| Message ID | 20190326204331.54352-1-bvanassche@acm.org (mailing list archive) |
|---|---|
| Headers | show |
| Series | Avoid that .queuecommand() gets called for a quiesced SCSI device | expand |
On Wed, Mar 27, 2019 at 4:44 AM Bart Van Assche <bvanassche@acm.org> wrote: > > Hi Martin, > > As explained in a recent LSF/MM topic proposal, it can happen that the SCSI > error handler calls .queuecommand() for quiesced SCSI devices. SCSI LLDs do > not expect this. Hence this patch series. Please consider this series for > kernel v5.2. > > Thanks, > > Bart. > > Bart Van Assche (2): > scsi: Avoid that .queuecommand() gets called for a quiesced SCSI > device > RDMA/srp: Fix a sleep-in-invalid-context bug > > drivers/infiniband/ulp/srp/ib_srp.c | 21 ++------------------- > drivers/scsi/scsi_error.c | 26 ++++++++++++++++++++++++-- > 2 files changed, 26 insertions(+), 21 deletions(-) > > -- > 2.21.0.196.g041f5ea1cf98 > Given it is a IB/SRP specific issue, could you consider the following one-line fix? https://marc.info/?l=linux-block&m=151585814931471&w=2 Thanks, Ming Lei
On Wed, Mar 27, 2019 at 09:58:16AM +0800, Ming Lei wrote: > Given it is a IB/SRP specific issue, could you consider the following > one-line fix? It isn't really a IB/SRP specific issue, it is just that the SRP maintainer actually tried to fix this race that others often ignored. And checking for context in queuecommand is a horrible hack that I'd like to get rid of.
On Wed, 2019-03-27 at 09:12 +0100, Christoph Hellwig wrote: > On Wed, Mar 27, 2019 at 09:58:16AM +0800, Ming Lei wrote: > > Given it is a IB/SRP specific issue, could you consider the following > > one-line fix? > > It isn't really a IB/SRP specific issue, it is just that the SRP > maintainer actually tried to fix this race that others often ignored. > > And checking for context in queuecommand is a horrible hack that I'd > like to get rid of. Thanks Christoph. I think that's a good summary. Bart.
Hi Martin, As explained in a recent LSF/MM topic proposal, it can happen that the SCSI error handler calls .queuecommand() for quiesced SCSI devices. SCSI LLDs do not expect this. Hence this patch series. Please consider this series for kernel v5.2. Thanks, Bart. Bart Van Assche (2): scsi: Avoid that .queuecommand() gets called for a quiesced SCSI device RDMA/srp: Fix a sleep-in-invalid-context bug drivers/infiniband/ulp/srp/ib_srp.c | 21 ++------------------- drivers/scsi/scsi_error.c | 26 ++++++++++++++++++++++++-- 2 files changed, 26 insertions(+), 21 deletions(-)