It should be safe to reenter qio_channel_yield() on io/channel read/write
path, so it's safe to reduce in_flight and allow attaching new aio
context. And no problem to allow drain itself: connection attempt is
not a guest request. Moreover, if remote server is down, we can hang
in negotiation, blocking drain section and provoking a dead lock.
How to reproduce the dead lock:
1. Create nbd-fault-injector.conf with the following contents:
[inject-error "mega1"]
event=data
io=readwrite
when=before
2. In one terminal run nbd-fault-injector in a loop, like this:
n=1; while true; do
echo $n; ((n++));
./nbd-fault-injector.py 127.0.0.1:10000 nbd-fault-injector.conf;
done
3. In another terminal run qemu-io in a loop, like this:
n=1; while true; do
echo $n; ((n++));
./qemu-io -c 'read 0 512' nbd://127.0.0.1:10000;
done
After some time, qemu-io will hang trying to drain, for example, like
this:
#3 aio_poll (ctx=0x55f006bdd890, blocking=true) at
util/aio-posix.c:600
#4 bdrv_do_drained_begin (bs=0x55f006bea710, recursive=false,
parent=0x0, ignore_bds_parents=false, poll=true) at block/io.c:427
#5 bdrv_drained_begin (bs=0x55f006bea710) at block/io.c:433
#6 blk_drain (blk=0x55f006befc80) at block/block-backend.c:1710
#7 blk_unref (blk=0x55f006befc80) at block/block-backend.c:498
#8 bdrv_open_inherit (filename=0x7fffba1563bc
"nbd+tcp://127.0.0.1:10000", reference=0x0, options=0x55f006be86d0,
flags=24578, parent=0x0, child_class=0x0, child_role=0,
errp=0x7fffba154620) at block.c:3491
#9 bdrv_open (filename=0x7fffba1563bc "nbd+tcp://127.0.0.1:10000",
reference=0x0, options=0x0, flags=16386, errp=0x7fffba154620) at
block.c:3513
#10 blk_new_open (filename=0x7fffba1563bc "nbd+tcp://127.0.0.1:10000",
reference=0x0, options=0x0, flags=16386, errp=0x7fffba154620) at
block/block-backend.c:421
And connection_co stack like this:
#0 qemu_coroutine_switch (from_=0x55f006bf2650, to_=0x7fe96e07d918,
action=COROUTINE_YIELD) at util/coroutine-ucontext.c:302
#1 qemu_coroutine_yield () at util/qemu-coroutine.c:193
#2 qio_channel_yield (ioc=0x55f006bb3c20, condition=G_IO_IN) at
io/channel.c:472
#3 qio_channel_readv_all_eof (ioc=0x55f006bb3c20, iov=0x7fe96d729bf0,
niov=1, errp=0x7fe96d729eb0) at io/channel.c:110
#4 qio_channel_readv_all (ioc=0x55f006bb3c20, iov=0x7fe96d729bf0,
niov=1, errp=0x7fe96d729eb0) at io/channel.c:143
#5 qio_channel_read_all (ioc=0x55f006bb3c20, buf=0x7fe96d729d28
"\300.\366\004\360U", buflen=8, errp=0x7fe96d729eb0) at
io/channel.c:247
#6 nbd_read (ioc=0x55f006bb3c20, buffer=0x7fe96d729d28, size=8,
desc=0x55f004f69644 "initial magic", errp=0x7fe96d729eb0) at
/work/src/qemu/master/include/block/nbd.h:365
#7 nbd_read64 (ioc=0x55f006bb3c20, val=0x7fe96d729d28,
desc=0x55f004f69644 "initial magic", errp=0x7fe96d729eb0) at
/work/src/qemu/master/include/block/nbd.h:391
#8 nbd_start_negotiate (aio_context=0x55f006bdd890,
ioc=0x55f006bb3c20, tlscreds=0x0, hostname=0x0,
outioc=0x55f006bf19f8, structured_reply=true,
zeroes=0x7fe96d729dca, errp=0x7fe96d729eb0) at nbd/client.c:904
#9 nbd_receive_negotiate (aio_context=0x55f006bdd890,
ioc=0x55f006bb3c20, tlscreds=0x0, hostname=0x0,
outioc=0x55f006bf19f8, info=0x55f006bf1a00, errp=0x7fe96d729eb0) at
nbd/client.c:1032
#10 nbd_client_connect (bs=0x55f006bea710, errp=0x7fe96d729eb0) at
block/nbd.c:1460
#11 nbd_reconnect_attempt (s=0x55f006bf19f0) at block/nbd.c:287
#12 nbd_co_reconnect_loop (s=0x55f006bf19f0) at block/nbd.c:309
#13 nbd_connection_entry (opaque=0x55f006bf19f0) at block/nbd.c:360
#14 coroutine_trampoline (i0=113190480, i1=22000) at
util/coroutine-ucontext.c:173
Note, that the hang may be
triggered by another bug, so the whole case is fixed only together with
commit "block/nbd: on shutdown terminate connection attempt".
