Am 01.05.2015 um 01:28 hat Don Slutz geschrieben:
On 04/30/15 16:15, Kevin Wolf wrote:
Am 30.04.2015 um 21:15 hat Eric Blake geschrieben:
[adding qemu-block]
On 04/30/2015 12:23 PM, Don Slutz wrote:
~/qemu/out/master/x86_64-softmmu/qemu-system-x86_64 -h | head
QEMU emulator version
Am 30.04.2015 um 20:07 hat John Snow geschrieben:
libqos.c:
-set_context for addressing which commands go where
-migrate performs the actual migration
malloc.c:
- Structure of the allocator is adjusted slightly with
a second-tier malloc to make swapping around the
Am 30.04.2015 um 20:07 hat John Snow geschrieben:
The day we all feared is here, and I am proposing we allow the migration
of the AHCI device tentatively for the 2.4 development window.
There are some more NCQ migration tests are needed, but I felt that it was
important to get migration
Kevin Wolf kw...@redhat.com writes:
Am 01.05.2015 um 01:28 hat Don Slutz geschrieben:
[...]
So do you want a more complex patch that allows the format to be
specified?
Only for 'disk_image'?
Include -hd* ?
I'm afraid that there is no nice way to improve the plain 'disk_image'
case.
The following sequence
int fd = open(argv[1], O_RDWR | O_CREAT | O_DIRECT, 0644);
for (i = 0; i 10; i++)
write(fd, buf, 4096);
iperforms 5% better if buf is aligned to 4096 bytes rather then to
512 bytes.
I have used the following program to test
#define _GNU_SOURCE
The following sequence
int fd = open(argv[1], O_RDWR | O_CREAT | O_DIRECT, 0644);
for (i = 0; i 10; i++)
write(fd, buf, 4096);
performs 5% better if buf is aligned to 4096 bytes.
The difference is quite reliable.
On the other hand we do not want at the moment to enforce
The patch introduces new concept: minimal memory alignment for bounce
buffers. Original so called optimal value is actually minimal required
value for aligment. It should be used for validation that the IOVec
is properly aligned and bounce buffer is not required.
Though, from the performance
Am 04.05.2015 um 15:39 hat Markus Armbruster geschrieben:
Kevin Wolf kw...@redhat.com writes:
Am 01.05.2015 um 01:28 hat Don Slutz geschrieben:
[...]
So do you want a more complex patch that allows the format to be
specified?
Only for 'disk_image'?
Include -hd* ?
I'm afraid
On 05/04/2015 08:29 AM, Kevin Wolf wrote:
Am 30.04.2015 um 20:07 hat John Snow geschrieben:
The day we all feared is here, and I am proposing we allow the migration
of the AHCI device tentatively for the 2.4 development window.
There are some more NCQ migration tests are needed, but I felt
On 05/04/2015 08:07 AM, Kevin Wolf wrote:
Am 30.04.2015 um 20:07 hat John Snow geschrieben:
libqos.c:
-set_context for addressing which commands go where
-migrate performs the actual migration
malloc.c:
- Structure of the allocator is adjusted slightly with
a
The existing qcow2 metadata overlap detection function used existing
structures to determine the location of the image metadata, from plain
fields such as l1_table_offset and l1_size in the BDRVQcowState, over
image structures in memory such as the L1 table for the L2 tables'
positions, or it even
This is a continuation of previous versions of this series. v2's cover
letter was the most elaborate, which you can find here (includes
benchmarks):
http://lists.nongnu.org/archive/html/qemu-devel/2014-11/msg03430.html
See patch 1 for an explanation of why this series exists and what it
does.
Create and destroy the metadata list on creation and destruction of a
qcow2 BDS, respectively. Skip creation if no overlap checks should be
performed.
Signed-off-by: Max Reitz mre...@redhat.com
---
block/qcow2.c | 11 +++
1 file changed, 11 insertions(+)
diff --git a/block/qcow2.c
Enter the image header into the metadata list to protect it against
accidental modifications.
Signed-off-by: Max Reitz mre...@redhat.com
Reviewed-by: Eric Blake ebl...@redhat.com
---
block/qcow2.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/block/qcow2.c b/block/qcow2.c
index
Keep track of the refcount table in the metadata list to protect it
against accidental modifications.
