Someone forgot something in commit 29c1a730... Documenting the right
return value is not enough, you also need to actually return it in the
code.
This bug sometimes causes error return values even when everything has
succeeded: The new offset of the refcount block is truncated to 32 bits
and interpreted as signed. At least with small cluster sizes it's easy
to get a negative return value this way.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
If do_alloc_cluster_offset() fails, the error handling code tried to
remove the request from the in-flight queue, to which it wasn't added
yet, resulting in a NULL pointer dereference.
m->nb_clusters really only becomes != 0 when the request is in the list.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
* kwolf/for-anthony: (46 commits)
qed: remove incoming live migration blocker
qed: honor BDRV_O_INCOMING for incoming live migration
migration: clear BDRV_O_INCOMING flags on end of incoming live migration
qed: add bdrv_invalidate_cache to be called after incoming live migration
blockdev: open images with BDRV_O_INCOMING on incoming live migration
block: add a function to clear incoming live migration flags
block: Add new BDRV_O_INCOMING flag to notice incoming live migration
block stream: close unused files and update ->backing_hd
qemu-iotests: Fix call syntax for qemu-io
qemu-iotests: Fix call syntax for qemu-img
qemu-iotests: Test unknown qcow2 header extensions
qemu-iotests: qcow2.py
sheepdog: fix send req helpers
sheepdog: implement SD_OP_FLUSH_VDI operation
block: bdrv_append() fixes
qed: track dirty flag status
qemu-img: add dirty flag status
qed: image fragmentation statistics
qemu-img: add image fragmentation statistics
block: document job API
...
From original commit with Patchwork-id: 31108 by
Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
"The QED image format includes a file header bit to mark images dirty.
QED normally checks dirty images on open and fixes inconsistent
metadata. This is undesirable during live migration since the dirty bit
may be set if the source host is modifying the image file. The check
should be postponed until migration completes.
Skip operations that modify the image file if the BDRV_O_INCOMING flag
is set."
Signed-off-by: Benoit Canet <benoit.canet@gmail.com>
Reviewed-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The QED image is reopened to flush metadata and check consistency.
Signed-off-by: Benoit Canet <benoit.canet@gmail.com>
Reviewed-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Close the now unused images that were part of the previous backing file
chain and adjust ->backing_hd, backing_filename and backing_format
properly.
Fixes https://bugzilla.redhat.com/show_bug.cgi?id=801449
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We should return if reading of the header fails.
Cc: Kevin Wolf <kwolf@redhat.com>
Cc: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Signed-off-by: Liu Yuan <tailai.ly@taobao.com>
Acked-by: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Flush operation is supposed to flush the write-back cache of
sheepdog cluster.
By issuing flush operation, we can assure the Guest of data
reaching the sheepdog cluster storage.
Cc: Kevin Wolf <kwolf@redhat.com>
Cc: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Signed-off-by: Liu Yuan <tailai.ly@taobao.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Dong Xu Wang <wdongxu@linux.vnet.ibm.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Dong Xu Wang <wdongxu@linux.vnet.ibm.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
There is no need to do this in every implementation of set_speed
(even though there is only one right now).
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Streaming can issue I/O while qcow2_close is running. This causes the
L2 caches to become very confused or, alternatively, could cause a
segfault when the streaming coroutine is reentered after closing its
block device. The fix is to cancel streaming jobs when closing their
underlying device.
The cancellation must be synchronous, on the other hand qemu_aio_wait
will not restart a coroutine that is sleeping in co_sleep. So add
a flag saying whether streaming has in-flight I/O. If the busy flag
is false, the coroutine is quiescent and, when cancelled, will not
issue any new I/O.
This protects streaming against closing, but not against deleting.
We have a reference count protecting us against concurrent deletion,
but I still added an assertion to ensure nothing bad happens.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Finally reindent all code and change goto statements to a loop.
Acked-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reads and writes to the underlying file can also occur with the simple
non-vectored I/O interfaces.
Acked-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
vdi.c really works as if it implemented bdrv_read and bdrv_write. However,
because only vector I/O is supported by the asynchronous callbacks, it
went through extra pain to bounce-buffer the I/O. This can be handled
by the block layer now that the format is coroutine-based.
Acked-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Most of the AIOCB really holds local variables that need to persist
across callback invocation. It can go away now.
