Release 6 limits the number of cases where software can cause UNDEFINED or
UNPREDICTABLE behaviour. In this case, when accessing reserved / unimplemented
CP0 register, writes are ignored and reads return 0.
In pre-R6 the behaviour is not specified, but generating RI exception is not
what the real HW does.
Additionally, remove CP0 Random register as it became reserved in Release 6.
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
Reviewed-by: Yongbok Kim <yongbok.kim@imgtec.com>
In Release 6 not all the values are allowed to be written to a register.
If the value is not valid or unsupported then it should stay unchanged.
For pre-R6 the existing behaviour has been changed only for CP0_Index register
as the current implementation does not seem to be correct - it looks like it
tries to limit the input value but the limit is higher than the actual
number of tlb entries.
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
Reviewed-by: Yongbok Kim <yongbok.kim@imgtec.com>
When conditional compact branch is encountered decode one more instruction in
current translation block - that will be forbidden slot. Instruction in
forbidden slot will be executed only if conditional compact branch is not taken.
Any control transfer instruction (CTI) which are branches, jumps, ERET,
DERET, WAIT and PAUSE will generate RI exception if executed in forbidden or
delay slot.
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
Reviewed-by: Yongbok Kim <yongbok.kim@imgtec.com>
SDBBP instruction Reserved Instruction control. The purpose of this field is
to restrict availability of SDBBP to kernel mode operation.
If the bit is set then SDBBP instruction can only be executed in kernel mode.
User execution of SDBBP will cause a Reserved Instruction exception.
Additionally add missing Config4 and Config5 cases for dm{f,t}c0.
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
Reviewed-by: Yongbok Kim <yongbok.kim@imgtec.com>
BadInstr Register (CP0 Register 8, Select 1)
The BadInstr register is a read-only register that capture the most recent
instruction which caused an exception.
BadInstrP Register (CP0 Register 8, Select 2)
The BadInstrP register contains the prior branch instruction, when the
faulting instruction is in a branch delay slot.
Using error_code to indicate whether AdEL or TLBL was triggered during
instruction fetch, in this case BadInstr is not updated as valid instruction
word is not available.
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
Reviewed-by: Yongbok Kim <yongbok.kim@imgtec.com>
For Standard TLB configuration (Config.MT=1):
TLBINV invalidates a set of TLB entries based on ASID. The virtual address is
ignored in the entry match. TLB entries which have their G bit set to 1 are not
modified.
TLBINVF causes all entries to be invalidated.
Single TLB entry can be marked as invalid on TLB entry write by having
EntryHi.EHINV set to 1.
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
Reviewed-by: Yongbok Kim <yongbok.kim@imgtec.com>
An Execute-Inhibit exception occurs when the virtual address of an instruction
fetch matches a TLB entry whose XI bit is set. This exception type can only
occur if the XI bit is implemented within the TLB and is enabled, this is
denoted by the PageGrain XIE bit.
An Read-Inhibit exception occurs when the virtual address of a memory load
reference matches a TLB entry whose RI bit is set. This exception type can
only occur if the RI bit is implemented within the TLB and is enabled, this is
denoted by the PageGrain RIE bit.
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
Reviewed-by: Yongbok Kim <yongbok.kim@imgtec.com>
PageGrain needs rw bitmask which differs between MIPS architectures.
In pre-R6 if RIXI is supported, PageGrain.XIE and PageGrain.RIE are writeable,
whereas in R6 they are read-only 1.
On MIPS64 mtc0 instruction left shifts bits 31:30 for MIPS32 backward
compatiblity, therefore there are separate mtc0 and dmtc0 helpers.
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
Reviewed-by: Yongbok Kim <yongbok.kim@imgtec.com>
In Revision 3 of the architecture, the RI and XI bits were added to the TLB
to enable more secure access of memory pages. These bits (along with the Dirty
bit) allow the implementation of read-only, write-only, no-execute access
policies for mapped pages.
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
Reviewed-by: Yongbok Kim <yongbok.kim@imgtec.com>
New MIPS features depend on the access type and enum is more convenient than
using the numbers directly.
