From: Dan Dennedy <dan@dennedy.org>
To: Nishanth Aravamudan <nacc@us.ibm.com>
Cc: Stefan Richter <stefanr@s5r6.in-berlin.de>,
Linux1394-Devel <linux1394-devel@lists.sourceforge.net>,
kj <kernel-janitors@lists.osdl.org>,
lkml <linux-kernel@vger.kernel.org>,
Ben Collins <bcollins@debian.org>
Subject: Re: [UPDATE PATCH] ieee1394/sbp2: use ssleep() instead of schedule_timeout()
Date: Thu, 13 Jan 2005 23:52:55 -0500 [thread overview]
Message-ID: <1105678375.7830.81.camel@kino.dennedy.org> (raw)
In-Reply-To: <20050110173945.GB3099@us.ibm.com>
On Mon, 2005-01-10 at 09:39 -0800, Nishanth Aravamudan wrote:
> On Sun, Jan 09, 2005 at 10:01:21AM +0100, Stefan Richter wrote:
> > Nishanth Aravamudan wrote:
> > >Description: Use ssleep() instead of schedule_timeout() to guarantee
> > >the task
> > >delays as expected. The existing code should not really need to run in
> > >TASK_INTERRUPTIBLE, as there is no check for signals (or even an
> > >early return
> > >value whatsoever). ssleep() takes care of these issues.
> >
> > >--- 2.6.10-v/drivers/ieee1394/sbp2.c 2004-12-24 13:34:00.000000000
> > >-0800
> > >+++ 2.6.10/drivers/ieee1394/sbp2.c 2005-01-05 14:23:05.000000000 -0800
> > >@@ -902,8 +902,7 @@ alloc_fail:
> > > * connected to the sbp2 device being removed. That host would
> > > * have a certain amount of time to relogin before the sbp2 device
> > > * allows someone else to login instead. One second makes sense. */
> > >- set_current_state(TASK_INTERRUPTIBLE);
> > >- schedule_timeout(HZ);
> > >+ ssleep(1);
> >
> > Maybe the current code is _deliberately_ accepting interruption by
> > signals but trying to complete sbp2_probe() anyway. However it seems
> > more plausible to me to abort the device probe, for example like this:
> > if (msleep_interruptible(1000)) {
> > sbp2_remove_device(scsi_id);
> > return -EINTR;
> > }
>
> You might be right, but I'd like to get Ben's input on this, as I honeslty am
Don't hold your breath waiting for Ben's input. However, I would like to
get one of the two proposed committed and tested by more users as this
is a sore spot. I am not in a position at this time to fully research
and test to make a call.
> unsure. To be fair, I am trying to audit all usage of schedule_timeout() and the
> semantic interpretation (to me) of using TASK_INTERRUPTIBLE is that you wish to
> sleep a certain amount of time, but also are prepared for an early return on
> either signals or wait-queue events. msleep_interruptible() cleanly removes this
> second issue, but still requires the caller to respond appropriately if there is
> a return value. Hence, I like your change. I think it makes the most sense.
> Since I didn't/don't know how the device works, I was not able to make the
> change myself. Thanks for your input!
Sounds like a sign-off. Any other input before I request Stefan to make
the final decision?
> > Anyway, signal handling does not appear to be critical there.
>
> Just out of curiousity, doesn't that run the risk, though, of
> signal_pending(current) being true for quite a bit of time following the
> timeout?
How much of this is "curiosity" vs a real risk?
next prev parent reply other threads:[~2005-01-14 4:55 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2004-12-25 0:48 [announce] 2.6.10-kj Domen Puncer
2005-01-07 19:33 ` [UPDATE PATCH] atm/ambassador: use msleep() instead of schedule_timeout() Nishanth Aravamudan
2005-01-07 19:40 ` [UPDATE PATCH] ide/ide-cd: use ssleep() " Nishanth Aravamudan
2005-01-07 19:47 ` Jens Axboe
2005-01-15 0:58 ` Bartlomiej Zolnierkiewicz
2005-01-07 21:34 ` [UPDATE PATCH] ieee1394/sbp2: " Nishanth Aravamudan
2005-01-09 9:01 ` Stefan Richter
2005-01-10 17:39 ` Nishanth Aravamudan
2005-01-14 4:52 ` Dan Dennedy [this message]
2005-01-14 11:16 ` Stefan Richter
2005-01-19 6:27 ` [KJ] " Nish Aravamudan
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=1105678375.7830.81.camel@kino.dennedy.org \
--to=dan@dennedy.org \
--cc=bcollins@debian.org \
--cc=kernel-janitors@lists.osdl.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux1394-devel@lists.sourceforge.net \
--cc=nacc@us.ibm.com \
--cc=stefanr@s5r6.in-berlin.de \
/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;
as well as URLs for NNTP newsgroup(s).