Acked-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Now inline the former AIO callbacks into vdi_co_readv and vdi_co_writev.
While many cleanups are possible, the code now really looks synchronous.
Acked-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The next step is to take code that only triggers after the first operation,
and move it at the end of vdi_aio_read_cb and vdi_aio_write_cb.
Acked-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Even a basic conversion changing the bdrv_aio_readv/bdrv_aio_writev calls
to bdrv_co_readv/bdrv_co_writev, and callbacks to goto statements can
eliminate a lot of code. This is because error handling is simplified
and indirections through bottom halves can go away.
After this patch, I/O to the underlying file already happens via
coroutines, but the code still looks a lot like if asynchronous I/O was
being used.
Acked-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
After validation check, the 'checksum' is not written back
to footer, which leave it with zero.
This results in errors while loadding it under Microsoft's
Hyper-V environment, and also errors from utilities like
Citrix's vhd-util.
Signed-off-by: Zhang Shengju <sean_zhang@trendmicro.com.cn>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Since everything goes through the cache, callers don't use the L2 table
offset any more.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
The function usleep is not available for all supported platforms:
at least some versions of MinGW don't support it.
usleep was also declared obsolete by POSIX.1-2001.
The function g_usleep is part of glib2.0, so it is available for
all supported platforms.
Using nanosleep would also be possible but needs more code.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
If the first part of a write request is allocated, but the second isn't
and it can be allocated so that the resulting area is contiguous, handle
it at once. This is a common case for sequential writes.
After this patch, alloc_cluster_offset() only checks if the clusters are
already allocated or how many new clusters can be allocated contigouosly.
The actual cluster allocation is split off into a new function
do_alloc_cluster_offset().
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
This function allows to allocate clusters at a given offset in the image
file. This is useful if you want to allocate the second part of an area
that must be contiguous.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
qemu-img resize has some limitations with qcow2, but the user is only
told that "this image format does not support resize". Quite confusing,
so add some more detailed error_report() calls and change "this image
format" into "this image".
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
The L2 table cache reduces QED metadata reads that would be required
when translating LBAs to offsets into the image file. Since requests
execute in parallel it is possible to share an L2 table between multiple
requests.
There is a potential data corruption issue when an in-use L2 table is
evicted from the cache because the following situation occurs:
1. An allocating write performs an update to L2 table "A".
2. Another request needs L2 table "B" and causes table "A" to be
evicted.
3. A new read request needs L2 table "A" but it is not cached.
As a result the L2 update from #1 can overlap with the L2 fetch from #3.
We must avoid doing overlapping I/O requests here since the worst case
outcome is that the L2 fetch completes before the L2 update and yields
stale data. In that case we would effectively discard the L2 update and
lose data clusters!
Thanks to Benoît Canet <benoit.canet@gmail.com> for extensive testing
and debugging which lead to discovery of this bug.
Reported-by: Benoît Canet <benoit.canet@gmail.com>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Tested-by: Benoît Canet <benoit.canet@gmail.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Image files that make qemu-img info read several gigabytes into the
unknown header extensions list are bad. Just fail opening the image
if an extension claims to be larger than the header extension area.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
The spec says that the length of extensions is padded to 8 bytes, not
the offset. Currently this is the same because the header size is a
multiple of 8, so this is only about compatibility with future changes
to the header size.
While touching it, move the calculation to a common place instead of
duplicating it for each header extension type.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
The co_recv coroutine has two things that will try to enter it:
1. The select(2) read callback on the sheepdog socket.
2. The aio_add_request() blocking operations, including a coroutine
mutex.
This patch fixes it by setting NULL to co_recv before sending data.
In future, we should make the sheepdog driver fully coroutine-based
and simplify request handling.
Signed-off-by: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
If we want header extensions to work as compatible extensions, we can't
destroy yet unknown header extensions when rewriting the header (e.g.
for changing the backing file). Save all unknown header extensions in a
list of blobs and include them in a new header.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
In order to switch the backing file, qcow2 issues multiple write
requests that only changed a part of the image header. Any failure after
the first one would leave the header in an corrupted state. With this
patch, the whole header is written at once, so we can't fail in the
middle.
