On Tue, Jun 23, 2015 at 01:44:55PM +0300, Dimitris Aragiorgis wrote:
> Hi all,
>
> These five patches make slight changes to the way QEMU handles SCSI
> generic devices to fix a number of small problems.
>
> I am sending them against the master branch, since I don't know if they
> can be considered bugfixes.
>
> Thanks,
> dimara
>
> v5 (rebased to current master):
> * First use stat() and then ioctl() in bdrv_is_sg() (Eric's comment)
> * Make bdrv_is_sg() return a bool and not an int (Eric's comment)
>
> v4 (rebased to current master):
> * Avoid errno clobbering with DPRINTF + strerror() (Eric's comment)
> * Use {} in #define macro even if it is not necessary (single commands)
>
> v3 (rebased to current master):
> * Avoid bit-rot in DPRINTF (adopt Eric's suggestion)
> * Address Kevin's comments (DEBUG_FLOPPY, line > 80 chars, SG device)
> * Mention Kevin's comment wrt disk flush in the corresponding commit
>
> v2:
> * remove duplicate check for sg inside iscsi_co_flush()
> * remove DEBUG_BLOCK_PRINT in block/raw-posix.c
> * use DPRINTF for debugging in block/raw-posix.c
>
> PS: Paolo suggested to use a tracepoint inside hdev_is_sg() but I chose
> DPRINTF
> instead. It would make sense to add a tracepoint for bdrv_is_sg() (just like
> most bdrv_* commands) but this is too much for now since it just returns the
> bs->sg flag (and is not an actual driver function). If you insist I'll change
> it in v3.
>
> Dimitris Aragiorgis (5):
> block: Use bdrv_is_sg() everywhere
> Fix migration in case of scsi-generic
> raw-posix: DPRINTF instead of DEBUG_BLOCK_PRINT
> raw-posix: Use DPRINTF for DEBUG_FLOPPY
> raw-posix: Introduce hdev_is_sg()
>
> block.c | 6 ++--
> block/io.c | 3 +-
> block/iscsi.c | 4 ---
> block/raw-posix.c | 91
> +++++++++++++++++++++++++++++++----------------------
> 4 files changed, 58 insertions(+), 46 deletions(-)Thanks, applied to my block tree: https://github.com/stefanha/qemu/commits/block Stefan
pgp2tgjyM1BCm.pgp
Description: PGP signature
