public inbox for linux-scsi@vger.kernel.org
 help / color / mirror / Atom feed
From: Douglas Gilbert <dougg@torque.net>
To: Olaf Hering <olh@suse.de>
Cc: linux-scsi@vger.kernel.org, patmans@ibm.com
Subject: Re: sysfs support for sg
Date: Thu, 08 Jan 2004 11:50:03 +1000	[thread overview]
Message-ID: <3FFCB74B.9090102@torque.net> (raw)
In-Reply-To: <20040107224028.GA24290@suse.de>

[-- Attachment #1: Type: text/plain, Size: 1856 bytes --]

Olaf Hering wrote:
>  On Wed, Jan 07, Olaf Hering wrote:
> 
> 
>>Hello,
>>
>>are there any plans to add sysfs support to the sg driver?
>>I see no 'dev' nodes for it.
>>Can stuff like 'sg_start /dev/sg2 0' be done in another way?
> 
> 
> I missed the /sys/block/sda/device/generic symlink. But still, no dev
> node. Is that supposed to work that way?

Olaf,
lk 2.6.0-rc1 has added sysfs support for sg with
/sys/cdev/major/sg<n> . In that directory
is a symlink to the corresponding SCSI device. When that symlink
is followed the target directory has a symlink called 'generic'
back to /sys/cdev/major/sg<n>. The st driver has similar cdev
bidirectional mapping (and hopefully osst will soon follow).

An example is attached. [BTW 'getcwd' returns the "true" path
as described in "man 3 getcwd".]

The "cdev" changes in the sg and st drivers allow more than
256 devices to be handled. The sg driver limits the number
of devices to 8192. I believe sd is still limited to
128 devices and I not sure about the device limits of sr.

Currently the sg driver lacks driver parameter support within sysfs
(e.g. 'allow_dio' as found in the /proc/scsi/sg directory).
Interestingly the st and osst drivers have /sys/bus/scsi/drivers/st
and osst directories (sd and st drivers also get directories if
their drivers are loaded). It is useful to be able to access
ULDs _prior_ to any (scsi) devices being attached which is not
possible under lk 2.4 .

I have been looking at the lsscsi program again which has been
dormant for 6 months. A symlink that I miss is from a scsi host
(enumerated in the /sys/class/scsi_host directory) to the
driver (and its name) that controls it.


BTW Is there a definitive libsysfs site anywhere? I noticed that
scsi_id provided in the "extras" directory of udev-009 doesn't
build with the libsysfs provided by udev-009.

Doug Gilbert




[-- Attachment #2: sg_sysfs.txt --]
[-- Type: text/plain, Size: 342 bytes --]

# cd /sys/cdev/major/sg0
# ls
device
# tree
.
`-- device -> ../../../devices/pci0000:00/0000:00:0e.0/host0/0:0:1:0

1 directory, 0 files
# cd device/
# ls
block   detach_state    generic  online  queue_depth  rev         type
delete  device_blocked  model    power   rescan       scsi_level  vendor
# cd generic/
# getcwd
/sys/cdev/major/sg0

  reply	other threads:[~2004-01-08  1:52 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2004-01-07 22:33 sysfs support for sg Olaf Hering
2004-01-07 22:40 ` Olaf Hering
2004-01-08  1:50   ` Douglas Gilbert [this message]
2004-01-08  2:11     ` Douglas Gilbert
2004-01-08  5:41     ` Daniel Stekloff

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=3FFCB74B.9090102@torque.net \
    --to=dougg@torque.net \
    --cc=linux-scsi@vger.kernel.org \
    --cc=olh@suse.de \
    --cc=patmans@ibm.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox