FYI, this causes segfaults when doing large streaming writes when
running against a sheepdog cluster which:
a) has relatively fast SSDs
and
b) uses buffered I/O.
Unfortunately I can't get a useful backtrace out of gdb. When running just
this commit I at least get some debugging messages:
Am 23.08.2011 19:14, schrieb MORITA Kazutaka:
> At Tue, 23 Aug 2011 14:29:50 +0200,
> Kevin Wolf wrote:
>>
>> Am 12.08.2011 14:33, schrieb MORITA Kazutaka:
>>> This makes the sheepdog block driver support bdrv_co_readv/writev
>>> instead of bdrv_aio_readv/writev.
>>>
>>> With this patch, Sheepdog n
At Tue, 23 Aug 2011 14:29:50 +0200,
Kevin Wolf wrote:
>
> Am 12.08.2011 14:33, schrieb MORITA Kazutaka:
> > This makes the sheepdog block driver support bdrv_co_readv/writev
> > instead of bdrv_aio_readv/writev.
> >
> > With this patch, Sheepdog network I/O becomes fully asynchronous. The
> > bl
Am 12.08.2011 14:33, schrieb MORITA Kazutaka:
> This makes the sheepdog block driver support bdrv_co_readv/writev
> instead of bdrv_aio_readv/writev.
>
> With this patch, Sheepdog network I/O becomes fully asynchronous. The
> block driver yields back when send/recv returns EAGAIN, and is resumed
This makes the sheepdog block driver support bdrv_co_readv/writev
instead of bdrv_aio_readv/writev.
With this patch, Sheepdog network I/O becomes fully asynchronous. The
block driver yields back when send/recv returns EAGAIN, and is resumed
when the sheepdog network connection is ready for the op
