Hi,
Can you elaborate how this can even happen?
Isn't the interrupt aggregation capability should attend for those cases?
Thanks,
Avri
> -Original Message-
> From: linux-scsi-ow...@vger.kernel.org [mailto:linux-scsi-
> ow...@vger.kernel.org] On Behalf Of Asutosh Das
> Sent: Tuesday,
Hi Martin,
> On Jan 30, 2018, at 7:04 PM, Martin K. Petersen
> wrote:
>
>
> Bart,
>
>> The patches in this series are what I came up with after having analyzed
>> the source code of the qla[24]xxx drivers with several source code analysis
>> tools
On Tue, Jan 30, 2018 at 08:22:27PM -0700, Jens Axboe wrote:
> On 1/30/18 8:21 PM, Bart Van Assche wrote:
> > On Tue, 2018-01-30 at 20:17 -0700, Jens Axboe wrote:
> >> BLK_STS_RESOURCE should always be safe to return, and it should work
> >> the same as STS_DEV_RESOURCE, except it may cause an
On 1/30/18 8:27 PM, Bart Van Assche wrote:
> On Tue, 2018-01-30 at 20:22 -0700, Jens Axboe wrote:
>> On 1/30/18 8:21 PM, Bart Van Assche wrote:
>>> On Tue, 2018-01-30 at 20:17 -0700, Jens Axboe wrote:
BLK_STS_RESOURCE should always be safe to return, and it should work
the same as
On Tue, 2018-01-30 at 20:22 -0700, Jens Axboe wrote:
> On 1/30/18 8:21 PM, Bart Van Assche wrote:
> > On Tue, 2018-01-30 at 20:17 -0700, Jens Axboe wrote:
> > > BLK_STS_RESOURCE should always be safe to return, and it should work
> > > the same as STS_DEV_RESOURCE, except it may cause an extra
Weiping,
> add user friendly command strings sd_wr_cache to enable/disable
> write cache. user can enable both write and read cache by one of
> the following commands:
I remain unconvinced that introducing redundant option strings is a
benefit.
These shorthand forms may seem obvious to you,
On 1/30/18 8:21 PM, Bart Van Assche wrote:
> On Tue, 2018-01-30 at 20:17 -0700, Jens Axboe wrote:
>> BLK_STS_RESOURCE should always be safe to return, and it should work
>> the same as STS_DEV_RESOURCE, except it may cause an extra queue
>> run.
>>
>> Well written drivers should use
On Tue, 2018-01-30 at 20:17 -0700, Jens Axboe wrote:
> BLK_STS_RESOURCE should always be safe to return, and it should work
> the same as STS_DEV_RESOURCE, except it may cause an extra queue
> run.
>
> Well written drivers should use STS_DEV_RESOURCE where it makes
> sense.
Hello Jens,
I would
On 1/30/18 8:04 PM, Mike Snitzer wrote:
> From: Ming Lei
>
> This status is returned from driver to block layer if device related
> resource is unavailable, but driver can guarantee that IO dispatch
> will be triggered in future when the resource is available.
>
> Convert
On 1/30/18 10:52 AM, Bart Van Assche wrote:
> On 01/30/18 06:24, Mike Snitzer wrote:
>> + *
>> + * If driver returns BLK_STS_RESOURCE and SCHED_RESTART
>> + * bit is set, run queue after a delay to avoid IO stalls
>> + * that could otherwise occur if
On Tue, Jan 30 2018 at 9:44P -0500,
Jens Axboe wrote:
> On 1/30/18 7:24 AM, Mike Snitzer wrote:
> > From: Ming Lei
> >
> > This status is returned from driver to block layer if device related
> > resource is unavailable, but driver can guarantee that IO
> Increase cmd_per_lun to allow more I/Os in progress per device,
> particularly for NVMe's. The Hyper-V host side can handle the
> higher count with no issues.
Long/KY/Cathy/Stephen: Please ack/review Michael's patches.
--
Martin K. Petersen Oracle Linux Engineering
From: Ming Lei
This status is returned from driver to block layer if device related
resource is unavailable, but driver can guarantee that IO dispatch
will be triggered in future when the resource is available.
Convert some drivers to return BLK_STS_DEV_RESOURCE. Also, if
Bart,
> The patches in this series are what I came up with after having analyzed
> the source code of the qla[24]xxx drivers with several source code analysis
> tools (scripts/kernel-doc, gcc, sparse and smatch). None of the patches in
> this series have been tested. Yet I'm asking you to
Bart,
> Since commit 3a025e1d1c2e ("Add optional check for bad kernel-doc
> comments") building with W=1 causes warnings to appear for issues
> in kernel-doc headers. This patch avoids that the following warnings
> are reported when building with W=1:
Applied to 4.16/scsi-fixes, thanks!