At the same time, this gives us a reusable functions that updates all
fields of the qcow2 header and not only the backing file.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The geometry calculation algorithm from the VHD spec rounds the image
size down if it doesn't exactly match a geometry. During image
conversion, this causes the image to be truncated. For dynamic images,
we already have code in place to round up instead, let's do the same for
fixed images.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The Virtual Hard Disk Image Format Specification allows for three
types of hard disk formats, Fixed, Dynamic, and Differencing. Qemu
currently only supports Dynamic disks. This patch adds support for
the Fixed Disk format.
Usage:
Example 1: qemu-img create -f vpc -o type=fixed <filename> [size]
Example 2: qemu-img convert -O vpc -o type=fixed <input filename> <output filename>
While it is also allowed to specify '-o type=dynamic', the default disk type
remains Dynamic and is what is used when the type is left unspecified.
Signed-off-by: Charles Arnold <carnold@suse.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This patch adds configuration variables for iSCSI to set
initiator-name to use when logging in to the target,
which type of header-digest to negotiate with the target
and username and password for CHAP authentication.
This allows specifying a initiator-name either from the command line
-iscsi initiator-name=iqn.2004-01.com.example:test
or from a configuration file included with -readconfig
[iscsi]
initiator-name = iqn.2004-01.com.example:test
header-digest = CRC32C|CRC32C-NONE|NONE-CRC32C|NONE
user = CHAP username
password = CHAP password
If you use several different targets, you can also configure this on a per
target basis by using a group name:
[iscsi "iqn.target.name"]
...
The configuration file can be read using -readconfig.
Example :
qemu-system-i386 -drive file=iscsi://127.0.0.1/iqn.ronnie.test/1
-readconfig iscsi.conf
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Zero writes are a dedicated interface for writing regions of zeroes into
the image file. If clusters are not yet allocated it is possible to use
an efficient metadata representation which keeps the image file compact
and does not store individual zero bytes.
Implementing this for the QED image format is fairly straightforward.
The only issue is that when a zero write touches an existing cluster we
have to allocate a bounce buffer and perform a regular write.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Per-request attributes like read/write are currently implemented as bool
fields in the QEDAIOCB struct. This becomes unwiedly as the number of
attributes grows. For example, the qed_aio_setup() function would have
to take multiple bool arguments and at call sites it would be hard to
distinguish the meaning of each bool.
Instead use a flags field with bitmask constants. This will be used
when zero write support is added.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Since common file operation functions lack of error detection and use
much more I/O syscalls, so change them to bdrv series functions and
reduce I/O request.
Signed-off-by: Li Zhi Hui <zhihuili@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The new block was filled with zero when it was allocated by g_malloc0,
but when it was reused later and only partially used, data from the
previously allocated block were still present and written to the new
block.
This caused the problems reported by bug #919242
(https://bugs.launchpad.net/qemu/+bug/919242).
Now the unused parts of the new block which are before and after the data
are always filled with zero, so it is no longer necessary to zero the whole
block with g_malloc0.
I also updated the copyright comment.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Add support for streaming data from an intermediate section of the
image chain (see patch and documentation for details).
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This patch implements rate-limiting for image streaming. If we've
exceeded the bandwidth quota for a 100 ms time slice we sleep the
coroutine until the next slice begins.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Most of the codebase as been converted to use glib memory allocation
functions. There are still a few instances of malloc/calloc in the
block layer and qemu-io. Replace them, especially since they do not
check the strdup/malloc/calloc return value.
Reported-by: Dr David Alan Gilbert <davidagilbert@uk.ibm.com>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
All files under GPLv2 will get GPLv2+ changes starting tomorrow.
event_notifier.c and exec-obsolete.h were only ever touched by Red Hat
employees and can be relicensed now.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Allow sending up to 16 requests, and drive the replies to the coroutine
that did the request. The code is written to be exactly the same as
before this patch when MAX_NBD_REQUESTS == 1 (modulo the extra mutex
and state).
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
qemu-nbd has a limit of slightly less than 1M per request. Work
around this in the nbd block driver.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Outside coroutines, avoid busy waiting on EAGAIN by temporarily
making the socket blocking.
The API of qemu_recvv/qemu_sendv is slightly different from
do_readv/do_writev because they do not handle coroutines. It
returns the number of bytes written before encountering an
EAGAIN. The specificity of yielding on EAGAIN is entirely in
qemu-coroutine.c.