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20200727184751.15704-3-vsementsov@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
We are going to implement non-blocking version of
nbd_establish_connection, which for a while will be used only for
nbd_reconnect_attempt, not for nbd_open, so we need to call it
separately.
Refactor nbd_reconnect_attempt in a way which makes next commit
simpler.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20200727184751.15704-2-vsementsov@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Add test for "qemu-img convert -O qcow2 -c" to NBD target. The tests
create a OVA file and write compressed qcow2 disk content directly into
the OVA file via qemu-nbd.
Signed-off-by: Nir Soffer <nsoffer@redhat.com>
Message-Id: <20200727215846.395443-5-nsoffer@redhat.com>
Tested-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Add 2 helpers for measuring and checking images:
- qemu_img_measure()
- qemu_img_check()
Both use --output-json and parse the returned json to make easy to use
in other tests. I'm going to use them in a new test, and I hope they
will be useful in may other tests.
Signed-off-by: Nir Soffer <nsoffer@redhat.com>
Message-Id: <20200727215846.395443-4-nsoffer@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Instead of duplicating the code to wait until the server is ready and
remember to terminate the server and wait for it, make it possible to
use like this:
with qemu_nbd_popen('-k', sock, image):
# Access image via qemu-nbd socket...
Only test 264 used this helper, but I had to modify the output since it
did not consistently when starting and stopping qemu-nbd.
Signed-off-by: Nir Soffer <nsoffer@redhat.com>
Message-Id: <20200727215846.395443-3-nsoffer@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
When converting to qcow2 compressed format, the last step is a special
zero length compressed write, ending in a call to bdrv_co_truncate(). This
call always fails for the nbd driver since it does not implement
bdrv_co_truncate().
For block devices, which have the same limits, the call succeeds since
the file driver implements bdrv_co_truncate(). If the caller asked to
truncate to the same or smaller size with exact=false, the truncate
succeeds. Implement the same logic for nbd.
Example failing without this change:
In one shell start qemu-nbd:
$ truncate -s 1g test.tar
$ qemu-nbd --socket=/tmp/nbd.sock --persistent --format=raw --offset 1536 test.tar
In another shell convert an image to qcow2 compressed via NBD:
$ echo "disk data" > disk.raw
$ truncate -s 1g disk.raw
$ qemu-img convert -f raw -O qcow2 -c disk1.raw nbd+unix:///?socket=/tmp/nbd.sock; echo $?
1
qemu-img failed, but the conversion was successful:
$ qemu-img info nbd+unix:///?socket=/tmp/nbd.sock
image: nbd+unix://?socket=/tmp/nbd.sock
file format: qcow2
virtual size: 1 GiB (1073741824 bytes)
...
$ qemu-img check nbd+unix:///?socket=/tmp/nbd.sock
No errors were found on the image.
1/16384 = 0.01% allocated, 100.00% fragmented, 100.00% compressed clusters
Image end offset: 393216
$ qemu-img compare disk.raw nbd+unix:///?socket=/tmp/nbd.sock
Images are identical.
