curreent build: internal compiler error: Segmentation fault

[CivilCK]

Hi,

I got this error message while building current distribution, 4.99.20



<built-in>:0: internal compiler error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://www.NetBSD.org/Misc/send-pr.html> for instructions.

*** Failed target:  ____semctl13.o



Regards,

[CivilCK]

ck74 wrote:

Hi,
I got this error message while building current distribution, 4.99.20

<built-in>:0: internal compiler error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://www.NetBSD.org/Misc/send-pr.html> for instructions.

*** Failed target:  ____semctl13.o

Regards,

*** Failed target:  ____semctl13.o
*** Failed command: /usr/src/../tools/bin/i386--netbsdelf-gcc -I/usr/src/lib/libc/arch/i386 -nostdinc -idirafter /usr/src/../obj/destdir.i386/usr/include -fstack-protector -Wstack-protector --param ssp-buffer-size=1 -D_LIBC -DLIBC_SCCS -DSYSLIBC_SCCS -D_REENTRANT -DHESIOD -DINET6 -DNLS -DYP -I/usr/src/lib/libc/include -I/usr/src/lib/libc -I/usr/src/sys -I/usr/src/lib/libc/compat/stdlib -I/usr/src/lib/libc/compat/../stdlib -I/usr/src/lib/libc/../../common/lib/libc/quad -I/usr/src/lib/libc/../../common/lib/libc/string -I/usr/src/lib/libc/../../common/lib/libc/arch/i386/string -D__DBINTERFACE_PRIVATE -DI18NMODULE_MAJOR=4 -DCITRUS -I/usr/src/libexec/ld.elf_so -I/usr/src/lib/libc/dlfcn -I/usr/src/lib/libc/gdtoa -DNO_FENV_H -I/usr/src/lib/libc/arch/i386/gdtoa -DCITRUS_ICONV -DWITH_RUNE -I/usr/src/lib/libc -DPOSIX_MISTAKE -DCOMPAT__RES -DUSE_POLL -DPORTMAP -DWIDE_DOUBLE -DALL_STATE -DUSG_COMPAT -D_FORTIFY_SOURCE=2 -nostdinc -isystem /usr/src/../obj/destdir.i386/usr/include -c -traditional-cpp ____semctl13.S -o ____semctl13.o
*** Error code 1

Stop.
nbmake: stopped in /usr/src/lib/libc

*** Failed target:  dependall
*** Failed command: cd /usr/src/lib/libc; /usr/src/../tools/bin/nbmake realall
*** Error code 1

[Steven Bellovin]

On Sun, 3 Jun 2007 16:30:02 -0700 (PDT)
ck74 <ck74@...> wrote:

>
>
>
> Hi,
>
> I got this error message while building current distribution, 4.99.20
>
>
>
> <built-in>:0: internal compiler error: Segmentation fault
> Please submit a full bug report,
> with preprocessed source if appropriate.
> See <URL:http://www.NetBSD.org/Misc/send-pr.html> for instructions.
>
> *** Failed target:  ____semctl13.o
>

Retry the build.  If that file compiles properly, you probably have
memory problems.  Install and run pkgsrc/sysutils/memtestplus...


                --Steve Bellovin, http://www.cs.columbia.edu/~smb

[CivilCK]

Hi,

It seems there is a problem within source... or at least not memory problem.

I have just did "# cd /usr ; cvs checkout -P src" and then tried to re-build and the result is same.

Also just after it fails:
-------------------------
Memory: 449M Act, 252M Inact, 1208K Wired, 70M Exec, 274M File, 69M Free
Swap: 1999M Total, 77M Used, 1922M Free
-------------------------


Regards,

Steven M. Bellovin wrote:

On Sun, 3 Jun 2007 16:30:02 -0700 (PDT)
ck74 <[e-mail]> wrote:

>
>
>
> Hi,
>
> I got this error message while building current distribution, 4.99.20
>
>
>
> <built-in>:0: internal compiler error: Segmentation fault
> Please submit a full bug report,
> with preprocessed source if appropriate.
> See <URL:http://www.NetBSD.org/Misc/send-pr.html> for instructions.
>
> *** Failed target:  ____semctl13.o
>
Retry the build.  If that file compiles properly, you probably have
memory problems.  Install and run pkgsrc/sysutils/memtestplus...