--
Corentin,
> Remove line using inexistant files which were removed in
> commit 642978beb483 ("[SCSI] remove m68k NCR53C9x based drivers")
Applied to 4.16/scsi-fixes, thank you.
--
Martin K. Petersen Oracle Linux Engineering
Corentin,
> aiclib.c is unused (and contain no code) since commit 1ff927306e08
> ("[SCSI] aic7xxx: remove aiclib.c")
> 13 years later, finish the cleaning by remove it from tree.
Applied to 4.16/scsi-fixes, thanks!
--
Martin K. Petersen Oracle Linux Engineering
Dan,
> My static checker complains about an out of bounds read:
>
> drivers/message/fusion/mptctl.c:2786 mptctl_hp_targetinfo()
> error: buffer overflow 'hd->sel_timeout' 255 <= u32max.
>
> It's true that we probably should have a bounds check here.
Applied to 4.16/scsi-fixes, thanks!
On 1/30/18 7:24 AM, Mike Snitzer wrote:
> From: Ming Lei
>
> This status is returned from driver to block layer if device related
> resource is unavailable, but driver can guarantee that IO dispatch
> will be triggered in future when the resource is available.
>
> Convert
Bart,
> A left shift must shift less than the bit width of the left argument.
> Avoid triggering undefined behavior if ha->mbx_count == 32.
Applied to 4.16/scsi-fixes. Thank you!
--
Martin K. Petersen Oracle Linux Engineering
Dan,
> We wanted to exit the loop with "div" set to zero, but instead, if we
> don't hit the break then "div" is -1 when we finish the loop. It
> leads to an array underflow a few lines later.
Applied to 4.16/scsi-fixes, thanks!
--
Martin K. Petersen Oracle Linux Engineering
Chad,
> When a request times out we set the io_req flag BNX2FC_FLAG_IO_COMPL so
> that if a subsequent completion comes in on that task ID we will ignore
> it. The issue is that in the check for this flag there is a missing
> return so we will continue to process a request which may have
Colin,
> The pointer ln is assigned a value that is never read, it is re-assigned
> a new value in the list_for_each loop hence the initialization is
> redundant and can be removed.
Applied to 4.16/scsi-fixes. Thank you.
--
Martin K. Petersen Oracle Linux Engineering
Asutosh,
> /* WRITE_SAME command is not supported*/
> sdev->no_write_same = 1;
>
> + /* WRITE_SAME command is not supported*/
> + sdev->no_write_same = 1;
> +
Uhm...
I applied this change to 4.16/scsi-fixes by hand.
--
Martin K. Petersen Oracle Linux Engineering
Tyrel,
> The fcp_rsp_info structure as defined in the FC spec has an initial 3 bytes
> reserved field. The ibmvfc driver mistakenly defined this field as 4 bytes
> resulting in the rsp_code field being defined in what should be the start of
> the second reserved field and thus always being
Himanshu,
> This patch fixes memory corrpution while performing
> HBA Reset test.
Applied to 4.16/scsi-fixes. Thank you!
--
Martin K. Petersen Oracle Linux Engineering
On Tue, Jan 30, 2018 at 09:52:31AM -0800, Bart Van Assche wrote:
> On 01/30/18 06:24, Mike Snitzer wrote:
> > +*
> > +* If driver returns BLK_STS_RESOURCE and SCHED_RESTART
> > +* bit is set, run queue after a delay to avoid IO stalls
> > +* that
Tomas,
> cpu_msix_table is allocated to store online cpus, but
> pci_irq_get_affinity may return cpu_possible_mask which is then used
> to access cpu_msix_table. That causes bad user experience. Fix
> limits access to only online cpus, I've also added an additonal test
> to protect from an
Make the attribute writeable.
Remove the ramp up to logic as its unnecessary, simply set depth.
Add debug message if depth changed, possibly reducing limit, yet
our outstanding count has yet to catch up with it.
Signed-off-by: Dick Kennedy
Signed-off-by: James Smart
The driver ignored checks on whether the link should be
kept administratively down after a link bounce. Correct the
checks.
Signed-off-by: Dick Kennedy
Signed-off-by: James Smart
Reviewed-by: Hannes Reinecke
---
Updated Copyright in files updated 11.4.0.7
Signed-off-by: Dick Kennedy
Signed-off-by: James Smart
---
v3:
Revised LPFC_COPYRIGHT string for 2018 as well.