Fixes: https://bugzilla.redhat.com/1860627
Signed-off-by: Nir Soffer <nsoffer@redhat.com>
Message-Id: <20200727215846.395443-2-nsoffer@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
[eblake: typo fixes]
Signed-off-by: Eric Blake <eblake@redhat.com>
Dr. David Alan Gilbert (1):
ip_stripoptions use memmove
Jindrich Novy (4):
Fix possible infinite loops and use-after-free
Use secure string copy to avoid overflow
Be sure to initialize sockaddr structure
Check lseek() for failure
Marc-André Lureau (2):
util: do not silently truncate
Merge branch 'stable-4.2' into 'stable-4.2'
Philippe Mathieu-Daudé (3):
Fix win32 builds by using the SLIRP_PACKED definition
Fix constness warnings
Remove unnecessary break
Ralf Haferkamp (2):
Drop bogus IPv6 messages
Fix MTU check
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
We are having issues debugging and bisecting this issue that happen
mostly on patchew. Let's make it abort where it failed to gather some
new informations.
Suggested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
* document use of -1 when pci_controller field can't be retrieved for
guest-get-fsinfo
* fix incorrect filesystem type reporting on w32 for guest-get-fsinfo
when a volume is not mounted
-----BEGIN PGP SIGNATURE-----
iQFOBAABCgA4FiEEzqzJ4VU066u4LT+gM1PJzvEItYQFAl8fYGIaHG1kcm90aEBs
aW51eC52bmV0LmlibS5jb20ACgkQM1PJzvEItYTOKgf9G5mM8cXW0MxKevy4Fc1Q
i6+/VGkvAUURxn6PALGl4an8sLA0E2uzgqZr3+IWZY+85DZmJ18tfj4QU9G/M9w0
4ByTksmvVR66ukwR0H6lN38cdqxTxzFqeAyuU2WWucYeyGl1l2QMlES2OUqlcgcq
297/ivLRTnUGGe2Ts5EfCNMS2RGExJxbPhq9ffhNYbas92Pu/e2nm0X5uHJpMHiW
TFw/cQ1EnSxxDtNnKJD7VnC/tlZxKKP/qXwJAgoDxW7ORB1XRxCLrSNkRRLBXMxV
F1VksXS4GAIc50nKTb2qHtexLebAjjrJW5c6iJAjWkKJlxhJbpcvXjo6d31Sri5D
6Q==
=Y/oG
-----END PGP SIGNATURE-----
Merge remote-tracking branch 'remotes/mdroth/tags/qga-pull-2020-07-27-tag' into staging
qemu-ga patch queue for hard-freeze
* document use of -1 when pci_controller field can't be retrieved for
guest-get-fsinfo
* fix incorrect filesystem type reporting on w32 for guest-get-fsinfo
when a volume is not mounted
# gpg: Signature made Tue 28 Jul 2020 00:16:50 BST
# gpg: using RSA key CEACC9E15534EBABB82D3FA03353C9CEF108B584
# gpg: issuer "mdroth@linux.vnet.ibm.com"
# gpg: Good signature from "Michael Roth <flukshun@gmail.com>" [full]
# gpg: aka "Michael Roth <mdroth@utexas.edu>" [full]
# gpg: aka "Michael Roth <mdroth@linux.vnet.ibm.com>" [full]
# Primary key fingerprint: CEAC C9E1 5534 EBAB B82D 3FA0 3353 C9CE F108 B584
* remotes/mdroth/tags/qga-pull-2020-07-27-tag:
qga/qapi-schema: Document -1 for invalid PCI address fields
qga-win: fix "guest-get-fsinfo" wrong filesystem type
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Although qemu as NBD client limits requests to <2G, the NBD protocol
allows clients to send requests almost all the way up to 4G. But
because our block layer is not yet 64-bit clean, we accidentally wrap
such requests into a negative size, and fail with EIO instead of
performing the intended operation.
