Re: Segfault compiling kernels
On 28 Dec, this message from Michael Schmitz echoed through cyberspace:
>> > > make[1]: *** [fastdep] Segmentation fault
>> > > make[1]: Leaving directory `/usr/src/linux/arch/ppc/boot'
>> > > make: *** [archdep] Error 2
>> > >
>> > > It seems make is to blame here??
>
> Might as well be mkdep:
>
> fastdep: dummy
> $(TOPDIR)/scripts/mkdep $(CFLAGS) $(EXTRA_CFLAGS) -- $(wildcard *.[chS]) > .depend
>
> Rebuild that (make distclean should throw away mkdep IIRC).
Done. Same thing.
> Please do also
> run mkdep under gdb control and report the resulting backtrace after
> segfault.
How do you find what command actually segfaults? make normally prints
out what it does before actually doing that. Well, for me it prints no
action after running the previous mkdep which completed ok. Something
like:
make[1] make <options
make [2] Entering blah/chrp
make [3] scripts/mkdep <options> > .depend
make [3] leaving blah/chrp
make [2] Segfault
make [1] *** Error blah
So the segfault happens between make recursion depth changes (that's
what's in brackets, right?). I tried running make under gdb, but since
it spawns another make (which spawns another one), those childs are not
running under gdb.
I'd run the command that segfaults directly, if I knew which one? Hm,
make -n comes to mind... I'll have to see what that gives.
>> > Spurious segfaults can only really be caused by libc, the kernel or
>> > hardware, or am I missing something?
For the record, I've run the same thing under a 2.4.16 kernel (compiled
somewhere else): same error.
>> I don't know if you can call these 'spurious', since they _always_ happen on kernel compile,
>> but (so far) on nothing else. I think memory has been exercised enough during install and
>> during first tests (running Mozilla and konqueror at the same time under Gnome ;-). It would
>> really be a very bad coincidence to always hit that bad memory block on kernel compile.
>>
>> Which leaves libc: but why doesn't anything else that uses libc segfault then?
>
> I'd blame libc though I really can't put a finger on why :-) If it's not a
> bad memory issue. mkdep uses realloc quite a lot.
Still: why is _nothing_ else crashing on this machine? During the
complete install process, running netscape, mozilla, konqueror, what
not... has not shown a single segfault.
Well, I'll keep on trying..
Cheers
Michel
-------------------------------------------------------------------------
Michel Lanners | " Read Philosophy. Study Art.
23, Rue Paul Henkes | Ask Questions. Make Mistakes.
L-1710 Luxembourg |
email mlan@cpu.lu |
http://www.cpu.lu/~mlan | Learn Always. "
Reply to: