On 07/16/2015 01:15 PM, Marciniszyn, Mike wrote:
Bart Van Assche help us and resulting patch you can see below:
diff --git a/drivers/infiniband/ulp/srp/ib_srp.c
b/drivers/infiniband/ulp/srp/ib_srp.c
index c5af67a..470eb54 100644
--- a/drivers/infiniband/ulp/srp/ib_srp.c
+++ b/drivers/infiniband/ulp/srp/ib_srp.c
@@ -59,6 +59,7 @@
#include "../../../../include/scsi/scsi_transport_srp.h"
#include "ib_srp.h"
+#include <linux/delay.h>
#define DRV_NAME "ib_srp"
#define PFX DRV_NAME ": "
@@ -621,11 +622,7 @@ static void srp_destroy_qp(struct srp_rdma_ch *ch)
if (ret)
goto out;
- init_completion(&ch->done);
- ret = ib_post_recv(ch->qp, &wr, &bad_wr);
- WARN_ONCE(ret, "ib_post_recv() returned %d\n", ret);
- if (ret == 0)
- wait_for_completion(&ch->done);
+ msleep(100);
out:
ib_destroy_qp(ch->qp);
Post receive can return a synchronous failure. The question would be which one and why?
Bart, any ideas?
Hello Mike,
ib_post_recv() can indeed return a synchronous failure. But I think a
synchronous failure should be handled properly by the SRP initiator
driver: if ret < 0 then the wait_for_completion() call is skipped. What
Alexey reported is that the wait_for_completion() call did not finish
which means that ib_post_recv() returned 0. Or did I perhaps
misinterpret something ?
Thanks,
Bart.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html