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
Reviewed-by: Thomas Huth <thuth@linux.vnet.ibm.com>
KScratch<n> Registers (CP0 Register 31, Selects 2 to 7)
The KScratch registers are read/write registers available for scratch pad
storage by kernel mode software. They are 32-bits in width for 32-bit
processors and 64-bits for 64-bit processors.
CP0Config4.KScrExist[2:7] bits indicate presence of CP0_KScratch1-6 registers.
For Release 6, all KScratch registers are required.
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
Reviewed-by: Yongbok Kim <yongbok.kim@imgtec.com>
Now that blockjobs use AioContext they are safe for use with dataplane.
Unblock them!
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-id: 1413889440-32577-12-git-send-email-stefanha@redhat.com
The commit block job must run in the BlockDriverState AioContext so that
it works with dataplane.
Acquire the AioContext in blockdev.c so starting the block job is safe.
One detail here is that the bdrv_drain_all() must be moved inside the
aio_context_acquire() region so requests cannot sneak in between the
drain and acquire.
The completion code in block/commit.c must perform backing chain
manipulation and bdrv_reopen() from the main loop. Use
block_job_defer_to_main_loop() to achieve that.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-id: 1413889440-32577-11-git-send-email-stefanha@redhat.com
The mirror block job must run in the BlockDriverState AioContext so that
it works with dataplane.
Acquire the AioContext in blockdev.c so starting the block job is safe.
Note that to_replace is treated separately from other BlockDriverStates
in that it does not need to be in the same AioContext. Explicitly
acquire/release to_replace's AioContext when accessing it.
The completion code in block/mirror.c must perform BDS graph
manipulation and bdrv_reopen() from the main loop. Use
block_job_defer_to_main_loop() to achieve that.
The bdrv_drain_all() call is not allowed outside the main loop since it
could lead to lock ordering problems. Use bdrv_drain(bs) instead
because we have acquired the AioContext so nothing else can sneak in
I/O.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-id: 1413889440-32577-10-git-send-email-stefanha@redhat.com
The stream block job must run in the BlockDriverState AioContext so that
it works with dataplane.
The basics of acquiring the AioContext are easy in blockdev.c.
The tricky part is the completion code which drops part of the backing
file chain. This must be done in the main loop where bdrv_unref() and
bdrv_close() are safe to call. Use block_job_defer_to_main_loop() to
achieve that.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-id: 1413889440-32577-9-git-send-email-stefanha@redhat.com
The backup block job must run in the BlockDriverState AioContext so that
it works with dataplane.
The basics of acquiring the AioContext are easy in blockdev.c.
The completion code in block/backup.c must call bdrv_unref() from the
main loop. Use block_job_defer_to_main_loop() to achieve that.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-id: 1413889440-32577-8-git-send-email-stefanha@redhat.com
Now that op blockers are in use, we can ensure that no other sources are
generating I/O on a BlockDriverState. Therefore it is possible to drain
requests for a single BDS.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-id: 1413889440-32577-7-git-send-email-stefanha@redhat.com
Block jobs will run in the BlockDriverState's AioContext, which may not
always be the QEMU main loop.
There are some block layer APIs that are either not thread-safe or risk
lock ordering problems. This includes bdrv_unref(), bdrv_close(), and
anything that calls bdrv_drain_all().
The block_job_defer_to_main_loop() API allows a block job to schedule a
function to run in the main loop with the BlockDriverState AioContext
held.
This function will be used to perform cleanup and backing chain
manipulations in block jobs.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-id: 1413889440-32577-6-git-send-email-stefanha@redhat.com
This function is correct but we should document the constraint that
everything must be thread-safe.
Emitting QMP events and scheduling BHs are both thread-safe so nothing
needs to be done here.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-id: 1413889440-32577-5-git-send-email-stefanha@redhat.com
When an emulated storage controller is unrealized it will call
blockdev_mark_auto_del(). This will cancel any running block job (and
that eventually releases its reference to the BDS so it can be freed).
