* [MPT3SAS bug] wrong queue setting for HDD. drive
@ 2021-01-27 22:38 Ming Lei
2021-01-28 7:15 ` Sreekanth Reddy
0 siblings, 1 reply; 2+ messages in thread
From: Ming Lei @ 2021-01-27 22:38 UTC (permalink / raw)
To: Suganath Prabu Subramani, Sathya Prakash, Sreekanth Reddy
Cc: Martin K. Petersen, MPT-FusionLinux.pdl, linux-scsi
Hello Guys,
The commit[1] is supposed for NVMe device only, but we found the change is
actually done on the following mpt3sas HDD. drive:
#sginfo /dev/sdc
INQUIRY response (cmd: 0x12)
----------------------------
Device Type 0
Vendor: SEAGATE
Product: ST16000NM002G
Revision level: E003)
So NOMERGES is enabled, and virt boundary limit is applied on this HDD.
device, and performance drop is observed, can you take a look at the
issue?
[1] commit d1b01d14b7baa8a4bb4c11305c8cca73456b2f7c
Author: Suganath Prabu Subramani <suganath-prabu.subramani@broadcom.com>
Date: Tue Oct 31 18:02:33 2017 +0530
scsi: mpt3sas: Set NVMe device queue depth as 128
Sets nvme device queue depth, name and displays device capabilities
Thanks,
Ming
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [MPT3SAS bug] wrong queue setting for HDD. drive
2021-01-27 22:38 [MPT3SAS bug] wrong queue setting for HDD. drive Ming Lei
@ 2021-01-28 7:15 ` Sreekanth Reddy
0 siblings, 0 replies; 2+ messages in thread
From: Sreekanth Reddy @ 2021-01-28 7:15 UTC (permalink / raw)
To: Ming Lei
Cc: Suganath Prabu Subramani, Sathya Prakash, Martin K. Petersen,
PDL-MPT-FUSIONLINUX, linux-scsi
[-- Attachment #1: Type: text/plain, Size: 2210 bytes --]
Ming,
Below referred commit changes apply only for NVMe drives.
Driver sets nomerges & virtual boundary under below if condition. And
the driver sets MPT_TARGET_FLAGS_PCIE_DEVICE bit in
sas_target_priv_data->flags only for NVMe drives.
if (sas_target_priv_data->flags & MPT_TARGET_FLAGS_PCIE_DEVICE) {
...
blk_queue_flag_set(QUEUE_FLAG_NOMERGES,
sdev->request_queue);
blk_queue_virt_boundary(sdev->request_queue,
ioc->page_size - 1);
}
on my local setup, I have both SAS HDD and NVMe drives as shown below,
[11:0:20:0] disk SEAGATE ST600MP0005 VS09 /dev/sdz
[11:2:0:0] disk NVMe INTEL SSDPE2MW40 0174 /dev/sdab
and default nomerge setting for these drives is as show below,
/sys/devices/pci0000:80/0000:80:03.0/0000:85:00.0/host11/port-11:0/expander-11:0/port-11:0:19/end_device-11:0:19/target11:0:20/11:0:20:0/block/sdz/queue/nomerges:0
/sys/devices/pci0000:80/0000:80:03.0/0000:85:00.0/host11/target11:2:0/11:2:0:0/block/sdab/queue/nomerges:2
so, for SAS HDD drive default nomerge setting is zero and for NVMe
drive the default nomerge setting is two.
Thanks,
Sreekanth
On Thu, Jan 28, 2021 at 4:08 AM Ming Lei <ming.lei@redhat.com> wrote:
>
> Hello Guys,
>
> The commit[1] is supposed for NVMe device only, but we found the change is
> actually done on the following mpt3sas HDD. drive:
>
> #sginfo /dev/sdc
> INQUIRY response (cmd: 0x12)
> ----------------------------
> Device Type 0
> Vendor: SEAGATE
> Product: ST16000NM002G
> Revision level: E003)
>
> So NOMERGES is enabled, and virt boundary limit is applied on this HDD.
> device, and performance drop is observed, can you take a look at the
> issue?
>
>
> [1] commit d1b01d14b7baa8a4bb4c11305c8cca73456b2f7c
> Author: Suganath Prabu Subramani <suganath-prabu.subramani@broadcom.com>
> Date: Tue Oct 31 18:02:33 2017 +0530
>
> scsi: mpt3sas: Set NVMe device queue depth as 128
>
> Sets nvme device queue depth, name and displays device capabilities
>
>
>
> Thanks,
> Ming
>
[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 4175 bytes --]
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2021-01-28 7:19 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-01-27 22:38 [MPT3SAS bug] wrong queue setting for HDD. drive Ming Lei
2021-01-28 7:15 ` Sreekanth Reddy
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.