Reset ->done and ->free when releasing the srb. There is a hidden
use-after-free bug in the driver which corrupts the srb memory pool
which originates from the cleanup callbacks. By explicitly resetting
the callbacks to NULL, we workaround the memory corruption.
An extensive search didn't bring any lights on the real problem. The
initial idea was to set both pointers to NULL and try to catch invalid
accesses. But instead the memory corruption was gone and the driver
didn't crash.
Signed-off-by: Daniel Wagner <[email protected]>
---
drivers/scsi/qla2xxx/qla_inline.h | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/scsi/qla2xxx/qla_inline.h b/drivers/scsi/qla2xxx/qla_inline.h
index 861dc522723c..6d41d758fc17 100644
--- a/drivers/scsi/qla2xxx/qla_inline.h
+++ b/drivers/scsi/qla2xxx/qla_inline.h
@@ -211,6 +211,8 @@ static inline void
qla2xxx_rel_qpair_sp(struct qla_qpair *qpair, srb_t *sp)
{
sp->qpair = NULL;
+ sp->done = NULL;
+ sp->free = NULL;
mempool_free(sp, qpair->srb_mempool);
QLA_QPAIR_MARK_NOT_BUSY(qpair);
}
--
2.16.4
On Thu, 2020-08-27 at 11:58 +0200, Daniel Wagner wrote:
> Reset ->done and ->free when releasing the srb. There is a hidden
> use-after-free bug in the driver which corrupts the srb memory pool
> which originates from the cleanup callbacks. By explicitly resetting
> the callbacks to NULL, we workaround the memory corruption.
>
> An extensive search didn't bring any lights on the real problem. The
> initial idea was to set both pointers to NULL and try to catch
> invalid
> accesses. But instead the memory corruption was gone and the driver
> didn't crash.
>
> Signed-off-by: Daniel Wagner <[email protected]>
> ---
> drivers/scsi/qla2xxx/qla_inline.h | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/drivers/scsi/qla2xxx/qla_inline.h
> b/drivers/scsi/qla2xxx/qla_inline.h
> index 861dc522723c..6d41d758fc17 100644
> --- a/drivers/scsi/qla2xxx/qla_inline.h
> +++ b/drivers/scsi/qla2xxx/qla_inline.h
> @@ -211,6 +211,8 @@ static inline void
> qla2xxx_rel_qpair_sp(struct qla_qpair *qpair, srb_t *sp)
> {
> sp->qpair = NULL;
> + sp->done = NULL;
> + sp->free = NULL;
> mempool_free(sp, qpair->srb_mempool);
> QLA_QPAIR_MARK_NOT_BUSY(qpair);
> }
Both sp->done() and sp->free() are called all over the place without
making sure the pointers are non-null. If these functions can be called
for freed sp's, wouldn't that mean we'd crash?
How about setting them to a dummy function that prints a big fat
warning?
Martin
> How about setting them to a dummy function that prints a big fat
> warning?
Good idea. I'll update the patch accordingly. This might even help to
find the real cause.