[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: linuxthreads & linux kernel 2.4.1
From: |
Oertel, Tim |
Subject: |
Re: linuxthreads & linux kernel 2.4.1 |
Date: |
Mon, 19 Feb 2001 14:13:54 -0500 |
Quoting Andreas Jaeger <address@hidden>:
> address@hidden writes:
>
> > Quoting Andreas Jaeger <address@hidden>:
> >
> > > address@hidden writes:
> > >
> > > > I am having difficulty building glibc-2.2.2 with linuxthreads.
> > > > The build works
> > > > completely, however in running the check, linuxthreads failed.
> > > > I did some investigation and found that mmap was failing with
> > > > ENOMEM. I hacked
> > > > __pthread_max_stacksize to be 0x003fffff, rather than the
> > > > original 0x7fffffff (based on rlim_cur).
> > > > This seemed to fix a portion of my difficulties, i.e. some of the
> > > > tests ran, but
> > > > some now fail with a core dump, and I am not sure gdb has given
> > > > me anything meaningful.
> > > >
> > > > I'm not convinced that it's a linuxthreads bug, but I thought
> > > > I'd ask.
> > > >
> > > > Any suggestions, either a fix, or suggestions for debugging this
> > > > further?
> > >
> > > Which architecture? What compiler?
> > >
> > > It works fine for me on i686-linux with gcc 2.95.2 and Linux
> > > 2.2.18,
> >
> > i686-pc-linux-gnu
> >
> > 2.95.2 gcc, freshly built
> > 3.79.1 make, freshly built
> > 2.10.1 binutils, freshly built
> > 2.4.1 Linux kernel
>
> Those should be fine.
>
> Did you run the glibc testsuite? Did it report any failures?
I ran 'make check' (the only testsuite, right?), and the thing ran fine (is
it obvious when something fails?) until linuxthreads. Initially, it hung
(or crashed, I don't remember ATM) in ex6, but in investigating, ex1-5 also
failed, due to the pthread_create() failing, due to the mmap failing.
After I lowered the __pthread_max_stacksize to 0x003fffff, ex1-4 ran
successfully, but I believe ex5 (and/or ex6) dumped core. I tried to gdb
it, but the stack didn't look correct, so either I didn't load it up
correctly, or the stack was hosed.
Since ex5 dumped core, I figured that there was something more fundemental
wrong than just the mmap(), thus my message here.
Tim
"He who possesses science and art also has religion;
but he who possesses neither of those two, let him have religion!" - Goethe