                --Steve Bellovin, http://www.cs.columbia.edu/~smb

[Matthias Scheler-2]

On Sun, Jun 03, 2007 at 05:34:45PM -0700, ck74 wrote:

> It seems there is a problem within source... or at least not memory problem.
>
> I have just did "# cd /usr ; cvs checkout -P src" and then tried to re-build
> and the result is same.
>
> Also just after it fails:
> -------------------------
> Memory: 449M Act, 252M Inact, 1208K Wired, 70M Exec, 274M File, 69M Free
> Swap: 1999M Total, 77M Used, 1922M Free
> -------------------------

Does it always fail on the same file? Or does it fail on random locations
which change every time you try?

If it fails and random locations and gets a little bit further everytime
you invoke "make" (or "build.sh") it is almost certainly a problem
with the memory in the box.

GCC is the best memory tester I know. It works much better than "memtest",
"memtestplus" and all the other tools.

        Kind regards

--
Matthias Scheler                                  http://zhadum.org.uk/

[Allen Briggs]

On Mon, Jun 04, 2007 at 12:55:38PM +0100, Matthias Scheler wrote:
> If it fails and random locations and gets a little bit further everytime
> you invoke "make" (or "build.sh") it is almost certainly a problem
> with the memory in the box.

I agree, however I just saw a couple of things that look a little
like memory errors, too.  In a box that's been solid for some time.
I was wondering if it might be related to my aac(4) changes (not
all of which are committed yet) or to something in -current.

A kernel from sources a week or two back paniced sometime last week
with a "bad dir" during the nightly script runs.  I was out of town
and didn't catch just when in paniced.

One build (-U -m amd64 -j2 on /i386) errored out with a problem
looking for and not finding src/bin/rcorder/Makefile (it's in sbin,
and there's no reason I could see for it to be looking in bin).
A subsequent build with the same options went fine--nothing more
than a 'cvs -q up -PdA' in between.  And a third build with the
same options seems to be running OK, too.

So it might still be a memory problem.  I have new RAM arriving
today (the intent being to go from 2G -> 6G and transition from
/i386 to /amd64 on this box).  So I can try swapping out sticks if
I can get to a fairly reproducible error.

-allen

--
Allen Briggs  |  http://www.ninthwonder.com/~briggs/  |  briggs@...

[CivilCK]

Yes, it was the same file... That was why i said "same". I will remove obj and tools dirs and will re-try.

Regards,

Matthias Scheler-2 wrote:

On Sun, Jun 03, 2007 at 05:34:45PM -0700, ck74 wrote:
> It seems there is a problem within source... or at least not memory problem.
>
> I have just did "# cd /usr ; cvs checkout -P src" and then tried to re-build
> and the result is same.
>
> Also just after it fails:
> -------------------------
> Memory: 449M Act, 252M Inact, 1208K Wired, 70M Exec, 274M File, 69M Free
> Swap: 1999M Total, 77M Used, 1922M Free
> -------------------------

Does it always fail on the same file? Or does it fail on random locations
which change every time you try?

If it fails and random locations and gets a little bit further everytime
you invoke "make" (or "build.sh") it is almost certainly a problem
with the memory in the box.

GCC is the best memory tester I know. It works much better than "memtest",
"memtestplus" and all the other tools.

        Kind regards

--
Matthias Scheler                                  http://zhadum.org.uk/

[Allen Briggs]

On Mon, Jun 04, 2007 at 08:29:34AM -0400, Allen Briggs wrote:
> A kernel from sources a week or two back paniced sometime last week
> with a "bad dir" during the nightly script runs.  I was out of town
> and didn't catch just when in paniced.
>
> One build (-U -m amd64 -j2 on /i386) errored out with a problem
> looking for and not finding src/bin/rcorder/Makefile (it's in sbin,
> and there's no reason I could see for it to be looking in bin).

It's possible that these were both bogons.  It seems that I zeroed
out the drive under a mounted filesystem (/mnt was a scratch fs) to
get the bad dir panic, and the build error could be explained by
an interrupted cvs update that left old, stale directories around
(so the make machinery picked up an empty directory).

My apologies for jumping the gun there.  I recalled the zeroing
when I went to work on that disk again today.

-allen

--
Allen Briggs  |  http://www.ninthwonder.com/~briggs/  |  briggs@...