970917 on mips-sgi-irix6.2, haifa disabled: many problems

[Zack Weinberg]

On Mon, 29 Sep 1997 11:25:29 +0100, Chris Richardson wrote:
> 
> On Sun, 28 Sep 1997 17:16:03 -0400, Zack Weinberg
>  wrote:
> > I've managed to reproduce the problem with closer monitoring, and it
> > looks like egcs tickles a kernel bug which causes any exec() to block
> > forever, unkillably.  960924 does this every time I run a `make check',
> > I think always at c-torture/execute/961125-1.c; however, I can't
> > duplicate the problem by compiling that file by hand.  IRIX has
> > historically been plagued by bugs of this kind; I will report it to SGI
> > and let you know if they respond.
> 
> I'm having a similar problem to the original poster (sorry, I've lost
> the attribution).  Making gcc, either by "make bootstrap" or in stages
> causes my machine to lock up completely at random stages in the
> compilation.  The point at which it locks up isn't predictable.  It has
[...]
> 
> I'm using the bog-standard SG C compiler from IRIX6.2 (not the MIPSPro)
> on an Indy R5000 with 64MByte RAM, several hundred MBytes free hard
> disk space.  It runs IRIX 6.2, with patches that were appropriate when
> it was purchased (this time last year).

Except for the compilers, that's exactly the same configuration I'm
having problems with.

What exactly do you mean by `lock up completely?' In my case all
programs that are running keep running, and fork() works, but anything
that calls exec() goes into the kernel and never comes out.  The
cpu states go to 99.9% "wait" and stay there.

If you start a `top' in one window and the make in the other, what
does the `top' window look like when you hit the wall?

zw

p.s. I have the most recent recommended patchset installed, so unless
there's an obscure patch, it's still a bug.

p.p.s.  Are you using NFS in any way?  It's usually my first suspect
when I have problems with SGIs.  In my case the machine is an nfs
client only, and the disk I'm compiling on is local.