On 23/11/2015 16:59, Kevin Wolf wrote:
> +bs->explicit_options = qdict_clone_shallow(options);
qdict_clone_shallow dereferences options, and other parts of
bdrv_open_inherit assume options != NULL. You can remove the first part
of this condition, some 50 lines below:
/* Check if any
On 01/08/2016 07:08 AM, P J P wrote:
> From: Prasad J Pandit
>
> When processing NCQ commands, ACHI device emulation prepares a
> NCQ transfer object; To which an aio control block(aiocb) object
> is assigned in 'execute_ncq_command'. In case, when the NCQ
> command is
+-- On Fri, 8 Jan 2016, John Snow wrote --+
| In both of these error pathways, AIOCB is actually never assigned to
| begin with.
True, it's mentioned in the commit message.
| So it's not necessarily a use-after-free.
Yes, right.
| I think it should be safe to put ncq_tfs->used = 0
add ahci_exec, which is a standard purpose flexible command dispatcher
and tester for the AHCI device. The intent is to eventually cut down on
the absurd amount of boilerplate inside of the AHCI qtest.
Signed-off-by: John Snow
---
tests/libqos/ahci.c | 76
Add pathways to tolerate ATAPI commands.
Notably, unlike ATA, each SCSI command's layout is a little different,
so support will have to be patched in for each command as we want to
test them in e.g. ahci_command_set_sizes and ahci_command_set_offset.
For now, I'm adding support for 0x28, READ
On 01/08/2016 02:48 PM, P J P wrote:
> From: Prasad J Pandit
>
> When processing NCQ commands, ACHI device emulation prepares a
> NCQ transfer object; To which an aio control block(aiocb) object
> is assigned in 'execute_ncq_command'. In case, when the NCQ
> command is
As part of streamlining the AHCI tests interface, it'd be nice
if specying a size of zero could be handled without special branches
and the allocator could handle this special case gracefully.
This lets me use the "ahci_io" macros for non-data commands, too,
which moves me forward towards
Use the proper free command to detroy an AHCICommand.
Signed-off-by: John Snow
---
tests/ahci-test.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/tests/ahci-test.c b/tests/ahci-test.c
index 0888506..f4945dc 100644
--- a/tests/ahci-test.c
+++
Add ATAPI support into libqos/ahci, and write a few tests for it.
This is the last "batch" of planned qtests for s/ata -- basic i/o
testing of HDDs and CDROMs on both PCI and AHCI should be complete
after this series.
v2: Rebase and resend for 2.6.
Organize the prototypes into nice little sections.
Signed-off-by: John Snow
---
tests/libqos/ahci.h | 36
1 file changed, 24 insertions(+), 12 deletions(-)
diff --git a/tests/libqos/ahci.h b/tests/libqos/ahci.h
index 2c2d2fc..69dc4d7
ATAPI commands are, unfortunately, weird in that they can
be either DMA or PIO depending on a header bit. In order to
accommodate them, I'll need to make AHCI command properties
mutable so we can toggle between which "flavor" of ATAPI command
we want to test.
The default ATAPI transfer mechanism
We need to say "hello!" to our ATAPI friends
in a slightly different manner.
Signed-off-by: John Snow
---
tests/ahci-test.c | 8 +++-
tests/libqos/ahci.c | 5 +
tests/libqos/ahci.h | 1 +
3 files changed, 13 insertions(+), 1 deletion(-)
diff --git
Simple I/O tests for DMA and PIO pathways in the AHCI HBA.
I believe at this point in time all of the common, major IO pathways
in BMDMA and AHCI are covered by qtests now.
Signed-off-by: John Snow
---
tests/ahci-test.c | 97
On 01/08/2016 02:55 PM, John Snow wrote:
> Add ATAPI support into libqos/ahci, and write a few tests for it.
>
> This is the last "batch" of planned qtests for s/ata -- basic i/o
> testing of HDDs and CDROMs on both PCI and AHCI should be complete
> after this series.
>
> v2: Rebase and resend
These variants try to set a data offset, even if you don't specify one.
In the cases where the offset is zero and it's a nondata command, just
ignore the instruction.
Signed-off-by: John Snow
---
tests/ahci-test.c | 14 ++
tests/libqos/ahci.c | 3 +++
2 files
On 01/08/2016 02:48 PM, P J P wrote:
> From: Prasad J Pandit
>
> When processing NCQ commands, ACHI device emulation prepares a
> NCQ transfer object; To which an aio control block(aiocb) object
> is assigned in 'execute_ncq_command'. In case, when the NCQ
> command is
On 7 January 2016 at 22:57, Max Reitz wrote:
> Kevin is on PTO this week, so I am filling in for him.
>
>
> The following changes since commit a7e00e2536941a6e570b45b7ab4afec4505ff67e:
>
> petalogix-ml605: Set the MicroBlaze CPU version to 8.10.a (2016-01-07
> 14:57:26
+-- On Fri, 8 Jan 2016, John Snow wrote --+
| > ide_state->status = READY_STAT | ERR_STAT;
| > ncq_tfs->drive->port_regs.scr_err |= (1 << ncq_tfs->tag);
| > +ncq_tfs->used = 0;
| > }
|
| Thanks, applied to my IDE tree:
|
| https://github.com/jnsnow/qemu/commits/ide
|
On 7 January 2016 at 22:43, Max Reitz wrote:
> I hope that the above explanation helped you understand why it bled into
> tray-less devices, from a technical perspective.
Yes, thanks, that was definitely a helpful explanation for why
the design is the way it is. I'm still not
From: Prasad J Pandit
When processing NCQ commands, ACHI device emulation prepares a
NCQ transfer object; To which an aio control block(aiocb) object
is assigned in 'execute_ncq_command'. In case, when the NCQ
command is invalid, the 'aiocb' object is not assigned, and
On Thu, Jan 7, 2016 at 10:09 AM, Peter Maydell wrote:
> On 20 December 2015 at 20:51, Peter Crosthwaite
> wrote:
>> On Sun, Dec 20, 2015 at 9:10 AM, Peter Maydell
>> wrote:
>>> For user-level back compat I think we
From: Roman Kagan
Factor out and expose the function to locate the floppy controller in
the system.
It will allow to dynamically populate the relevant objects in the ACPI
tables.
Signed-off-by: Roman Kagan
Cc: "Michael S. Tsirkin"
On Wed, Dec 30, 2015 at 01:49:25PM +0800, Fam Zheng wrote:
> In preparation for an async implementation, introduce a callback and
> move the shutdown/close to the function.
>
> Signed-off-by: Fam Zheng
> ---
> blockdev-nbd.c | 5 ++---
> include/block/nbd.h | 6 --
>
Instead of delaying blk_detach_dev and blockdev_auto_del until
the object is finalized and properties are released, do that
as soon as possible.
This patch replaces blockdev_mark_auto_del calls with blk_detach_dev
and blockdev_del_drive (the latter is a combination of the former
In short, this patch gets rid of blockdev_mark_auto_del and
blockdev_auto_del.
With these patches, it is possible to create a new -drive with the same
id as soon as the DEVICE_DELETED event is delivered (which equals to
unrealize).
I'm sorry I'm not able to explain the history (and probably do
Currently, blockdev_del_drive does a blk_unref (and before it
blockdev_auto_del did the same) that will cause blk_delete to be called
and the DriveInfo to be freed. But really, we want to free the drive info
as soon as the device is detached, even if there are other references
for whatever
26 matches
Mail list logo