Re: [PATCH for-rc] IB/hfi1, qib: Ensure RCU is locked when accessing list

Dennis Dalessandro <dennis.dalessandro@xxxxxxxxx> · Mon, 2 Mar 2020 09:52:03 -0500

On 3/2/2020 8:29 AM, Jason Gunthorpe wrote:
On Mon, Mar 02, 2020 at 08:14:52AM -0500, Dennis Dalessandro wrote:
On 2/28/2020 11:15 AM, Jason Gunthorpe wrote:
On Tue, Feb 25, 2020 at 02:54:45PM -0500, Dennis Dalessandro wrote:
The packet handling function, specifically the iteration of the qp list
for mad packet processing misses locking RCU before running through the
list. Not only is this incorrect, but the list_for_each_entry_rcu() call
can not be called with a conditional check for lock dependency. Remedy
this by invoking the rcu lock and unlock around the critical section.

This brings MAD packet processing in line with what is done for non-MAD
packets.

Cc: Madhuparna Bhowmik <madhuparnabhowmik04@xxxxxxxxx>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@xxxxxxxxx>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@xxxxxxxxx>
   drivers/infiniband/hw/hfi1/verbs.c    |    4 +++-
   drivers/infiniband/hw/qib/qib_verbs.c |    2 ++
   2 files changed, 5 insertions(+), 1 deletion(-)

Applied to for-next, thanks

Maybe it should have went to -rc?

It doesn't even have a fixes line. If you want patches in -rc send
better commit message. I keep repeating this again and again..

Jason

Crap, yeah my bad on that. However there really isn't a good fixes line 
for this. It's pretty much just the initial commit of the driver, does 
that really help? It's still just in your WIP branch so is it too late 
to add:

Fixes: 7724105686e7 ("IB/hfi1: add driver files")

Other than a fixes line what else do you need for the commit message? 
Messed up locking is pretty self explanatory I would think.

-Denny