---
drivers/scsi/lpfc/lpfc.h | 2 +-
drivers/scsi/lpfc/lpfc_attr.c |
During link bounce testing in a point-to-point topology, the
host may enter a soft lockup on the lpfc_worker thread:
Call Trace:
lpfc_work_done+0x1f3/0x1390 [lpfc]
lpfc_do_work+0x16f/0x180 [lpfc]
kthread+0xc7/0xe0
ret_from_fork+0x3f/0x70
The driver was simultaneously
The lpfc driver does not discover a target when the topology
changes from switched-fabric to direct-connect. The target
rejects the PRLI from the initiator in direct-connect as the
driver is using the old S_ID from the switched topology.
The driver was inappropriately clearing the VP bit to
In a test that is doing large numbers of cable swaps on the target,
the nvme controllers wouldn't reconnect.
During the cable swaps, the targets n_port_id would change. This
information was passed to the nvme-fc transport, in the new remoteport
registration. However, the nvme-fc transport didn't
When using the special option to suppress the response iu, ensure
the adapter fully supports the feature by checking feature flags
from the adapter and validating the support when formatting the
WQE.
Signed-off-by: Dick Kennedy
Signed-off-by: James Smart
Update the driver version to 11.4.0.7
Signed-off-by: Dick Kennedy
Signed-off-by: James Smart
Reviewed-by: Hannes Reinecke
---
drivers/scsi/lpfc/lpfc_version.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git
When nvme target deferred receive logic waits for exchange
resources, the corresponding receive buffer is not replenished
with the hardware. This can result in a lack of asynchronous
receive buffer resources in the hardware, resulting in a
"2885 Port Status Event: ... error 1=0x52004a01 ..."
Currently, write underruns (mismatch of amount transferred vs scsi
status and its residual) detected by the adapter are not being
flagged as an error. Its expected the target controls the data
transfer and would appropriately set the RSP values. Only read
underruns are treated as errors.
Revise
During SCSI error handling escalation to host reset, the SCSI io
routines were moved off the txcmplq, but the individual io's
ON_CMPLQ flag wasn't cleared. Thus, a background thread saw the
io and attempted to access it as if on the txcmplq.
Clear the flag upon removal.
Signed-off-by: Dick
A stress test repeatedly resetting the adapter while performing
io would eventually report I/O failures and missing nvme namespaces.
The driver was setting the nvmefc_fcp_req->private pointer to NULL
during the IO completion routine before upcalling done().
If the transport was also running an
I/O conditions on the nvme target may have the driver submitting
to a full hardware wq. The hardware wq is a shared resource among
all nvme controllers. When the driver hit a full wq, it failed the
io posting back to the nvme-fc transport, which then escalated it
into errors.
Correct by
Revise the NVME PRLI to indicate CONF support.
Signed-off-by: Dick Kennedy
Signed-off-by: James Smart
Reviewed-by: Hannes Reinecke
---
drivers/scsi/lpfc/lpfc_els.c | 3 ++-
drivers/scsi/lpfc/lpfc_hw4.h | 6 +++---
The driver was inappropriately pulling in the nvme host's
nvme.h header. What it really needed was the standard
header.
Signed-off-by: Dick Kennedy
Signed-off-by: James Smart
Reviewed-by: Hannes Reinecke
---
Existing code was using the wrong field for the completion status
when comparing whether to increment abort statistics
Signed-off-by: Dick Kennedy
Signed-off-by: James Smart
Reviewed-by: Hannes Reinecke
---
Increased CQ and WQ sizes for SCSI FCP, matching those used
for NVMe development.
Signed-off-by: Dick Kennedy
Signed-off-by: James Smart
---
v3:
lpfc_init.c:
Corrected indentation of comments after patching
Removed unnecessary
The driver controls when the hardware sends completions that
communicate consumption of elements from the WQ. This is done by
setting a WQEC bit on a WQE.
The current driver sets it on every Nth WQE posting. However, the
driver isn't clearing the bit if the WQE is reused. Thus, if the
queue depth
Ensure nvme localports/targetports are torn down before
dismantling the adapter sli interface on driver detachment.
This aids leaving interfaces live while nvme may be making
callbacks to abort it.
Signed-off-by: Dick Kennedy
Signed-off-by: James Smart
This patch set provides a number of fixes for the driver.
The patches were cut against the Martin's 4.16/scsi-queue tree.
