mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: devik <devik@cdi.cz>
To: "Randy.Dunlap" <rddunlap@osdl.org>
Cc: <linux-kernel@vger.kernel.org>, <piggin@cyberone.com.au>
Subject: Re: sched isolcpus=1 related OOPS in 2.6.9
Date: Fri, 3 Dec 2004 18:46:52 +0100 (CET)	[thread overview]
Message-ID: <Pine.LNX.4.33.0412031839270.493-100000@devix> (raw)
In-Reply-To: <41B09FFD.6070706@osdl.org>

> > Well, I have more info. I setup bochs smp emulator and hacked
> > printk to output into e9 port which is then directed to a file.
> > Also I turned sched_domains debugging. From the result (below)
> > is clear that there is bug in isolated domains setup.
>
> You are correct, of course.  If "isolcpus" is used, the isolated
> cpu(s) (in <cpu_isolated_map>) are not init like the remaining
> cpus are.

The problem seems worse to me. See:
Brought up 2 CPUs
<7>CPU0:  online
<7> domain 0: span 3
<7>  groups: 1[128] 2[0]
<7>ERROR groups don't span domain->span
<7>CPU1:  online
<7> domain 0: span 2
<7>ERROR domain->cpu_power not set
<7>  groups: 2[0]
<1>divide error: 0000 [#1]

I added code which dumps cpu_power for each group (in brackets)
and it seems that only for the first group the power is computed
(even for regular non isolated domain).
Also the span for cpu0 should be 1 (it can be fixed by:
  sd->span = cpu_possible_map; => sd->span = cpu_default_map;
at line 4484).
Even then the group list is still bad. I'll dig more into it...

devik


  reply	other threads:[~2004-12-03 17:47 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2004-12-02 15:42 devik
2004-12-03 16:28 ` devik
2004-12-03 17:18   ` Randy.Dunlap
2004-12-03 17:46     ` devik [this message]
2004-12-03 18:15     ` [PATCH] " devik
2004-12-03 19:47       ` Dimitri Sivanich
2004-12-03 20:21         ` devik

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=Pine.LNX.4.33.0412031839270.493-100000@devix \
    --to=devik@cdi.cz \
    --cc=linux-kernel@vger.kernel.org \
    --cc=piggin@cyberone.com.au \
    --cc=rddunlap@osdl.org \
    /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

all inboxes | Powered by JetHome®