Reviewed-by: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The caller expects psn_tab to be NULL when there are no snapshots or
an error occurs. This results in calling g_free on an invalid address.
Reported-by: Oliver Francke <Oliver@filoo.de>
Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Li Zhi Hui <zhihuili@linux.vnet.ibm.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Initially done with the following semantic patch:
@ rule1 @
expression E;
statement S;
@@
E = qemu_aio_get (...);
(
- if (E == NULL) { ... }
|
- if (E)
{ <... S ...> }
)
which however missed occurrences in linux-aio.c and posix-aio-compat.c.
Those were done by hand.
The change in vdi_aio_setup's caller was also done by hand.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Initially done with the following semantic patch:
@ rule1 @
expression E;
statement S;
@@
E =
(
bdrv_aio_readv
| bdrv_aio_writev
| bdrv_aio_flush
| bdrv_aio_discard
| bdrv_aio_ioctl
)
(...);
(
- if (E == NULL) { ... }
|
- if (E)
{ <... S ...> }
)
which however missed the occurrence in block/blkverify.c
(as it should have done), and left behind some unused
variables.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Double semicolons should be single.
Signed-off-by: Dong Xu Wang <wdongxu@linux.vnet.ibm.com>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Now that bdrv_co_is_allocated() is available we can use it instead of
the synchronous bdrv_is_allocated() interface. This is a follow-up that
Kevin Wolf <kwolf@redhat.com> pointed out after applying the series that
introduces bdrv_co_is_allocated().
It is safe to make cow_read() a coroutine_fn because its only caller is
a coroutine_fn.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
It's common to wake up all waiting coroutines. Introduce the
qemu_co_queue_restart_all() function to do this instead of looping over
qemu_co_queue_next() in every caller.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The cow block driver does not keep internal state for cluster lookups.
This means it is safe to perform cluster lookups in coroutine context
without risk of race conditions that corrupt internal state.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
It is trivial to switch from the synchronous .bdrv_is_allocated()
interface to .bdrv_co_is_allocated() since vdi_is_allocated() does not
block.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
It is trivial to switch from the synchronous .bdrv_is_allocated()
interface to .bdrv_co_is_allocated() since vvfat_is_allocated() does not
block.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The qcow2, qcow, and vmdk block drivers are based on coroutines. They have a
coroutine mutex which protects internal state. We can convert the
.bdrv_is_allocated() function to .bdrv_co_is_allocated() by holding the mutex
around the cluster lookup operation.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The bdrv_qed_is_allocated() function is a synchronous wrapper around
qed_find_cluster(), which performs the cluster lookup. In order to
convert the synchronous function to a coroutine function we yield
instead of using qemu_aio_wait(). Note that QED's cache is already safe
for parallel requests so no locking is needed.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
If the bdrv_read() of the snapshot's L1 table fails, return the right
error code and make sure that the old L1 table is still loaded and we
don't break the BlockDriverState completely.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
First the snapshot must be deleted and only then the refcounts can be
decreased.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
The refcount updates must be moved so that in the worst case we can get
cluster leaks, but refcounts may never be too low.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Besides fixing the return code, this adds some comments that make clear
how the code works and that it potentially breaks images if we fail in
the wrong place. Actually fixing this is left for the next patch.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Increase refcounts only after allocating a new L1 table has succeeded in
order to make leaks less likely. If writing the snapshot table fails,
revert in-memory state to be consistent with that on disk.
While at it, make it return the real error codes instead of -1.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
sn->id_str could be leaked before this. The rest of this patch changes
comments, fixes coding style or removes checks that are unnecessary with
g_malloc.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Failing in the middle wouldn't help with the integrity of the image, so
doing everything in a single request seems better.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Doesn't immediately fix anything as the callers don't use the return
value, but they will be fixed next.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Looks better when reviewing these source files.
Signed-off-by: Dong Xu Wang <wdongxu@linux.vnet.ibm.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Since common file operation functions lack of error detection,
so change them to bdrv series functions.
Signed-off-by: Li Zhi Hui <zhihuili@linux.vnet.ibm.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This patch is only to refactor some lines of codes to get better and more robust codes.
