On 8/28/2018 1:50 PM, Leon Romanovsky wrote:
On Tue, Aug 28, 2018 at 01:08:55PM -0400, Dennis Dalessandro wrote:
On 8/28/2018 7:45 AM, Leon Romanovsky wrote:
From: Parav Pandit <parav@xxxxxxxxxxxx>
When AF_IB addresses are used during rdma_resolve_addr(), currently lock
is not held. A cma device can get removed while such list traversal is
in progress which may lead to crash.
Therefore, hold a lock while traversing the list which avoids such
situation.
This patch is not tested. It was discovered during code review.
Cc: <stable@xxxxxxxxxxxxxxx> # 3.10
Fixes: f17df3b0d ("RDMA/cma: Add support for AF_IB to rdma_resolve_addr()")
Signed-off-by: Parav Pandit <parav@xxxxxxxxxxxx>
Reviewed-by: Daniel Jurgens <danielj@xxxxxxxxxxxx>
Signed-off-by: Leon Romanovsky <leonro@xxxxxxxxxxxx>
Really? You are sending something to stable that hasn't been tested?
Why are you saying that?
Thanks,
I know I personally wouldn't let any patches from our drivers hit the
list without being tested, and certainly patches that touch common code
and may impact other drivers would definitely be tested.
And I would guess this probably was tested or at least run through your
QA, in which case, just strike that sentence from the commit message. We
are trying to establish/maintain a good reputation here for our
subsystem and when the casual reader of the git log comes along and sees
this it tells them, hey RDMA folks take anything! Not the message we
want to send I don't think.
-Denny