Since the block job may be executing in another AioContext we must
acquire/release to ensure thread safety.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-id: 1413889440-32577-4-git-send-email-stefanha@redhat.com
Make sure that query-block-jobs acquires the BlockDriverState
AioContext so that the blockjob isn't running in another thread while we
access its state.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-id: 1413889440-32577-3-git-send-email-stefanha@redhat.com
block-job-set-speed, block-job-cancel, block-job-pause,
block-job-resume, and block-job-complete must acquire the
BlockDriverState AioContext so that it is safe to access bs.
At the moment bs->job is always NULL when dataplane is active because op
blockers prevent blockjobs from starting. Once the rest of the blockjob
API has been made aware of AioContext we can drop the op blocker.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-id: 1413889440-32577-2-git-send-email-stefanha@redhat.com
Add some tests for progress output to 061.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Benoît Canet <benoit.canet@nodalink.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Message-id: 1414404776-4919-8-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Currently, we have a bitmap for keeping track of which clusters have
been created during the zero cluster expansion process. This was
necessary because we need to properly increase the refcount for shared
L2 tables.
However, now we can simply take the L2 refcount and use it for the
cluster allocated for expansion. This will be the correct refcount and
therefore we don't have to remember that cluster having been allocated
any more.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Benoît Canet <benoit.canet@nodalink.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Message-id: 1414404776-4919-7-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reading the refcount of a cluster is an operation which can be useful in
all of the qcow2 code, so make that function globally available.
While touching this function, amend the comment describing the "addend"
parameter: It is (no longer, if it ever was) necessary to have it set to
-1 or 1; any value is fine.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Benoît Canet <benoit.canet@nodalink.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Message-id: 1414404776-4919-6-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
The only really time-consuming operation potentially performed by
qcow2_amend_options() is zero cluster expansion when downgrading qcow2
images from compat=1.1 to compat=0.10, so report status of that
operation and that operation only through the status CB.
For this, approximate the progress as the number of L1 entries visited
during the operation.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Benoît Canet <benoit.canet@nodalink.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Message-id: 1414404776-4919-5-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
As soon as options is set in img_amend(), it needs to be freed before
the function returns. This leak is rather insignificant, as qemu-img
will exit subsequently anyway, but there's no point in not fixing it.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Benoît Canet <benoit.canet@nodalink.com>
Message-id: 1414404776-4919-4-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Now that bdrv_amend_options() supports a status callback, use it to
display a progress report.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Benoît Canet <benoit.canet@nodalink.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-id: 1414404776-4919-3-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Depending on the changed options and the image format,
bdrv_amend_options() may take a significant amount of time. In these
cases, a way to be informed about the operation's status is desirable.
Since the operation is rather complex and may fundamentally change the
image, implementing it as AIO or a coroutine does not seem feasible. On
the other hand, implementing it as a block job would be significantly
more difficult than a simple callback and would not add benefits other
than progress report to the amending operation, because it should not
actually be run as a block job at all.
A callback may not be very pretty, but it's very easy to implement and
perfectly fits its purpose here.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Benoît Canet <benoit.canet@nodalink.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-id: 1414404776-4919-2-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
As discussed during review a follow up for Max's fix.
Signed-off-by: Peter Lieven <pl@kamp.de>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-id: 1414249537-29257-1-git-send-email-pl@kamp.de
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Add a test for qcow2's fast bdrv_make_empty implementation on images
without internal snapshots.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 1414159063-25977-15-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Add a test for qemu-img commit on backing chains with more than two
images. This test also checks whether the top image is emptied (unless
this is prevented by specifying either -d or -b) and does therefore not
work for qed and vmdk which requires it to be separate from 020.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-id: 1414159063-25977-14-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
As different image formats most probably map guest addresses to
different host addresses, add a filter to filter the host addresses out;
also, the image filename should be filtered.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-id: 1414159063-25977-13-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Introduce a new parameter for qemu-img commit which may be used to
explicitly specify the backing file into which an image should be
committed if the backing chain has more than a single layer.