As you have seen, in qed_read_table_cb() it's nice to
use qiov->size because that function doesn't obviously use a single
struct iovec.
In other two functions, if qiov use more than one struct iovec, the existing way will get wrong nb_sectors.
To make the code more robust, it will be nicer to refactor the existing way as below.
Signed-off-by: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com>
Acked-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
A BlockDriverState should not issue requests on itself through the
public block layer interface. Nested, or reentrant, requests are
problematic because they do I/O throttling and request tracking twice.
Features like block layer copy-on-read use request tracking to avoid
race conditions between concurrent requests. The reentrant request will
have to "wait" for its parent request to complete. But the parent is
waiting for the reentrant request to make progress so we have reached
deadlock.
The solution is for block drivers to avoid the public block layer
interfaces for reentrant requests. Instead they should call their own
internal functions if they wish to perform reentrant requests.
This is also a good opportunity to make copy_sectors() a true
coroutine_fn. That means calling bdrv_co_writev() instead of
bdrv_write(). Behavior is unchanged but we're being explicit that this
executes in coroutine context.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Unlocking during COW allows for more parallelism. One change it requires is
that buffers are dynamically allocated instead of just using a per-image
buffer.
While touching the code, drop the synchronous qcow2_read() function and replace
it by a bdrv_read() call.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Cc: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Dong Xu Wang <wdongxu@linux.vnet.ibm.com>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
vvfat caches more or less everything when in writable mode. For migration
to work, it would have to be invalidated. Block migration for now when
in writable mode (default is readonly).
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
vdi caches the block map. For migration to work, it would have to be
invalidated. Block migration for now.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
s->lock should be unlocked before leaving add_aio_request.
Signed-off-by: Dong Xu Wang <wdongxu@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
zlib.h is not a local include file, therefore it should be included
using <> instead of "".
Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Now when you try to migrate with qed, you get:
(qemu) migrate tcp:localhost:1025
Block format 'qed' used by device 'ide0-hd0' does not support feature 'live migration'
(qemu)
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
We don't reopen the actual file, but instead invoke the close and open routines.
We specifically ignore the backing file since it's contents are read-only and
therefore immutable.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
qcow2 has a writeback metadata cache, so flushing a qcow2 image actually
consists of writing back that cache to the protocol and only then flushes the
protocol in order to get everything stable on disk.
This introduces a separate bdrv_co_flush_to_os to reflect the split.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
There are two different types of flush that you can do: Flushing one level up
to the OS (i.e. writing data to the host page cache) or flushing it all the way
down to the disk. The existing functions flush to the disk, reflect this in the
function name.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The Data Offset field in the Dynamic Disk Header is an 8 byte field.
Although the specification (2006-10-11) gives an example of initializing
only the first 4 bytes, images generated by Microsoft on Windows initialize
all 8 bytes.
Failure to initialize all 8 bytes results in errors from utilities
like Citrix's vhd-util which checks specifically for the proper Data
Offset field initialization.
Signed-off-by: Charles Arnold <carnold@suse.com>
Reviewed-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
vvfat used to directly call into the qcow2 block driver instead of using the
block.c wrappers. With the coroutine conversion, this stopped working.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
First determine FAT12/16/32, then compute geometry from that for both
FDD and HDD. For 1.44MB floppies, and 2.88MB floppies using FAT16,
change to 1 sector/cluster. The default remains 2.88MB with FAT12
and 2 sectors/cluster. Both DOS and mkdosfs by default format a 2.88MB
floppy as FAT12.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The sector count is stored in the partition and hence must not include the
sectors before its start. At the same time, remove the useless special
casing for 1.44 MB floppies. This fixes fsck on VVFAT hard disks,
which otherwise tries to seek past the end of the disk.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This is consistent with what "real" floppies have, so file(1)
now actually recognizes the VVFAT image as a 1.44 MB floppy.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
If the number of "faked sectors" + the number of sectors that are
part of a cluster does not sum up to the total number of sectors,
qemu-img convert fails. Read these spare sectors as all zeros.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
When reading the address of the first free entry, you cannot
use array_get without first marking all entries as occupied.
This is visible if you change the sectors per cluster on a
floppy from 2 to 1.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Fix mismatching allocation and deallocation: g_free should be used to pair with
g_malloc.
