Skip to contentSkip to conversation

nvme/pci: Check segement valid for SGL use

PATCH New Keith Busch 4 participants 7 messages Jan 15, 2018 NVMe

lore
Share & tools
lore

Keith Busch

author
#0 lore raw

PATCH 1/1 nvme/pci: Check segement valid for SGL use

The driver needs to verify there is a payload with a command before seeing if it should use SGLs to map it.

Fixes 955b1b5a00ba ("nvme-pci: move use_sgl initialization to nvme_init_iod()")
Reported-by Paul Menzel <pmenzel+linux-nvme at molgen.mpg.de>
Signed-off-by Keith Busch <keith.busch at intel.com>
---
 drivers/nvme/host/pci.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
drivers/nvme/host/pci.c +5 -2
index d53550e612bc..a7e94cc3c70e 100644
@@ -451,10 +451,13 @@ static void **nvme_pci_iod_list(struct request *req)
 static inline bool nvme_pci_use_sgls(struct nvme_dev *dev, struct request *req)
 {
 	struct nvme_iod *iod = blk_mq_rq_to_pdu(req);
+	int nseg = blk_rq_nr_phys_segments(req);
 	unsigned int avg_seg_size;
 
-	avg_seg_size = DIV_ROUND_UP(blk_rq_payload_bytes(req),
-			blk_rq_nr_phys_segments(req));
+	if (nseg == 0)
+		return false;
+
+	avg_seg_size = DIV_ROUND_UP(blk_rq_payload_bytes(req), nseg);
 
 	if (!(dev->ctrl.sgls & ((1 << 0) | (1 << 1))))
 		return false;
Signature · 1 line
2.13.6
#1 lore raw

PATCH 1/1 nvme/pci: Check segement valid for SGL use

Dear Keith,

Quoted 22 lines · Keith Busch

The driver needs to verify there is a payload with a command before seeing if it should use SGLs to map it.

Fixes 955b1b5a00ba ("nvme-pci: move use_sgl initialization to nvme_init_iod()")
Reported-by Paul Menzel <pmenzel+linux-nvme at molgen.mpg.de>
Signed-off-by Keith Busch <keith.busch at intel.com>
---
  drivers/nvme/host/pci.c | 7 +++++--
  1 file changed, 5 insertions(+), 2 deletions(-)
drivers/nvme/host/pci.c +1 -0
index d53550e612bc..a7e94cc3c70e 100644
@@ -451,10 +451,13 @@ static void **nvme_pci_iod_list(struct request *req)
  static inline bool nvme_pci_use_sgls(struct nvme_dev *dev, struct request *req)
  {
  	struct nvme_iod *iod = blk_mq_rq_to_pdu(req);
+	int nseg = blk_rq_nr_phys_segments(req);
  	unsigned int avg_seg_size;

drivers/nvme/host/pci.c static void **nvme_pci_iod_list(struct request *req) · line 463

Quoted 10 lines
drivers/nvme/host/pci.c +4 -2
 
-	avg_seg_size = DIV_ROUND_UP(blk_rq_payload_bytes(req),
-			blk_rq_nr_phys_segments(req));
+	if (nseg == 0)
+		return false;
+
+	avg_seg_size = DIV_ROUND_UP(blk_rq_payload_bytes(req), nseg);
 
  	if (!(dev->ctrl.sgls & ((1 << 0) | (1 << 1))))
  		return false;

Thank you for the quick reaction.

Reviewed-by Paul Menzel <pmenzel+linux-nvme at molgen.mpg.de>

Kind regards,

Paul

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 5174 bytes
Desc: S/MIME Cryptographic Signature
URL: <http://lists.infradead.org/pipermail/linux-nvme/attachments/20180115/51c8b8d8/attachment.p7s>

maintainer
#3 lore raw

PATCH 1/1 nvme/pci: Check segement valid for SGL use

In reply to Keith Busch

Actually this needs to go into nvme-4.15. The offending commit is only in 4.15-rc, and not in nvme-4.16 or Jens' for-4.16 tree.

maintainer
#4 lore raw

PATCH 1/1 nvme/pci: Check segement valid for SGL use

Quoted 2 lines · Christoph Hellwig

Actually this needs to go into nvme-4.15. The offending commit is only in 4.15-rc, and not in nvme-4.16 or Jens' for-4.16 tree.

I can queue it up for 4.15, there's already been one report of this...

Signature · 1 line
Jens Axboe
maintainer
#5 lore raw

PATCH 1/1 nvme/pci: Check segement valid for SGL use

Quoted 6 lines · Jens Axboe
Quoted 2 lines · Christoph Hellwig

Actually this needs to go into nvme-4.15. The offending commit is only in 4.15-rc, and not in nvme-4.16 or Jens' for-4.16 tree.

I can queue it up for 4.15, there's already been one report of this...

I think we'll need at least another related fix as well. I'll send you a pull request by tomorrow or so.

maintainer
#6 lore raw

PATCH 1/1 nvme/pci: Check segement valid for SGL use

Quoted 10 lines · Christoph Hellwig
Quoted 6 lines · Jens Axboe
Quoted 2 lines · Christoph Hellwig

Actually this needs to go into nvme-4.15. The offending commit is only in 4.15-rc, and not in nvme-4.16 or Jens' for-4.16 tree.

I can queue it up for 4.15, there's already been one report of this...

I think we'll need at least another related fix as well. I'll send you a pull request by tomorrow or so.

Sounds good, thanks.

Signature · 1 line
Jens Axboe