The bug is visible in modern systems with something as simple as:
$ qemu-img create -f qcow2 /tmp/image.img 5G
$ sudo qemu-nbd --connect=/dev/nbd0 /tmp/image.img
$ sudo blkdiscard /dev/nbd0
or with user-space only:
$ truncate --size=3G file
$ qemu-nbd -f raw file
$ nbdsh -u nbd://localhost:10809 -c 'h.trim(3*1024*1024*1024,0)'
Although both blk_co_pdiscard and blk_pwrite_zeroes currently return 0
on success, this is also a good time to fix our code to a more robust
paradigm that treats all non-negative values as success.
Alas, our iotests do not currently make it easy to add external
dependencies on blkdiscard or nbdsh, so we have to rely on manual
testing for now.
This patch can be reverted when we later improve the overall block
layer to be 64-bit clean, but for now, a minimal fix was deemed less
risky prior to release.
CC: qemu-stable@nongnu.org
Fixes: 1f4d6d18ed
Fixes: 1c6c4bb7f0
Fixes: https://github.com/systemd/systemd/issues/16242
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200722212231.535072-1-eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
[eblake: rework success tests to use >=0]
- Improve handling of various post-copy bitmap migration scenarios. A lost
bitmap should merely mean that the next backup must be full rather than
incremental, rather than abruptly breaking the entire guest migration.
- Associated iotest improvements
-----BEGIN PGP SIGNATURE-----
iQEzBAABCAAdFiEEccLMIrHEYCkn0vOqp6FrSiUnQ2oFAl8fPRkACgkQp6FrSiUn
Q2qanQf/dRTrqZ7/hs8aENySf44o0dBzOLZr+FBcrqEj2sd0c6jPzV2X5CVtnA1v
gBgKJJGLpti3mSeNQDbaXZIQrsesBAuxvJsc6vZ9npDCdMYnK/qPE3Zfw1bx12qR
cb39ba28P4izgs216h92ZACtUewnvjkxyJgN7zfmCJdNcwZINMUItAS183tSbQjn
n39Wb7a+umsRgV9HQv/6cXlQIPqFMyAOl5kkzV3evuw7EBoHFnNq4cjPrUnjkqiD
xf2pcSomaedYd37SpvoH57JxfL3z/90OBcuXhFvbqFk4FgQ63rJ32nRve2ZbIDI0
XPbohnYjYoFv6Xs/jtTzctZCbZ+jTg==
=1dmz
-----END PGP SIGNATURE-----
Merge remote-tracking branch 'remotes/ericb/tags/pull-bitmaps-2020-07-27' into staging
bitmaps patches for 2020-07-27
- Improve handling of various post-copy bitmap migration scenarios. A lost
bitmap should merely mean that the next backup must be full rather than
incremental, rather than abruptly breaking the entire guest migration.
- Associated iotest improvements
# gpg: Signature made Mon 27 Jul 2020 21:46:17 BST
# gpg: using RSA key 71C2CC22B1C4602927D2F3AAA7A16B4A2527436A
# gpg: Good signature from "Eric Blake <eblake@redhat.com>" [full]
# gpg: aka "Eric Blake (Free Software Programmer) <ebb9@byu.net>" [full]
# gpg: aka "[jpeg image of size 6874]" [full]
# Primary key fingerprint: 71C2 CC22 B1C4 6029 27D2 F3AA A7A1 6B4A 2527 436A
* remotes/ericb/tags/pull-bitmaps-2020-07-27: (24 commits)
migration: Fix typos in bitmap migration comments
iotests: Adjust which migration tests are quick
qemu-iotests/199: add source-killed case to bitmaps postcopy
qemu-iotests/199: add early shutdown case to bitmaps postcopy
qemu-iotests/199: check persistent bitmaps
qemu-iotests/199: prepare for new test-cases addition
migration/savevm: don't worry if bitmap migration postcopy failed
migration/block-dirty-bitmap: cancel migration on shutdown
migration/block-dirty-bitmap: relax error handling in incoming part
migration/block-dirty-bitmap: keep bitmap state for all bitmaps
migration/block-dirty-bitmap: simplify dirty_bitmap_load_complete
migration/block-dirty-bitmap: rename finish_lock to just lock
migration/block-dirty-bitmap: refactor state global variables
migration/block-dirty-bitmap: move mutex init to dirty_bitmap_mig_init
migration/block-dirty-bitmap: rename dirty_bitmap_mig_cleanup
migration/block-dirty-bitmap: rename state structure types
migration/block-dirty-bitmap: fix dirty_bitmap_mig_before_vm_start
qemu-iotests/199: increase postcopy period
qemu-iotests/199: change discard patterns
qemu-iotests/199: improve performance: set bitmap by discard
...
