() drains the block node, and this
means drainig the block job, too. The second part can be done in the
.set_aio_ctx callback of the block job BdrvChildRole.
The notifiers were problematic because they poll the AioContext while
the graph is in an inconsistent state with some nodes already in the new
context, but others still in the old context. So removing the notifiers
not only simplifies the code, but actually makes the code safer.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Block jobs require that all of the nodes the job is using are in the
same AioContext. Therefore all BdrvChild objects of the job propagate
.(can_)set_aio_context to all other job nodes, so that the switch is
checked and performed consistently even if both nodes are in different
subtrees.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Some users (like block jobs) can tolerate an AioContext change for their
BlockBackend. Add a function that tells the BlockBackend that it can
allow changes.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
bdrv_try_set_aio_context() currently fails if a BlockBackend is attached
to a node because it doesn't implement the BdrvChildRole callbacks for
AioContext management.
We can allow changing the AioContext of monitor-owned BlockBackends as
long as no device is attached to them.
When setting the AioContext of the root node of a BlockBackend, we now
need to pass blk->root as an ignored child because we don't want the
root node to recursively call back into BlockBackend and execute
blk_do_set_aio_context() a second time.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
All block nodes and users in any connected component of the block graph
must be in the same AioContext, so changing the AioContext of one node
must not only change all of its children, but all of its parents (and
in turn their children etc.) as well.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Instead of having two recursions, in bdrv_attach_aio_context() and in
bdrv_detach_aio_context(), just having one recursion is enough. Said
functions are only about a single node now.
It is important that the recursion doesn't happen between detaching and
attaching a context to the current node because the nested call will
drain the node, and draining with a NULL context would segfault.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Since commit b97511c7bc, there is no reason for block drivers any more
to call these functions (see the function comment in block_int.h). They
are now just internal helper functions for bdrv_set_aio_context()
and can be made static.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Eventually, we want to make sure that all parents and all children of a
node are in the same AioContext as the node itself. This means that
changing the AioContext may fail because one of the other involved
parties (e.g. a guest device that was configured with an iothread)
cannot allow switching to a different AioContext.
Introduce a set of functions that allow to first check whether all
involved nodes can switch to a new context and only then do the actual
switch. The check recursively covers children and parents.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
