8-current: linux_dist-gentoo-stage3 wonn't bootstrap

[Arno J. Klaassen]

"Arno J. Klaassen" <arno at heho.snv.jussieu.fr> writes:

> [NB, subject changed; this bug is produce now on -current ]
> 
> Roman Divacky <rdivacky at freebsd.org> writes:
> 
> > On Sun, Dec 28, 2008 at 10:42:44PM +0100, Arno J. Klaassen wrote:
> > > Roman Divacky <rdivacky at freebsd.org> writes:
> > > 
> > > > On Sun, Dec 28, 2008 at 09:38:09PM +0100, Arno J. Klaassen wrote:
> > > > > 
> > > > > Roman Divacky <rdivacky at freebsd.org> writes:
> > > > > 
> > > > > > > > 
> > > > > > > > does this patch fix the hang? 
> > > > > > > > 
> > > > > > > > www.vlakno.cz/~rdivacky/linprocfs.patch
> > > > > > > 
> > > > > > > nope, though it does fix the lock order reversal 
> > > > > > > (I attach the slightly modified patch for linprocfs.c to
> > > > > > > make it compile)
> > > > > >  
> > > > > > the LOR is probably harmless.... dont bother with that patch :)
> > > > > > 
> > > > > > > funny enough, it again hangs in compiling gconv_simple.c 
> > > > > > > with cc1 in pipewr state and no assembler showing up in
> > > > > > > ps(1) after having succesfully compiled a bunch of other
> > > > > > > files.
> > > > > > 
> > > > > > you mean native cc1? or linux one?
> > > > > 
> > > > > linux :
> > > > > 
> > > > > # ps axuww | fgrep 1129
> > > > > root       11290  0.0  0.1  2296  1432   0  D     8:29PM   0:00.01 [i486-pc-linux-gnu-g]
> > > > > root       11291  0.0  0.8 18152 16808   0  I     8:29PM   0:00.96 [cc1]
> > > > > 
> > > > > and no trace of gconv_simple.o
> > > > > 
> > > > > last line of log-file is :
> > > > > 
> > > > > i486-pc-linux-gnu-gcc gconv_simple.c -c -std=gnu99  -O2 -Wall -Winline -Wwrite-strings -fmerge-all-constants -fno-strict-aliasing -mtune=i686 -pipe -Wstrict-prototypes -mpreferred-stack-boundary=2     -I../include -I/var/tmp/portage/sys-libs/glibc-2.6.1/work/build-default-i486-pc-linux-gnu-nptl/iconv -I/var/tmp/portage/sys-libs/glibc-2.6.1/work/build-default-i486-pc-linux-gnu-nptl -I../sysdeps/i386/elf -I../nptl/sysdeps/unix/sysv/linux/i386/i486 -I../nptl/sysdeps/unix/sysv/linux/i386 -I../sysdeps/unix/sysv/linux/i386 -I../nptl/sysdeps/unix/sysv/linux -I../nptl/sysdeps/pthread -I../sysdeps/pthread -I../ports/sysdeps/unix/sysv/linux -I../sysdeps/unix/sysv/linux -I../sysdeps/gnu -I../sysdeps/unix/common -I../sysdeps/unix/mman -I../sysdeps/unix/inet -I../ports/sysdeps/unix/sysv/i386 -I../sysdeps/unix/sysv/i386 -I../nptl/sysdeps/unix/sysv -I../ports/sysdeps/unix/sysv -I../sysdeps/unix/sysv -I../sysdeps/unix/i386 -I../nptl/sysdeps/unix -I../ports/sysdeps/unix -I../sysdep!
 s/!
>  unix!
> > >   -I.!
> > > > >  ./sysdeps/posix -I../nptl/sysdeps/i386/i486 -I../sysdeps/i386/i486 -I../sysdeps/i386/fpu -I../nptl/sysdeps/i386 -I../sysdeps/i386 -I../sysdeps/wordsize-32 -I../sysdeps/ieee754/ldbl-96 -I../sysdeps/ieee754/dbl-64 -I../sysdeps/ieee754/flt-32 -I../sysdeps/ieee754 -I../sysdeps/generic/elf -I../sysdeps/generic -I../nptl -I../ports  -I.. -I../libio -I. -nostdinc -isystem /usr/lib/gcc/i486-pc-linux-gnu/4.1.2/include -isystem /usr/lib/gcc/i486-pc-linux-gnu/4.1.2/include-fixed -isystem /usr/include -D_LIBC_REENTRANT -include ../include/libc-symbols.h        -o /var/tmp/portage/sys-libs/glibc-2.6.1/work/build-default-i486-pc-linux-gnu-nptl/iconv/gconv_simple.o -MD -MP -MF /var/tmp/portage/sys-libs/glibc-2.6.1/work/build-default-i486-pc-linux-gnu-nptl/iconv/gconv_simple.o.dt -MT /var/tmp/portage/sys-libs/glibc-2.6.1/work/build-default-i486-pc-linux-gnu-nptl/iconv/gconv_simple.o
> > > > > 
> > > > > 
> > > > > ++, Arno
> > > > >  
> > > > 
> > > > can you break into DDB and extract a backtrace of the stuck process?
> > > 
> > > OK, I'll hook up a serconsole tomorrow ..
> > > 
> > > for now some good old quick n dirty copy paste by pen and paper :
> > > 
> > >  11290 (i486-pc-linux-gnu-gcc]
> > > 
> > >   sched_switch
> > >   mi_switch
> > >   sleepq_switch
> > >   sleepq_wait
> > >   _sleep
> > >   linux_vfork
> > >   ia32_syscall
> > >   Xint0x80_syscall
> > >    syscall (190, Linux ELF32, linux_vfork)
> > > 
> > > 
> > >  11291 (cc1)
> > > 
> > >   sched_switch
> > >   mi_switch
> > >   sleepq_switch
> > >   sleepq_catch_signals
> > >   sleepq_wait_sig
> > >   _sleep
> > >   pipe_write
> > >   dofilewrite
> > >   kern_writev
> > >   write
> > >   ia32_syscall
> > >   Xint80_syscall
> > >     syscall (4, Linux ELF32, write)
> > 
> > I dont see anything obvious.. can you do "ps axl" and see what the MWCHAN is?
> > that might shed some light to this...
> 
> bon, in fact when I launce the command by hand adding a '-v ' to gcc, output
> says :
> 
> /usr/libexec/gcc/i486-pc-linux-gnu/4.1.2/cc1 ... | /usr/lib/gcc/i486-pc-linux-gnu/4.1.2/../../../../i486-pc-linux-gnu/bin/as -V -Qy -o /var/tmp/portage/sys-libs/glibc-2.6.1/work/build-default-i486-pc-linux-gnu-nptl/iconv/gconv_simple.o -
> 
> 
> but, cc1 is stuck in 'pipewr' and no .../bin/as process is running (any longer)
> nor the file .../gconv_simple.o created 

I investigated this a bit more (still chrooted to /usr/local/gentoo-stage3) :

  i486-pc-linux-gnu-gcc gconv_open.c -c                 : OK
  i486-pc-linux-gnu-gcc gconv_open.c -c -pipe           : OK

  i486-pc-linux-gnu-gcc gconv_simple.c -c               : OK
  i486-pc-linux-gnu-gcc gconv_simple.c -c -pipe         : pipe to as(1) fails

I tried to play with kern.ipc.maxpipekva (I vaguely remember that
was necessary for linux testsuites) pumping it up to "536608768", but
no difference. If someone has another suggestion?

Thanx, Arno