From e461338b6cd4074e39a0d5fdd1dc5582fbca1520 Mon Sep 17 00:00:00 2001 From: "Martin K. Petersen" Date: Fri, 5 Dec 2014 08:58:03 -0500 Subject: [PATCH 1/3] sd: tweak discard heuristics to work around QEMU SCSI issue 7985090aa020 changed the discard heuristics to give preference to the WRITE SAME commands that (unlike UNMAP) guarantee deterministic results. Ming Lei discovered that QEMU SCSI's WRITE SAME implementation internally relied on limits that were only communicated for the UNMAP case. And therefore discard commands backed by WRITE SAME would fail. Tweak the heuristics so we still pick UNMAP in the LBPRZ=0 case and only prefer the WRITE SAME variants if the device has the LBPRZ flag set. Reported-by: Ming Lei Tested-by: Ming Lei Signed-off-by: Martin K. Petersen Acked-by: Paolo Bonzini Signed-off-by: Christoph Hellwig --- drivers/scsi/sd.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c index fedab3c21ddf..399516925d80 100644 --- a/drivers/scsi/sd.c +++ b/drivers/scsi/sd.c @@ -2623,8 +2623,9 @@ static void sd_read_block_limits(struct scsi_disk *sdkp) sd_config_discard(sdkp, SD_LBP_WS16); } else { /* LBP VPD page tells us what to use */ - - if (sdkp->lbpws) + if (sdkp->lbpu && sdkp->max_unmap_blocks && !sdkp->lbprz) + sd_config_discard(sdkp, SD_LBP_UNMAP); + else if (sdkp->lbpws) sd_config_discard(sdkp, SD_LBP_WS16); else if (sdkp->lbpws10) sd_config_discard(sdkp, SD_LBP_WS10); From efc7a288382cffc76d6cdb9678f643db37991906 Mon Sep 17 00:00:00 2001 From: "Anil Chintalapati (achintal)" Date: Tue, 23 Dec 2014 19:40:00 +0000 Subject: [PATCH 2/3] fnic: IOMMU Fault occurs when IO and abort IO is out of order When I/O is aborted by mid-layer, fnic FW will complete the I/O before completing the abort task. In some cases abort request is completed before the I/O, which could lead to inconsistent driver and firmware states. In this case firmware reset would clear the inconsistent state. Signed-off-by: Anil Chintalapati Signed-off-by: Sesidhar Baddela Signed-off-by: Hiral Shah Signed-off-by: Christoph Hellwig --- drivers/scsi/fnic/fnic.h | 2 +- drivers/scsi/fnic/fnic_scsi.c | 15 +++++++++++++++ 2 files changed, 16 insertions(+), 1 deletion(-) diff --git a/drivers/scsi/fnic/fnic.h b/drivers/scsi/fnic/fnic.h index 3b73b96619e2..26270c351624 100644 --- a/drivers/scsi/fnic/fnic.h +++ b/drivers/scsi/fnic/fnic.h @@ -39,7 +39,7 @@ #define DRV_NAME "fnic" #define DRV_DESCRIPTION "Cisco FCoE HBA Driver" -#define DRV_VERSION "1.6.0.16" +#define DRV_VERSION "1.6.0.17" #define PFX DRV_NAME ": " #define DFX DRV_NAME "%d: " diff --git a/drivers/scsi/fnic/fnic_scsi.c b/drivers/scsi/fnic/fnic_scsi.c index 2097de42a147..155b286f1a9d 100644 --- a/drivers/scsi/fnic/fnic_scsi.c +++ b/drivers/scsi/fnic/fnic_scsi.c @@ -1892,6 +1892,21 @@ int fnic_abort_cmd(struct scsi_cmnd *sc) goto fnic_abort_cmd_end; } + /* IO out of order */ + + if (!(CMD_FLAGS(sc) & (FNIC_IO_ABORTED | FNIC_IO_DONE))) { + spin_unlock_irqrestore(io_lock, flags); + FNIC_SCSI_DBG(KERN_DEBUG, fnic->lport->host, + "Issuing Host reset due to out of order IO\n"); + + if (fnic_host_reset(sc) == FAILED) { + FNIC_SCSI_DBG(KERN_DEBUG, fnic->lport->host, + "fnic_host_reset failed.\n"); + } + ret = FAILED; + goto fnic_abort_cmd_end; + } + CMD_STATE(sc) = FNIC_IOREQ_ABTS_COMPLETE; /* From 511833acfc06c013d453e288f483c682c60ffbff Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Fri, 21 Nov 2014 10:44:49 -0500 Subject: [PATCH 3/3] SCSI: fix regression in scsi_send_eh_cmnd() MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Commit ac61d1955934 (scsi: set correct completion code in scsi_send_eh_cmnd()) introduced a bug. It changed the stored return value from a queuecommand call, but it didn't take into account that the return value was used again later on. This patch fixes the bug by changing the later usage. There is a big comment in the middle of scsi_send_eh_cmnd() which does a good job of explaining how the routine works. But it mentions a "rtn = FAILURE" value that doesn't exist in the code. This patch adjusts the code to match the comment (I assume the comment is right and the code is wrong). This fixes Bugzilla #88341. Signed-off-by: Alan Stern Reported-by: Андрей Аладьев Tested-by: Андрей Аладьев Fixes: ac61d19559349e205dad7b5122b281419aa74a82 Acked-by: Hannes Reinecke Cc: Signed-off-by: James Bottomley --- drivers/scsi/scsi_error.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/scsi/scsi_error.c b/drivers/scsi/scsi_error.c index e42fff6e8c10..8afb01604d51 100644 --- a/drivers/scsi/scsi_error.c +++ b/drivers/scsi/scsi_error.c @@ -1041,7 +1041,7 @@ retry: } /* signal not to enter either branch of the if () below */ timeleft = 0; - rtn = NEEDS_RETRY; + rtn = FAILED; } else { timeleft = wait_for_completion_timeout(&done, timeout); rtn = SUCCESS; @@ -1081,7 +1081,7 @@ retry: rtn = FAILED; break; } - } else if (!rtn) { + } else if (rtn != FAILED) { scsi_abort_eh_cmnd(scmd); rtn = FAILED; }