[lkml]   [1997]   [Oct]   [6]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: QNX+2.1.57+SMP=OPPS

On Mon, 6 Oct 1997, linux kernel account wrote:

> Tyan Tomcat - III with 2/ 166MMXes..
> >
> > Can you give more details of the oops? (this could be mean copying down the
> > values by hand and then doing 'ksymoops ...')
> I should have a chance tommorrow evening..
> > Btw- I'm compiling with PGCC and -O2 so if this isn't happening for
> > anyone else then it might be because of that..
> >
> > Hmmm... could do. I stop using pgcc a while back because funny things
> > occurred, but since then I hear it has gotten much more stable.
> It's been real stable for me real recently.. I would find it suprising if
> it were PGCC esp since its -O2.. I'll try changing it too.. In about
> 8months of using PGCC I've never had it do something like that.. But.. You
> never know..

There are some problems with pgcc and embedded assembly in the kernel. I
doubt that your problem is the one I found last week since it was related
to inlined string-handling functions, and happened both at -O2 and -O6.
However, I have identified potential problems in some important kernel
source files. It is possible that the changes in the scheduler code
trigger a similar problem. The one I found was very easy to track down
(when compared to classical compiler bugs, which are ususally a nightmare).

The problem is not the compiler (at least in my case), but the fact that
it is much less forgiving than standard gcc is about not specifying
correctly the rules of the game on "asm" statements.


P.S: I tried to send this by private email to <>
but I have systematically received error messages from my mailer.

 \ /
  Last update: 2005-03-22 13:40    [W:0.037 / U:3.732 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site