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
While 197 is very much a qcow2 test, and it looks like the partial
cluster case at the end (introduced in b0ddcbbb36) is specifically
a qcow2 case, the whole test scripts actually marks itself to work with
generic formats (and generic protocols, even).
Said partial cluster case happened to work with non-qcow2 formats as
well (mostly by accident), but 1855536256 broke that, because it sets
the compat option, which does not work for non-qcow2 formats.
So go the whole way and force IMGFMT=qcow2 and IMGPROTO=file, as done in
other places in this test.
Fixes: 1855536256
("iotests/197: Fix for compat=0.10")
Reported-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
Message-Id: <20200728131134.902519-1-mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Since these functions take a @qiov_offset, they must always take it into
account when working with @qiov. There are a couple of places where
they do not, but they should.
Fixes: 65cd4424b9
("block/io: bdrv_aligned_preadv: use and support qiov_offset")
Fixes: 28c4da2869
("block/io: bdrv_aligned_pwritev: use and support qiov_offset")
Reported-by: Claudio Fontana <cfontana@suse.de>
Reported-by: Bruce Rogers <brogers@suse.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Max Reitz <mreitz@redhat.com>
Message-Id: <20200728120806.265916-2-mreitz@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Tested-by: Claudio Fontana <cfontana@suse.de>
Tested-by: Bruce Rogers <brogers@suse.com>
The memory API allows DMA into NIC's MMIO area. This means the NIC's
RX routine must be reentrant. Instead of auditing all the NIC, we can
simply detect the reentrancy and return early. The queue->delivering
is set and cleared by qemu_net_queue_deliver() for other queue helpers
to know whether the delivering in on going (NIC's receive is being
called). We can check it and return early in qemu_net_queue_flush() to
forbid reentrant RX.
Signed-off-by: Jason Wang <jasowang@redhat.com>
We try to check whether a peer is VDPA in order to get config from
there - with no peer, this leads to a NULL
pointer dereference. Add a check before trying to access the peer
type. No peer means not VDPA.
Fixes: 108a64818e ("vhost-vdpa: introduce vhost-vdpa backend")
Cc: Cindy Lu <lulu@redhat.com>
Tested-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
We should use the index passed by the caller instead of the queue_sel
when checking the enablement of a specific virtqueue.
This is reported in https://bugzilla.redhat.com/show_bug.cgi?id=1702608
Fixes: f19bcdfedd ("virtio-pci: implement queue_enabled method")
Signed-off-by: Yuri Benditovich <yuri.benditovich@daynix.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
The "guest-get-fsinfo" could also be used for non-PCI devices in the
future. And the code in GuestPCIAddress() in qga/commands-win32.c seems
to be using "-1" for fields that it can not determine already. Thus
let's properly document "-1" as value for invalid PCI address fields.
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
This patch handles the case where unmounted volumes exist,
where in that case GetVolumePathNamesForVolumeName returns
empty path, GetVolumeInformation will use the current working
directory instead.
This patch fixes the issue by opening a handle to the volumes,
and using GetVolumeInformationByHandleW instead.
Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=1746667
Signed-off-by: Basil Salman <bsalman@redhat.com>
Signed-off-by: Basil Salman <basil@daynix.com>
*fix crash when guest_build_fsinfo() sets errp multiple times
*make new error message more distinct from existing ones
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Noticed while reviewing the file for newer patches.