There are no outside dependencies and are expected to be pulled
via Martins tree.
v2:
respin patch 14 "lpfc: Validate adapter support for SRIU option"
for snippet that
On Tue, Jan 30 2018 at 2:42pm -0500,
Bart Van Assche wrote:
> On Tue, 2018-01-30 at 14:33 -0500, Mike Snitzer wrote:
> > On Tue, Jan 30 2018 at 12:52pm -0500,
> > Bart Van Assche wrote:
> >
> > > - This patch does not fix any bugs nor makes
On Tue, 2018-01-30 at 14:33 -0500, Mike Snitzer wrote:
> On Tue, Jan 30 2018 at 12:52pm -0500,
> Bart Van Assche wrote:
>
> > - This patch does not fix any bugs nor makes block drivers easier to
> > read or to implement. So why is this patch considered useful?
>
> It
On Tue, Jan 30 2018 at 12:52pm -0500,
Bart Van Assche wrote:
> On 01/30/18 06:24, Mike Snitzer wrote:
> >+ *
> >+ * If driver returns BLK_STS_RESOURCE and SCHED_RESTART
> >+ * bit is set, run queue after a delay to avoid IO stalls
> >+
On Tue, 2018-01-30 at 16:07 +0100, Hannes Reinecke wrote:
> On 01/29/2018 10:08 PM, Mike Snitzer wrote:
> > We currently don't restack the queue_limits if the lowest, or
> > intermediate, layer of an IO stack changes.
> >
> > This is particularly unfortunate in the case of FLUSH/FUA which may
> >
On Tue, 2018-01-30 at 09:52 -0800, Bart Van Assche wrote:
> On 01/30/18 06:24, Mike Snitzer wrote:
> > + *
> > + * If driver returns BLK_STS_RESOURCE and
> > SCHED_RESTART
> > + * bit is set, run queue after a delay to avoid IO
> > stalls
> > + * that
On 01/30/18 06:24, Mike Snitzer wrote:
+*
+* If driver returns BLK_STS_RESOURCE and SCHED_RESTART
+* bit is set, run queue after a delay to avoid IO stalls
+* that could otherwise occur if the queue is idle.
*/
-
On 01/29/2018 10:08 PM, Mike Snitzer wrote:
> We currently don't restack the queue_limits if the lowest, or
> intermediate, layer of an IO stack changes.
>
> This is particularly unfortunate in the case of FLUSH/FUA which may
> change if/when a HW controller's BBU fails; whereby requiring the
From: Ming Lei
This status is returned from driver to block layer if device related
resource is unavailable, but driver can guarantee that IO dispatch
will be triggered in future when the resource is available.
Convert some drivers to return BLK_STS_DEV_RESOURCE. Also, if
On Tue, Jan 30, 2018 at 11:08:28AM +0100, Johannes Thumshirn wrote:
> [+Cc Mel]
> Jens Axboe writes:
> > On 1/29/18 1:56 PM, James Bottomley wrote:
> >> On Mon, 2018-01-29 at 23:46 +0800, Ming Lei wrote:
> >> [...]
> >>> 2. When to enable SCSI_MQ at default again?
> >>
> >> I'm
Hi Himanshu,
Today I tried several times and have some news.
Before I insmod the qla2xxx.ko , if I shutdown then start the FC switch port
connected to the HBA card, the qla2xxx.ko works well.
It seems that the issue has relation to the FC switch port. Maybe some old
status causes the
On 30/01/2018 01:24, Ming Lei wrote:
On Mon, Jan 29, 2018 at 12:56:30PM -0800, James Bottomley wrote:
On Mon, 2018-01-29 at 23:46 +0800, Ming Lei wrote:
[...]
2. When to enable SCSI_MQ at default again?
I'm not sure there's much to discuss ... I think the basic answer is as
soon as Christoph
Introduce a new blist flag that indicates the device may return certain
sense code/ASC/ASCQ combinations that indicate different treatment than
normal. In particular, some devices need unconditional retry (aka
ADD_TO_MLQUEUE) under certain conditions; otherwise path failures may be
falsely
[+Cc Mel]
Jens Axboe writes:
> On 1/29/18 1:56 PM, James Bottomley wrote:
>> On Mon, 2018-01-29 at 23:46 +0800, Ming Lei wrote:
>> [...]
>>> 2. When to enable SCSI_MQ at default again?
>>
>> I'm not sure there's much to discuss ... I think the basic answer is as
>> soon as
Ming Lei - 30.01.18, 02:24:
> > > SCSI_MQ is enabled on V3.17 firstly, but disabled at default. In
> > > V4.13-rc1, it is enabled at default, but later the patch is reverted
> > > in V4.13-rc7, and becomes disabled at default too.
> > >
> > > Now both the original reported PM issue(actually SCSI
62 matches
Mail list logo