Re: [racket-dev] racket 6.3 build failures on armel
Is it possible to get the output of
gcc -E -dM - < /dev/null
on that machine?
The JIT is sensitive to a number of preprocessor definitions, and I
might be able to provoke the buffer overflow by using the same values.
Segfaults or freezes could be the same problem, so the porterbox values
might also be useful.
Thanks!
At Wed, 09 Dec 2015 08:44:34 -0400, David Bremner wrote:
>
> I'm stuck figuring out some build-failures on armel.
>
> On the autobuilders, I get (twice, on two different autobuilders)
>
> ,----
> | Copying /«PKGBUILDDIR»/collects/racket/private/kernstruct.rkt to
> /«PKGBUILDDIR»/build/racket/gc2/xform-collects/racket/private/kernstruct.rkt
> | Copying /«PKGBUILDDIR»/collects/racket/private/norm-arity.rkt to
> /«PKGBUILDDIR»/build/racket/gc2/xform-collects/racket/private/norm-arity.rkt
> | Copying /«PKGBUILDDIR»/collects/racket/private/top-int.rkt to internal error
> in JIT;
> | ending address 0xb35bf650 not in [0xb35bf548,0xb35bf648] (0)
> | internal error: JIT buffer overflow
> `----
>
> There are several also a few warnings, I'm not sure if there is some
> kind of architecture detection failure.
>
>
> https://buildd.debian.org/status/fetch.php?pkg=racket&arch=armel&ver=6.3-1&stam
> p=1449355720
>
> On the porterbox (abel.debian.org), I can't duplicate that failure, but
> I get occasional (less than 1 build in 10) segfaults in racketcgc
>
> ,----
> | env XFORM_USE_PRECOMP=xsrc/precomp.h ../racketcgc -cqu
> /home/bremner/racket-6.3/src/racket/gc2/xform.rkt --setup . --depends \
> | --cpp "gcc -E -I./.. -I/home/bremner/racket-6.3/src/racket/gc2/../include
> -D_FORTIFY_SOURCE=2 -DUSE_SENORA_GC -D_LARGEFILE\
> | _SOURCE -D_FILE_OFFSET_BITS=64 " --keep-lines -o xsrc/dynext.c
> /home/bremner/racket-6.3/src/racket/gc2/../src/dynext.c
> | Segmentation fault
> | Makefile:216: recipe for target 'xsrc/compile.c' failed
> | make[5]: *** [xsrc/compile.c] Error 139
> `----
>
> Finally, once (out of about 25 builds), racketcgc was still running 9.5h
> later, and ignoring SIGTERM. It was aparently processing char.c
>
> --
> You received this message because you are subscribed to the Google Groups
> "Racket Developers" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to racket-dev+unsubscribe@googlegroups.com.
> To post to this group, send email to racket-dev@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/racket-dev/87h9jryeod.fsf%40zancas.localnet.
> For more options, visit https://groups.google.com/d/optout.
Reply to: