build sage-6.6 R fail

[William Stein]

Hi,

I'm trying to build Sage-6.6 on the latest Ubuntu 14.10
(SageMathCloud). The first problem was with libgd, which another
patch somebody posted May 1 got past. Now there is some mysterious
error/crash when building R which means nothing to me. See attached
log. I'm building with MAKE not set and a very high RAM/cpu quota...

gcc -std=gnu99 -shared -L../../../../lib
-L/projects/8111fd7e-e470-49ed-996d-8bba01a6cd3e/sage-6.6/local/lib/
-o tools.so text.o init.o Rmd5.o md5.o signals.o install.o getfmts.o
http.o gramLatex.o gramRd.o -L../../../../lib -lR
make[9]: Entering directory
'/projects/8111fd7e-e470-49ed-996d-8bba01a6cd3e/sage-6.6/local/var/tmp/sage/build/r-3.1.2.p0/src/src/library/tools/src'
mkdir -p -- ../../../../library/tools/libs
make[9]: Leaving directory
'/projects/8111fd7e-e470-49ed-996d-8bba01a6cd3e/sage-6.6/local/var/tmp/sage/build/r-3.1.2.p0/src/src/library/tools/src'
make[8]: Leaving directory
'/projects/8111fd7e-e470-49ed-996d-8bba01a6cd3e/sage-6.6/local/var/tmp/sage/build/r-3.1.2.p0/src/src/library/tools/src'
make[7]: Leaving directory
'/projects/8111fd7e-e470-49ed-996d-8bba01a6cd3e/sage-6.6/local/var/tmp/sage/build/r-3.1.2.p0/src/src/library/tools'
/bin/bash: line 1: 11246 Done echo
"tools:::.install_package_description('.',
'"../../../library/tools"')"
11247 Aborted (core dumped) |
R_DEFAULT_PACKAGES=NULL ../../../bin/R --vanilla --slave > /dev/null
Makefile:30: recipe for target 'all' failed



--
William (http://wstein.org)

[leif]

William Stein wrote:
> I'm trying to build Sage-6.6 on the latest Ubuntu 14.10
> (SageMathCloud). The first problem was with libgd, which another
> patch somebody posted May 1 got past. Now there is some mysterious
> error/crash when building R which means nothing to me. See attached
> log. I'm building with MAKE not set and a very high RAM/cpu quota...

14.10 is not an LTS release... ;-)

Looks like Ubuntu's GCC is broken; I've seen such (the R binary
crashing) in the past when experimenting with LTO.

I'd play a little with CFLAGS. There's btw. a new R package at #18229
you could perhaps try; no idea whether that will make a difference.


-leif

[leif]

leif wrote:
> William Stein wrote:
>> I'm trying to build Sage-6.6 on the latest Ubuntu 14.10
>> (SageMathCloud). The first problem was with libgd, which another
>> patch somebody posted May 1 got past. Now there is some mysterious
>> error/crash when building R which means nothing to me. See attached
>> log. I'm building with MAKE not set and a very high RAM/cpu quota...
>
> 14.10 is not an LTS release... ;-)
>
> Looks like Ubuntu's GCC is broken; I've seen such (the R binary
> crashing) in the past when experimenting with LTO.
>
> I'd play a little with CFLAGS. There's btw. a new R package at #18229
> you could perhaps try; no idea whether that will make a difference.

P.S.: Another option is of course trying to build with
SAGE_INSTALL_GCC=yes (vanilla FSF GCC 4.9.2).

[Jeroen Demeyer]

On 2015-05-04 09:05, leif wrote:
> William Stein wrote:
>> I'm trying to build Sage-6.6 on the latest Ubuntu 14.10
>> (SageMathCloud). The first problem was with libgd, which another
>> patch somebody posted May 1 got past. Now there is some mysterious
>> error/crash when building R which means nothing to me. See attached
>> log. I'm building with MAKE not set and a very high RAM/cpu quota...
>
> 14.10 is not an LTS release... ;-)
>
> Looks like Ubuntu's GCC is broken; I've seen such (the R binary
> crashing) in the past when experimenting with LTO.

I cannot comment on this specific case, but I know from experience that
Debian has a history of shipping broken compilers...

[Volker Braun]

The libgd issue is fixed in #18293

[John Foster]

Just a suggestion:
Try a different make, as you know there are several. I see at the bottom
there is a core dump but no justification. Could be a memory addressing
issue. best wishes.

John Foster
JW Foster & Associates

[William Stein]

Hi,

Thanks for the feedback. I decided to (1) use sage-6.7.beta4, and
(2) upgrade to Ubuntu-15.04. The Sage build has gone perfectly with
this combination.

-- William

> --
> You received this message because you are subscribed to the Google Groups
> "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to sage-devel+...@googlegroups.com.
> To post to this group, send email to sage-...@googlegroups.com.
> Visit this group at http://groups.google.com/group/sage-devel.
> For more options, visit https://groups.google.com/d/optout.



--
William (http://wstein.org)

[leif]

John Foster wrote:
> Just a suggestion:
> Try a different make, as you know there are several.

You mean a different 'make' program?

There's nothing wrong with 'make' here; the R binary "crashed":


$ printf '#include<stdlib.h>\nint main(){abort();}' | gcc -o dump_core
-x c -

$ printf 'all:\n\t@echo | ./dump_core' | make -f - SHELL=/bin/bash
/bin/bash: line 1: 25600 Done echo
25601 Aborted | ./dump_core
make: *** [all] Error 134

$ ulimit -c unlimited

$ printf 'all:\n\t@echo | ./dump_core' | make -f - SHELL=/bin/bash
/bin/bash: line 1: 25607 Done echo
25608 Aborted (core dumped) | ./dump_core
make: *** [all] Error 134

> I see at the bottom
> there is a core dump but no justification. Could be a memory addressing
> issue. best wishes.

Memory addressing errors usually give SIGSEGV. The justification for
dumping a core file is SIGABRT, since that's the default action for it.


-leif

[William Stein]

On Wed, May 6, 2015 at 4:54 PM, William Stein <wst...@gmail.com> wrote:
> Hi,
>
> Thanks for the feedback. I decided to (1) use sage-6.7.beta4, and
> (2) upgrade to Ubuntu-15.04. The Sage build has gone perfectly with
> this combination.

This worked and all long tests passed.

William

--
William (http://wstein.org)