Reviewed-by: Andreas Färber <afaerber@suse.de>
Reviewed_by: Ray Wang <raywang@linux.vnet.ibm.com>
Signed-off-by: Dong Xu Wang <wdongxu@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Fix coding style in block/cloop.c.
Reviewed-by: Andreas Färber <afaerber@suse.de>
Reviewed_by: Ray Wang <raywang@linux.vnet.ibm.com>
Signed-off-by: Dong Xu Wang <wdongxu@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
If qcow2_cache_flush failed, s->lock will not be unlock.
Signed-off-by: Dong Xu Wang <wdongxu@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Data we read from the disk isn't necessarily null terminated and may not
contain the string we're looking for. The code needs to be a bit more careful
here.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
An entry in the VDI block map will hold an offset to the actual block if
the block is allocated, or one of two specially-interpreted values if
not allocated. Using VirtualBox terminology, value VDI_IMAGE_BLOCK_FREE
(0xffffffff) represents a never-allocated block (semantically arbitrary
content). VDI_IMAGE_BLOCK_ZERO (0xfffffffe) represents a "discarded"
block (semantically zero-filled). block/vdi knows only about
VDI_IMAGE_BLOCK_FREE. Teach it about VDI_IMAGE_BLOCK_ZERO.
Signed-off-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This provides built-in support for iSCSI to QEMU.
This has the advantage that the iSCSI devices need not be made visible to the host, which is useful if you have very many virtual machines and very many iscsi devices.
It also has the benefit that non-root users of QEMU can access iSCSI devices across the network without requiring root privilege on the host.
This driver interfaces with the multiplatform posix library for iscsi initiator/client access to iscsi devices hosted at
git://github.com/sahlberg/libiscsi.git
The patch adds the driver to interface with the iscsi library.
It also updated the configure script to
* by default, probe is libiscsi is available and if so, build
qemu against libiscsi.
* --enable-libiscsi
Force a build against libiscsi. If libiscsi is not available
the build will fail.
* --disable-libiscsi
Do not link against libiscsi, even if it is available.
When linked with libiscsi, qemu gains support to access iscsi resources such as disks and cdrom directly, without having to make the devices visible to the host.
You can specify devices using a iscsi url of the form :
iscsi://[<username>[:<password>@]]<host>[:<port]/<target-iqn-name>/<lun>
When using authentication, the password can optionally be set with
LIBISCSI_CHAP_PASSWORD="password" to avoid it showing up in the process list
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
'ret' is unconditionally overwitten by qed_read_l1_table_sync()
Spotted by Clang Analyzer
Signed-off-by: Pavel Borzenkov <pavel.borzenkov@gmail.com>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Spotted by Clang Analyzer
[Note this memcpy call has always been safe because the length will be 0
when the pointer is NULL]
Signed-off-by: Pavel Borzenkov <pavel.borzenkov@gmail.com>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Since coroutine operation is now mandatory, convert both bdrv_discard
implementations to coroutines. For qcow2, this means taking the lock
around the operation. raw-posix remains synchronous.
The bdrv_discard callback is then unused and can be eliminated.
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Since coroutine operation is now mandatory, convert all bdrv_flush
implementations to coroutines. For qcow2, this means taking the lock.
Other implementations are simpler and just forward bdrv_flush to the
underlying protocol, so they can avoid the lock.
The bdrv_flush callback is then unused and can be eliminated.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This does the first part of the conversion to coroutines, by
wrapping bdrv_write implementations to take the mutex.
Drivers that implement bdrv_write rather than bdrv_co_writev can
then benefit from asynchronous operation (at least if the underlying
protocol supports it, which is not the case for raw-win32), even
though they still operate with a bounce buffer.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This does the first part of the conversion to coroutines, by
wrapping bdrv_read implementations to take the mutex.
Drivers that implement bdrv_read rather than bdrv_co_readv can
then benefit from asynchronous operation (at least if the underlying
protocol supports it, which is not the case for raw-win32), even
though they still operate with a bounce buffer.
raw-win32 does not need the lock, because it cannot yield.
nbd also doesn't probably, but better be safe.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The big conversion of bdrv_read/write to coroutines caused the two
homonymous callbacks in BlockDriver to become reentrant. It goes
like this:
1) bdrv_read is now called in a coroutine, and calls bdrv_read or
bdrv_pread.