Signed-off-by: Max Reitz mre...@redhat.com
Reviewed-by: Eric Blake ebl...@redhat.com
---
block/qcow2-refcount.c | 18 ++
block/qcow2.c | 4
2 files changed, 22
Keep track of the inactive L1 tables in the metadata list to protect
them against accidental modifications.
Signed-off-by: Max Reitz mre...@redhat.com
Reviewed-by: Eric Blake ebl...@redhat.com
---
block/qcow2-snapshot.c | 25 +
1 file changed, 25 insertions(+)
diff --git
Keep track of the inactive L2 tables in the metadata list to protect
them against accidental modifications.
Signed-off-by: Max Reitz mre...@redhat.com
---
block/qcow2-refcount.c | 20
block/qcow2-snapshot.c | 43 ---
2 files changed,
Add runtime options to qcow2 to control the size of the structures used
for metadata overlap prevention (one option to control the size of the
bitmap cache, another one to control the total memory size limit).
Signed-off-by: Max Reitz mre...@redhat.com
---
block/qcow2.c | 35
Later, a mechanism to set a limit on how much memory may be used for the
overlap prevention structures will be introduced. If that limit is about
to be exceeded, a QMP event should be emitted. This very event is
specified by this patch.
Signed-off-by: Max Reitz mre...@redhat.com
---
Keep track of the active L1 table in the metadata list to protect it
against accidental modifications.
Signed-off-by: Max Reitz mre...@redhat.com
Reviewed-by: Eric Blake ebl...@redhat.com
---
block/qcow2-cluster.c | 11 +++
block/qcow2-snapshot.c | 10 ++
block/qcow2.c
Keep track of the refcount blocks in the metadata list to protect them
against accidental modifications.
Signed-off-by: Max Reitz mre...@redhat.com
Reviewed-by: Eric Blake ebl...@redhat.com
---
block/qcow2-refcount.c | 38 +-
1 file changed, 37 insertions(+),
Make the static new overlap check function global and drop the old
function.
Signed-off-by: Max Reitz mre...@redhat.com
---
block/qcow2-overlap.c | 8 +---
block/qcow2-refcount.c | 120 -
2 files changed, 2 insertions(+), 126 deletions(-)
diff
This adds an adjustable limit for the total memory usage of the overlap
prevention structures.
Signed-off-by: Max Reitz mre...@redhat.com
---
block/qcow2-overlap.c | 145 +++---
block/qcow2.c | 2 +-
block/qcow2.h | 2 +-
3 files
This patch adds some test cases for the memory limit concerning the
in-memory structures used to detect and prevent accidental metadata
overlaps.
Signed-off-by: Max Reitz mre...@redhat.com
---
tests/qemu-iotests/060 | 222 +
Expose the two new options for controlling the memory usage of the
overlap check implementation via QAPI.
Signed-off-by: Max Reitz mre...@redhat.com
---
qapi/block-core.json | 37 +
1 file changed, 37 insertions(+)
diff --git a/qapi/block-core.json
On 05/04/2015 01:15 PM, Max Reitz wrote:
Later, a mechanism to set a limit on how much memory may be used for the
overlap prevention structures will be introduced. If that limit is about
to be exceeded, a QMP event should be emitted. This very event is
specified by this patch.
On Thu, 04/30 15:34, Kevin Wolf wrote:
Am 27.04.2015 um 15:18 hat Fam Zheng geschrieben:
An unaligned zero write causes NULL deferencing in bdrv_co_do_pwritev. That
path is reachable from bdrv_co_write_zeroes and bdrv_aio_write_zeroes.
You can easily trigger through the former with
For zero write, qiov passed by callers (qemu-io write -z and scsi-disk
write same) is NULL.
Commit fc3959e466 fixed bdrv_co_write_zeroes which is the common case
for this bug, but it still exists in bdrv_aio_write_zeroes. A simpler
fix would be in bdrv_co_do_pwritev which is the NULL dereference
An unaligned zero write causes NULL deferencing in bdrv_co_do_pwritev. That
path is reachable from bdrv_co_write_zeroes and bdrv_aio_write_zeroes.
You can easily trigger through the former with qemu-io, as the test case added
by 61815d6e0aa. For bdrv_aio_write_zeroes, in common cases there's
On 05/04/2015 01:15 PM, Max Reitz wrote:
Expose the two new options for controlling the memory usage of the
overlap check implementation via QAPI.
Signed-off-by: Max Reitz mre...@redhat.com
---
qapi/block-core.json | 37 +
1 file changed, 37
30 matches
Mail list logo