Remove fixed 16 sg_request object array and replace with an active
rq_list plus a request free list. Add finer grained spin lock to
sg_request and do a major rework on locking. sg_request objects now
are only de-allocated when the owning file descriptor is closed.
This simplifies locking issues.
Add a SG_SET_GET_EXTENDED ioctl control for whether commands
will be queued_at_head or queued_at_tail by the block layer
(together with the scsi mid-level). It has file scope.
Also add a read_value integer the can be used by write a
value from the SG_SEIRV_* group then the corresponding value
Complete the locking and structure changes of ioctl and debug
('cat /proc/scsi/sg/debug') handling.
Signed-off-by: Douglas Gilbert
---
This was the code that was "#if 0'-ed out 2 patches ago. It
also shuts checkpatch.pl up as it doesn't like that technique
but offers no viable substitute.
On 10/2/18 11:52 PM, Bart Van Assche wrote:
Hello Martin,
During the 2018 edition of LSF/MM there was a session about increasing SCSI
disk probing concurrency. This patch series implements what has been proposed
during that session, namely:
- Make sure that the driver core is aware of
https://bugzilla.kernel.org/show_bug.cgi?id=201199
--- Comment #8 from Seb Lu (se...@seblu.net) ---
Hello,
Any idea? Any test I can do to find what's wrong?
--
You are receiving this mail because:
You are the assignee for the bug.
On Fri, 2018-10-26 at 15:58 +0200, Hannes Reinecke wrote:
> On 10/2/18 11:52 PM, Bart Van Assche wrote:
> > Hello Martin,
> >
> > During the 2018 edition of LSF/MM there was a session about increasing SCSI
> > disk probing concurrency. This patch series implements what has been
> > proposed
> >
Split scsi/sg.h into a smaller scsi/sg.h which includes a new header:
uapi/scsi/sg.h . Overall expand the twin header files with new
functionality in this patchset and functionality to be added in the
next patchset to implement SG_IOSUBMIT and friends. Adjust format to
modern kernel conventions.
Introduce the new sg_rq_state enumerations for tracking the
lifetime of a sg_request. SG_RQ_DATA_THRESHOLD is a
default value that if the data length of a request exceeds
then, after that request is completed, the data buffer will
be freed up as the sg_request object is placed on the free
list.
Rework ioctl handling, report clearly to the log which ioctl has
been invoked. Add a new "IOWR" ioctl: SG_SET_GET_EXTENDED which
permits several integer and boolean values to be "_SET_" (i.e.
passed into the driver, potentially changing its actions) and/or
read from the driver (the "_GET_" part)
Introduce the SG_LOG macro to replace long-winded
'SCSI_LOG_TIMEOUT(3, sg_printk ...' debug messages. Use __func__
wherever appropriate to make the debug messages more portable.
Signed-off-by: Douglas Gilbert
---
Reviewers want SCSI_LOG_* style logging replaced. But with
what? One suggestion is
Remove typedefs and use better type names like bool and u8 where
appropriate. Rename some variables and functions for clarity.
Adjust formatting (e.g. function definitions) to be more
consistent across the driver.
Signed-off-by: Douglas Gilbert
---
The intention is to move to sg_version_num
The intention is to add two new ioctls as proposed by Linus Torvalds:
SG_IOSUBMIT and SG_IORECEIVE to replace the write()/read() async
interface.
But first, clean up the driver and remove the SG_MAX_QUEUE limit of
no more than 16 queued commands on a file descriptor at a time. A
free list has
Hi James,
> On Oct 25, 2018, at 11:23 AM, James Smart wrote:
>
> External Email
>
> On 9/28/2018 3:46 PM, Himanshu Madhani wrote:
>> + .target_features= NVMET_FCTGTFEAT_READDATA_RSP |
>> + NVMET_FCTGTFEAT_CMD_IN_ISR |
>> +
On 10/26/18 10:06 AM, Benjamin Block wrote:
> On Thu, Oct 25, 2018 at 05:12:59PM -0600, Jens Axboe wrote:
>> On 10/23/18 12:07 PM, Jens Axboe wrote:
>>> On 10/23/18 11:40 AM, Benjamin Block wrote:
So I tried 4.19.0 with only the two patches from you:
On Thu, Oct 25, 2018 at 05:12:59PM -0600, Jens Axboe wrote:
> On 10/23/18 12:07 PM, Jens Axboe wrote:
> > On 10/23/18 11:40 AM, Benjamin Block wrote:
> >>
> >> So I tried 4.19.0 with only the two patches from you:
> >>
Add a SG_SET_GET_EXTENDED ioctl control for whether commands
will be queued_at_head or queued_at_tail by the block layer
(together with the scsi mid-level). It has file scope.
Also add a read_value integer the can be used by write a
value from the SG_SEIRV_* group then the corresponding value
/0day-ci/linux/commits/linux-scsi-owner-vger-kernel-org/sg-types-and-naming-cleanup/20181026-220008
base: https://git.kernel.org/pub/scm/linux/kernel/git/mkp/scsi.git for-next
config: x86_64-randconfig-r0-10262157 (attached as .config)
compiler: gcc-6 (Debian 6.4.0-9) 6.4.0 20171026
reproduce
On Fri, Oct 26, 2018 at 10:08:30AM -0600, Jens Axboe wrote:
> On 10/26/18 10:06 AM, Benjamin Block wrote:
> > On Thu, Oct 25, 2018 at 05:12:59PM -0600, Jens Axboe wrote:
> >> On 10/23/18 12:07 PM, Jens Axboe wrote:
> >>> On 10/23/18 11:40 AM, Benjamin Block wrote:
>
> So I tried 4.19.0
On 2018-10-26 12:48 p.m., Douglas Gilbert wrote:
Add a SG_SET_GET_EXTENDED ioctl control for whether commands
will be queued_at_head or queued_at_tail by the block layer
(together with the scsi mid-level). It has file scope.
Also add a read_value integer the can be used by write a
value from
On Thu, Oct 18, 2018 at 09:43:25PM -0700, Bart Van Assche wrote:
> Hi Christoph,
>
> Have you considered to use put_unaligned_le64() instead of storing the
> lower and upper 32 bits separately?
I really don't want to touch this old driver all that much, just
get rid of the buggy existing
20 matches
Mail list logo