Message ID | 1401801062-9154-16-git-send-email-kwolf@redhat.com |
---|---|
State | New |
Headers | show |
The Tuesday 03 Jun 2014 à 15:10:56 (+0200), Kevin Wolf wrote : > Some code in the block layer makes potentially huge allocations. Failure > is not completely unexpected there, so avoid aborting qemu and handle > out-of-memory situations gracefully. > > This patch addresses the allocations in the rbd block driver. > > Signed-off-by: Kevin Wolf <kwolf@redhat.com> > Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> > --- > block/rbd.c | 7 +++++-- > 1 file changed, 5 insertions(+), 2 deletions(-) > > diff --git a/block/rbd.c b/block/rbd.c > index 09af484..d0b2329 100644 > --- a/block/rbd.c > +++ b/block/rbd.c > @@ -623,7 +623,7 @@ static BlockDriverAIOCB *rbd_start_aio(BlockDriverState *bs, > RBDAIOCmd cmd) > { > RBDAIOCB *acb; > - RADOSCB *rcb; > + RADOSCB *rcb = NULL; > rbd_completion_t c; > int64_t off, size; > char *buf; > @@ -637,7 +637,10 @@ static BlockDriverAIOCB *rbd_start_aio(BlockDriverState *bs, > if (cmd == RBD_AIO_DISCARD || cmd == RBD_AIO_FLUSH) { > acb->bounce = NULL; > } else { > - acb->bounce = qemu_blockalign(bs, qiov->size); > + acb->bounce = qemu_try_blockalign(bs, qiov->size); I am under the impression that acb->bounce will be leaked in next goto failed. > + if (acb->bounce == NULL) { > + goto failed; > + } > } > acb->ret = 0; > acb->error = 0; > -- > 1.8.3.1 > >
Am 03.06.2014 um 17:43 hat Benoît Canet geschrieben: > The Tuesday 03 Jun 2014 à 15:10:56 (+0200), Kevin Wolf wrote : > > Some code in the block layer makes potentially huge allocations. Failure > > is not completely unexpected there, so avoid aborting qemu and handle > > out-of-memory situations gracefully. > > > > This patch addresses the allocations in the rbd block driver. > > > > Signed-off-by: Kevin Wolf <kwolf@redhat.com> > > Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> > > --- > > block/rbd.c | 7 +++++-- > > 1 file changed, 5 insertions(+), 2 deletions(-) > > > > diff --git a/block/rbd.c b/block/rbd.c > > index 09af484..d0b2329 100644 > > --- a/block/rbd.c > > +++ b/block/rbd.c > > @@ -623,7 +623,7 @@ static BlockDriverAIOCB *rbd_start_aio(BlockDriverState *bs, > > RBDAIOCmd cmd) > > { > > RBDAIOCB *acb; > > - RADOSCB *rcb; > > + RADOSCB *rcb = NULL; > > rbd_completion_t c; > > int64_t off, size; > > char *buf; > > @@ -637,7 +637,10 @@ static BlockDriverAIOCB *rbd_start_aio(BlockDriverState *bs, > > if (cmd == RBD_AIO_DISCARD || cmd == RBD_AIO_FLUSH) { > > acb->bounce = NULL; > > } else { > > - acb->bounce = qemu_blockalign(bs, qiov->size); > > + acb->bounce = qemu_try_blockalign(bs, qiov->size); > I am under the impression that acb->bounce will be leaked in next goto failed. Yes, I think you're right. That's a preexisting problem, though. I'll fix it in a patch independent from this series. Kevin
diff --git a/block/rbd.c b/block/rbd.c index 09af484..d0b2329 100644 --- a/block/rbd.c +++ b/block/rbd.c @@ -623,7 +623,7 @@ static BlockDriverAIOCB *rbd_start_aio(BlockDriverState *bs, RBDAIOCmd cmd) { RBDAIOCB *acb; - RADOSCB *rcb; + RADOSCB *rcb = NULL; rbd_completion_t c; int64_t off, size; char *buf; @@ -637,7 +637,10 @@ static BlockDriverAIOCB *rbd_start_aio(BlockDriverState *bs, if (cmd == RBD_AIO_DISCARD || cmd == RBD_AIO_FLUSH) { acb->bounce = NULL; } else { - acb->bounce = qemu_blockalign(bs, qiov->size); + acb->bounce = qemu_try_blockalign(bs, qiov->size); + if (acb->bounce == NULL) { + goto failed; + } } acb->ret = 0; acb->error = 0;