| Message ID | 1466681836-1123-1-git-send-email-den@openvz.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 06/23/2016 02:37 PM, Denis V. Lunev wrote: > Partial write most likely means that there is not space rather than > "something wrong happens". Thus it would be more natural to return > ENOSPC rather than EINVAL. > > The problem actually happens with NBD server, which has reported EINVAL > rather then ENOSPC on the first error using its protocol, which makes > report to the user wrong. > > Signed-off-by: Denis V. Lunev <den@openvz.org> > CC: Pavel Borzenkov <pborzenkov@virtuozzo.com> > CC: Kevin Wolf <kwolf@redhat.com> > CC: Max Reitz <mreitz@redhat.com> > --- > block/linux-aio.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/block/linux-aio.c b/block/linux-aio.c > index e468960..7df8651 100644 > --- a/block/linux-aio.c > +++ b/block/linux-aio.c > @@ -87,7 +87,7 @@ static void qemu_laio_process_completion(struct qemu_laiocb *laiocb) > qemu_iovec_memset(laiocb->qiov, ret, 0, > laiocb->qiov->size - ret); > } else { > - ret = -EINVAL; > + ret = -ENOSPC; > } > } > } ping
Am 23.06.2016 um 13:37 hat Denis V. Lunev geschrieben: > Partial write most likely means that there is not space rather than > "something wrong happens". Thus it would be more natural to return > ENOSPC rather than EINVAL. > > The problem actually happens with NBD server, which has reported EINVAL > rather then ENOSPC on the first error using its protocol, which makes > report to the user wrong. > > Signed-off-by: Denis V. Lunev <den@openvz.org> > CC: Pavel Borzenkov <pborzenkov@virtuozzo.com> > CC: Kevin Wolf <kwolf@redhat.com> > CC: Max Reitz <mreitz@redhat.com> Thanks, applied to the block branch. Kevin
On 06/28/2016 05:21 AM, Denis V. Lunev wrote: > On 06/23/2016 02:37 PM, Denis V. Lunev wrote: >> Partial write most likely means that there is not space rather than >> "something wrong happens". Thus it would be more natural to return >> ENOSPC rather than EINVAL. >> >> The problem actually happens with NBD server, which has reported EINVAL >> rather then ENOSPC on the first error using its protocol, which makes >> report to the user wrong. I'm not stating that this patch is wrong, but I'm also wondering if NBD code needs to be patched (the NBD protocol specifically recommends ENOSPC rather than EINVAL on attempts to write larger than size). In fact, commit 29b6c3b3 may have been that fix.
On 06/28/2016 05:22 PM, Eric Blake wrote: > On 06/28/2016 05:21 AM, Denis V. Lunev wrote: >> On 06/23/2016 02:37 PM, Denis V. Lunev wrote: >>> Partial write most likely means that there is not space rather than >>> "something wrong happens". Thus it would be more natural to return >>> ENOSPC rather than EINVAL. >>> >>> The problem actually happens with NBD server, which has reported EINVAL >>> rather then ENOSPC on the first error using its protocol, which makes >>> report to the user wrong. > I'm not stating that this patch is wrong, but I'm also wondering if NBD > code needs to be patched (the NBD protocol specifically recommends > ENOSPC rather than EINVAL on attempts to write larger than size). In > fact, commit 29b6c3b3 may have been that fix. > this patch does not patch return code in NBD layer directly. Here the command comes normally to QCOW2 and then below to raw-posix/linux-aio. From the point of low level there is no error at all, the write has partial success status. In this case completion handler raises EINVAL and this code is reported to the caller, i.e. to NBD and through NBD to the client. In terms of NBD the patch altered return code from blk_pwrite in nbd_trip(). The same code will be reported to all other users and thus VM will be properly put to pause if configured on the first failed request without reporting the error to guest. NBD is just an example. Den
diff --git a/block/linux-aio.c b/block/linux-aio.c index e468960..7df8651 100644 --- a/block/linux-aio.c +++ b/block/linux-aio.c @@ -87,7 +87,7 @@ static void qemu_laio_process_completion(struct qemu_laiocb *laiocb) qemu_iovec_memset(laiocb->qiov, ret, 0, laiocb->qiov->size - ret); } else { - ret = -EINVAL; + ret = -ENOSPC; } } }
Partial write most likely means that there is not space rather than "something wrong happens". Thus it would be more natural to return ENOSPC rather than EINVAL. The problem actually happens with NBD server, which has reported EINVAL rather then ENOSPC on the first error using its protocol, which makes report to the user wrong. Signed-off-by: Denis V. Lunev <den@openvz.org> CC: Pavel Borzenkov <pborzenkov@virtuozzo.com> CC: Kevin Wolf <kwolf@redhat.com> CC: Max Reitz <mreitz@redhat.com> --- block/linux-aio.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)