2) the nested bdrv_read goes through the fast path in bdrv_rw_co_entry;
3) in the common case when the protocol is file, bdrv_co_do_readv calls
bdrv_co_readv_em (and from here goes to bdrv_co_io_em), which yields
until the AIO operation is complete;
4) if bdrv_read had been called from a bottom half, the main loop
is free to iterate again: a device model or another bottom half
can then come and call bdrv_read again.
This applies to all four of read/write/flush/discard. It would also
apply to is_allocated, but it is not used from within coroutines:
besides qemu-img.c and qemu-io.c, which operate synchronously, the
only user is the monitor. Copy-on-read will introduce a use in the
block layer, and will require converting it.
The solution is "simply" to convert all drivers to coroutines! We
just need to add a CoMutex that is taken around affected operations.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Move vmdk_parent_open to vmdk_open. There's another path how
vmdk_parent_open can be reached:
vmdk_parse_extents() -> vmdk_open_sparse() -> vmdk_open_vmdk4() ->
vmdk_open_desc_file().
If that can happen, however, the code is bogus. vmdk_parent_open
reads from bs->file:
if (bdrv_pread(bs->file, s->desc_offset, desc, DESC_SIZE) != DESC_SIZE) {
but it is always called with s->desc_offset == 0 and with the same
bs->file. So the data that vmdk_parent_open reads comes always from the
same place, and anyway there is only one place where it can write it,
namely bs->backing_file.
So, if it cannot happen, the patched code is okay.
It is also possible that the recursive call can happen, but only once. In
that case there would still be a bug in vmdk_open_desc_file setting
s->desc_offset = 0, but the patched code is okay.
Finally, in the case where multiple recursive calls can happen the code
would need to be rewritten anyway. It is likely that this would anyway
involve adding several parameters to vmdk_parent_open, and calling it from
vmdk_open_vmdk4.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
While vmdk_open_desc_file (touched by the patch) correctly changed -1
to -EINVAL, vmdk_open did not. Fix it directly in vmdk_parent_open.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
If during allocation of compressed clusters the cluster was already allocated
uncompressed, fail and properly release the l2_table (the latter avoids a
failed assertion).
While at it, make it return some real error numbers instead of -1.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Dong Xu Wang <wdongxu@linux.vnet.ibm.com>
This similarly adds support for coroutine and asynchronous discard.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Block drivers now only need to provide either of .bdrv_co_flush,
.bdrv_aio_flush() or for legacy drivers .bdrv_flush(). Remove
the redundant .bdrv_flush() implementations.
[Paolo Bonzini: change raw driver to bdrv_co_flush]
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This makes the following patch easier to review.
Cc: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The raw format delegates all operations to bs->file (the protocol).
Previously this block driver exposed both sync and aio interfaces.
Since the block layer now works in terms of coroutines, expose the
coroutine interfaces and drop the others. This avoids unnecessary
emulation of sync and aio interfaces.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Block drivers only need to provide one of sync, aio, or coroutine
interfaces. Since raw-posix.c provides aio interfaces, simply drop the
synchronous interfaces since they can be emulated using aio and
coroutines.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
cppcheck reported this error:
qemu/block/qcow.c:599: error: Mismatching allocation and deallocation: cluster_data
Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Reviewed-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Dong Xu Wang <wdongxu@linux.vnet.ibm.com>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
The unused code was detected using cppcheck.
Cc: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Weil <weil@mail.berlios.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
cppcheck reported memory leaks and mismatched g_malloc() with free()
instead of g_free().
Fix these errors.
Cc: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Weil <weil@mail.berlios.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Allow to resize images that reside on host devices up to the available
space. This allows to grow images after resizing the device manually or
vice versa.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
QED's metadata caching strategy allows two parallel requests to race for
metadata lookup. The first one to complete will populate the metadata
cache and the second one will drop the data it just read in favor of the
cached data.
There is a use-after-free in qed_read_l2_table_cb() and
qed_commit_l2_update() where l2_table->offset was used after the
l2_table may have been freed due to a metadata lookup race. Fix this by
keeping the l2_offset in a local variable and not reaching into the
possibly freed l2_table.
Reported-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>