Fixes: b35ebdf076
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200727203206.134996-1-eblake@redhat.com>
A quick run of './check -qcow2 -g migration' shows that test 169 is
NOT quick, but meanwhile several other tests ARE quick. Let's adjust
the test designations accordingly.
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200727195117.132151-1-eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Previous patches fixes behavior of bitmaps migration, so that errors
are handled by just removing unfinished bitmaps, and not fail or try to
recover postcopy migration. Add corresponding test.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Tested-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200727194236.19551-22-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Previous patches fixed two crashes which may occur on shutdown prior to
bitmaps postcopy finished. Check that it works now.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Tested-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200727194236.19551-21-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Check that persistent bitmaps are not stored on source and that bitmaps
are persistent on destination.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Message-Id: <20200727194236.19551-20-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Move future common part to start_postcopy() method. Move checking
number of bitmaps to check_bitmap().
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Message-Id: <20200727194236.19551-19-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
First, if only bitmaps postcopy is enabled (and not ram postcopy)
postcopy_pause_incoming crashes on an assertion
assert(mis->to_src_file).
And anyway, bitmaps postcopy is not prepared to be somehow recovered.
The original idea instead is that if bitmaps postcopy failed, we just
lose some bitmaps, which is not critical. So, on failure we just need
to remove unfinished bitmaps and guest should continue execution on
destination.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200727194236.19551-18-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
If target is turned off prior to postcopy finished, target crashes
because busy bitmaps are found at shutdown.
Canceling incoming migration helps, as it removes all unfinished (and
therefore busy) bitmaps.
Similarly on source we crash in bdrv_close_all which asserts that all
bdrv states are removed, because bdrv states involved into dirty bitmap
migration are referenced by it. So, we need to cancel outgoing
migration as well.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Message-Id: <20200727194236.19551-17-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Bitmaps data is not critical, and we should not fail the migration (or
use postcopy recovering) because of dirty-bitmaps migration failure.
Instead we should just lose unfinished bitmaps.
Still we have to report io stream violation errors, as they affect the
whole migration stream.
While touching this, tighten code that was previously blindly calling
malloc on a size read from the migration stream, as a corrupted stream
(perhaps from a malicious user) should not be able to convince us to
allocate an inordinate amount of memory.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20200727194236.19551-16-vsementsov@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
[eblake: typo fixes, enhance commit message]
Signed-off-by: Eric Blake <eblake@redhat.com>
Keep bitmap state for disabled bitmaps too. Keep the state until the
end of the process. It's needed for the following commit to implement
bitmap postcopy canceling.
To clean-up the new list the following logic is used:
We need two events to consider bitmap migration finished:
1. chunk with DIRTY_BITMAP_MIG_FLAG_COMPLETE flag should be received
2. dirty_bitmap_mig_before_vm_start should be called
These two events may come in any order, so we understand which one is
last, and on the last of them we remove bitmap migration state from the
list.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Message-Id: <20200727194236.19551-15-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
bdrv_enable_dirty_bitmap_locked() call does nothing, as if we are in
postcopy, bitmap successor must be enabled, and reclaim operation will
enable the bitmap.
So, actually we need just call _reclaim_ in both if branches, and
making differences only to add an assertion seems not really good. The
logic becomes simple: on load complete we do reclaim and that's all.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Message-Id: <20200727194236.19551-14-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
finish_lock is bad name, as lock used not only on process end.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Message-Id: <20200727194236.19551-13-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Move all state variables into one global struct. Reduce global
variable usage, utilizing opaque pointer where possible.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Message-Id: <20200727194236.19551-12-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
No reasons to keep two public init functions.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20200727194236.19551-11-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Rename dirty_bitmap_mig_cleanup to dirty_bitmap_do_save_cleanup, to
stress that it is on save part.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200727194236.19551-10-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Rename types to be symmetrical for load/save part and shorter.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200727194236.19551-9-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Using the _locked version of bdrv_enable_dirty_bitmap to bypass locking
is wrong as we do not already own the mutex. Moreover, the adjacent
call to bdrv_dirty_bitmap_enable_successor grabs the mutex.