[Applied Eric Blake's qemu-img.texi documentation rewording
--Stefan]
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 1414159063-25977-12-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Implement progress output for the commit command by querying the
progress of the block job.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-id: 1414159063-25977-11-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
After the top image has been committed, it should be emptied unless
specified otherwise.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-id: 1414159063-25977-10-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
qemu-img should use QMP commands whenever possible in order to ensure
feature completeness of both online and offline image operations. As
qemu-img itself has no access to QMP (since this would basically require
just everything being linked into qemu-img), imitate QMP's
implementation of block-commit by using commit_active_start() and then
waiting for the block job to finish.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 1414159063-25977-9-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Instead of taking the total length of the block device as the block
job's length, use the number of dirty sectors. The progress is now the
number of sectors mirrored to the target block device. Note that this
may result in the job's length increasing during operation, which is
however in fact desirable.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-id: 1414159063-25977-8-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
As of a follow-up patch to this one, the length of a mirror block job
will no longer directly depend on the size of the block device;
therefore, drop these checks from this test. Instead, just check whether
the final offset equals the block job length.
As 041 uses the wait_until_completed function from iotests.py, the same
applies there as well which in turn affects tests 030, 055 and 056. On
the other hand, a block job's length does not have to be related to the
length of the image file in the first place, so that check was
questionable anyway.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 1414159063-25977-7-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
When a block job signals readiness, this is currently reported only
through QMP. If qemu wants to use block jobs for internal tasks, there
needs to be another way to correctly detect when a block job may be
completed.
For this reason, introduce a bool "ready" which is set when the block
job may be completed.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-id: 1414159063-25977-6-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Implement block_job_complete_sync() by doing the exact same thing as
block_job_cancel_sync() does, only with calling block_job_complete()
instead of block_job_cancel().
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 1414159063-25977-5-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
bdrv_make_empty() is currently only called if the current image
represents an external snapshot that has been committed to its base
image; it is therefore unlikely to have internal snapshots. In this
case, bdrv_make_empty() can be greatly sped up by emptying the L1 and
refcount table (while having the dirty flag set, which only works for
compat=1.1) and creating a trivial refcount structure.
If there are snapshots or for compat=0.10, fall back to the simple
implementation (discard all clusters).
[Applied s/clusters/cluster/ typo fix suggested by Eric Blake
--Stefan]
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 1414159063-25977-4-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Implement this function by making all clusters in the image file fall
through to the backing file (by using the recently extended discard).
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-id: 1414159063-25977-3-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Normally, discarded sectors should read back as zero. However, there are
cases in which a sector (or rather cluster) should be discarded as if
they were never written in the first place, that is, reading them should
fall through to the backing file again.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-id: 1414159063-25977-2-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
It should not be happening, but it is possible to truncate an image
outside of qemu while qemu is running (or any of the qemu tools using
the block layer. raw_co_get_block_status() should not break then.
While touching this test, replace the existing "truncate" invocation by
"$QEMU_IMG convert -f raw".
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 1414148280-17949-4-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Instead of generating the full return value thrice in try_fiemap(),
try_seek_hole() and as a fall-back in raw_co_get_block_status() itself,
generate the value only in raw_co_get_block_status().
While at it, also remove the pnum parameter from try_fiemap() and
try_seek_hole().
Suggested-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-id: 1414148280-17949-3-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
As its comment states, raw_co_get_block_status() should unconditionally
return 0 and set *pnum to 0 for after EOF.
An assertion after lseek(..., SEEK_HOLE) tried to catch this case by
asserting that errno != -ENXIO (which would indicate a position after
the EOF); but it should be errno != ENXIO instead. Regardless of that,
there should be no such assertion at all. If bdrv_getlength() returned
an outdated value and the image has been resized outside of qemu,
lseek() will return with errno == ENXIO. Just return that value as an
error then.
Setting *pnum to 0 and returning 0 should not be done here, as in that
case we should update the device length as well. So, from qemu's
perspective, the file has not been resized; it's just that there was an
error querying sectors beyond a certain point (the actual file size).
Additionally, nb_sectors should be clamped against the image end. This
was probably not an issue if FIEMAP or SEEK_HOLE/SEEK_DATA worked, but
the fallback did not take this case into account.
Reported-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-id: 1414148280-17949-2-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
