summaryrefslogtreecommitdiff
path: root/drivers/scsi/sg.c
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/scsi/sg.c')
-rw-r--r--drivers/scsi/sg.c17
1 files changed, 16 insertions, 1 deletions
diff --git a/drivers/scsi/sg.c b/drivers/scsi/sg.c
index 208bf3c8a16c..5cb2a5e80b2b 100644
--- a/drivers/scsi/sg.c
+++ b/drivers/scsi/sg.c
@@ -1680,7 +1680,22 @@ sg_start_req(Sg_request *srp, unsigned char *cmd)
return -ENOMEM;
}
- rq = blk_get_request(q, rw, GFP_ATOMIC);
+ /*
+ * NOTE
+ *
+ * With scsi-mq enabled, there are a fixed number of preallocated
+ * requests equal in number to shost->can_queue. If all of the
+ * preallocated requests are already in use, then using GFP_ATOMIC with
+ * blk_get_request() will return -EWOULDBLOCK, whereas using GFP_KERNEL
+ * will cause blk_get_request() to sleep until an active command
+ * completes, freeing up a request. Neither option is ideal, but
+ * GFP_KERNEL is the better choice to prevent userspace from getting an
+ * unexpected EWOULDBLOCK.
+ *
+ * With scsi-mq disabled, blk_get_request() with GFP_KERNEL usually
+ * does not sleep except under memory pressure.
+ */
+ rq = blk_get_request(q, rw, GFP_KERNEL);
if (IS_ERR(rq)) {
kfree(long_cmdp);
return PTR_ERR(rq);