Fixes: 58f72b965e9e1q
Cc: qemu-stable@nongnu.org # v3.0
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200727194236.19551-8-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
The test wants to force a bitmap postcopy. Still, the resulting
postcopy period is very small. Let's increase it by adding more
bitmaps to migrate. Also, test disabled bitmaps migration.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Tested-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200727194236.19551-7-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
iotest 199 works too long because of many discard operations. At the
same time, postcopy period is very short, in spite of all these
efforts.
So, let's use less discards (and with more interesting patterns) to
reduce test timing. In the next commit we'll increase postcopy period.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Tested-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200727194236.19551-6-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Discard dirties dirty-bitmap as well as write, but works faster. Let's
use it instead.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Tested-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200727194236.19551-5-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
The test aims to test _postcopy_ migration, and wants to do some write
operations during postcopy time.
Test considers migrate status=complete event on source as start of
postcopy. This is completely wrong, completion is completion of the
whole migration process. Let's instead consider destination start as
start of postcopy, and use RESUME event for it.
Next, as migration finish, let's use migration status=complete event on
target, as such method is closer to what libvirt or another user will
do, than tracking number of dirty-bitmaps.
Finally, add a possibility to dump events for debug. And if
set debug to True, we see, that actual postcopy period is very small
relatively to the whole test duration time (~0.2 seconds to >40 seconds
for me). This means, that test is very inefficient in what it supposed
to do. Let's improve it in following commits.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Tested-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200727194236.19551-4-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
We don't need any specific format constraints here. Still keep qcow2
for two reasons:
1. No extra calls of format-unrelated test
2. Add some check around persistent bitmap in future (require qcow2)
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Tested-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200727194236.19551-3-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Make the capitalization of the hexadecimal numbers consistent for the
QCOW2 header extension constants in docs/interop/qcow2.txt.
Suggested-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <1594973699-781898-2-git-send-email-andrey.shinkevich@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
The glibc getcwd function returns different errors than the getcwd
syscall, which triggers an assertion failure in the glibc getcwd function
when running under the emulation.
When the syscall returns ENAMETOOLONG, the glibc wrapper uses a fallback
implementation that potentially handles an unlimited path length, and
returns with ERANGE if the provided buffer is too small. The qemu
emulation cannot distinguish the two cases, and thus always returns ERANGE.
This is unexpected by the glibc wrapper.
Signed-off-by: Andreas Schwab <schwab@suse.de>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <mvmmu3qplvi.fsf@suse.de>
[lv: updated description]
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Implementation of 'rt_sigtimedwait()' in 'syscall.c' uses the
function 'target_to_host_timespec()' to transfer the value of
'struct timespec' from target to host. However, the implementation
doesn't check whether this conversion succeeds and thus can cause
an unaproppriate error instead of the 'EFAULT (Bad address)' which
is supposed to be set if the conversion from target to host fails.
This was confirmed with the LTP test for rt_sigtimedwait:
"/testcases/kernel/syscalls/rt_sigtimedwait/rt_sigtimedwait01.c"
which causes an unapropriate error in test case "test_bad_adress3"
which is run with a bad adress for the 'struct timespec' argument:
FAIL: test_bad_address3 (349): Unexpected failure: EAGAIN/EWOULDBLOCK (11)
The test fails with an unexptected errno 'EAGAIN/EWOULDBLOCK' instead
of the expected EFAULT.
After the changes from this patch, the test case is executed successfully
along with the other LTP test cases for 'rt_sigtimedwait()':
PASS: test_bad_address3 (349): Test passed
Signed-off-by: Filip Bozuta <Filip.Bozuta@syrmia.com>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20200724181651.167819-1-Filip.Bozuta@syrmia.com>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
When the chroot does not have /proc mounted, we can read neither
/proc/sys/vm/mmap_min_addr nor /proc/sys/maps.
The enforcement of mmap_min_addr in the host kernel is done by
the security module, and so does not apply to processes owned
by root. Which leads pgd_find_hole_fallback to succeed in probing
a reservation at address 0. Which confuses pgb_reserved_va to
believe that guest_base has not actually been initialized.
We don't actually want NULL addresses to become accessible, so
make sure that mmap_min_addr is initialized with a non-zero value.
Buglink: https://bugs.launchpad.net/qemu/+bug/1888728
Reported-by: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Acked-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20200724212314.545877-1-richard.henderson@linaro.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Minor bugfixes all over the places, including one CVE.
Additionally, a fix for an ancient bug in migration -
one has to wonder how come no one noticed.
The fix is also non-trivial since we dare not break all
existing machine types with pci - we have a work around
in the works, for now we just skip the work-around for
old machine types.
Great job by Hogan Wang noticing, debugging and fixing it,
and thanks to Dr. David Alan Gilbert for reviewing the patches.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
-----BEGIN PGP SIGNATURE-----
iQFDBAABCAAtFiEEXQn9CHHI+FuUyooNKB8NuNKNVGkFAl8e9CIPHG1zdEByZWRo
YXQuY29tAAoJECgfDbjSjVRpsAIH/2EEq9rLpjqMJdzRvjq3/UAHsvm42zeTnJl7
81cM887Mrg2Nd7MXFoxurLK5UEehTzlD2DRTvaDFfJaJlrtkPM2QEU2X/6c3syAS
GbmOQaljQtR4zEFE81t84mZQS025Gp0s+uble7KvtXakgp1A/vdu93OEvJkhtRY8
JBdRMlTt2T0eizvHn1obBKjaQN7tAUKl5KagHWxP1ApGU0YibUbrBadpJ18ZcKMl
vwB3dwmoi4f7AjuC0GnxYKp7kC/MMhUPFoDxQKI7d+wMGFnbsAF4sBIN9EZKeOkv
xT2InNSAzk/PTSuQpnDnZQjmrf4dPuL/GNJ8vQk27eaFfVchJyc=
=Bu6o
-----END PGP SIGNATURE-----
Merge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging
virtio,pci: bugfixes
Minor bugfixes all over the places, including one CVE.
Additionally, a fix for an ancient bug in migration -
one has to wonder how come no one noticed.
The fix is also non-trivial since we dare not break all
existing machine types with pci - we have a work around
in the works, for now we just skip the work-around for
old machine types.
Great job by Hogan Wang noticing, debugging and fixing it,
and thanks to Dr. David Alan Gilbert for reviewing the patches.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# gpg: Signature made Mon 27 Jul 2020 16:34:58 BST
# gpg: using RSA key 5D09FD0871C8F85B94CA8A0D281F0DB8D28D5469
# gpg: issuer "mst@redhat.com"
# gpg: Good signature from "Michael S. Tsirkin <mst@kernel.org>" [full]
# gpg: aka "Michael S. Tsirkin <mst@redhat.com>" [full]
# Primary key fingerprint: 0270 606B 6F3C DF3D 0B17 0970 C350 3912 AFBE 8E67
# Subkey fingerprint: 5D09 FD08 71C8 F85B 94CA 8A0D 281F 0DB8 D28D 5469
* remotes/mst/tags/for_upstream:
virtio-pci: fix virtio_pci_queue_enabled()
MAINTAINERS: Cover the firmware JSON schema
vhost-vdpa :Fix Coverity CID 1430270 / CID 1420267
libvhost-user: Report descriptor index on panic
Fix vhost-user buffer over-read on ram hot-unplug
hw/pci-host: save/restore pci host config register
virtio-mem-pci: force virtio version 1
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Vladimir Sementsov-Ogievskiy
Nir Soffer
Marc-André Lureau
Peter Maydell
Eric Blake
Max Reitz
Jason Wang
Yuri Benditovich
Thomas Huth
Basil Salman
Andrey Shinkevich
Andreas Schwab
Filip Bozuta
Richard Henderson