696299 - Valgrind jobs are failing in js/src/shell/

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Jesse Ruderman]

https://tbpl.mozilla.org/php/getParsedLog.php?id=6947304&tree=Firefox&full=1#error0

`.L3239' referenced in section `.gnu.linkonce.r._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o): defined in discarded section `.gnu.linkonce.t._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o)
`.L3239' referenced in section `.gnu.linkonce.r._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o): defined in discarded section `.gnu.linkonce.t._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o)
`.L3240' referenced in section `.gnu.linkonce.r._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o): defined in discarded section `.gnu.linkonce.t._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o)
`.L3240' referenced in section `.gnu.linkonce.r._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o): defined in discarded section `.gnu.linkonce.t._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o)
`.L3241' referenced in section `.gnu.linkonce.r._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o): defined in discarded section `.gnu.linkonce.t._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o)
`.L3241' referenced in section `.gnu.linkonce.r._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o): defined in discarded section `.gnu.linkonce.t._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o)
`.L3242' referenced in section `.gnu.linkonce.r._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o): defined in discarded section `.gnu.linkonce.t._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o)
`.L3242' referenced in section `.gnu.linkonce.r._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o): defined in discarded section `.gnu.linkonce.t._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o)
`.L3243' referenced in section `.gnu.linkonce.r._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o): defined in discarded section `.gnu.linkonce.t._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o)
`.L3243' referenced in section `.gnu.linkonce.r._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o): defined in discarded section `.gnu.linkonce.t._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o)
`.L3244' referenced in section `.gnu.linkonce.r._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o): defined in discarded section `.gnu.linkonce.t._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o)
`.L3244' referenced in section `.gnu.linkonce.r._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o): defined in discarded section `.gnu.linkonce.t._ZN2js12StringObject6createEP9JSContextP8JSString' of ../libjs_static.a(jsstr.o)
collect2: ld returned 1 exit status
make[4]: Leaving directory `/builds/slave/m-cen-lnx-valgrind/objdir/js/src/shell'
make[4]: *** [js] Error 1
make[3]: *** [libs] Error 2
make[3]: Leaving directory `/builds/slave/m-cen-lnx-valgrind/objdir/js/src'
make[2]: *** [libs_tier_js] Error 2
make[2]: Leaving directory `/builds/slave/m-cen-lnx-valgrind/objdir'
make[1]: *** [tier_js] Error 2
make[1]: Leaving directory `/builds/slave/m-cen-lnx-valgrind/objdir'
make: *** [default] Error 2
program finished with exit code 2

Comment 1

[[e-mail] (:bhearsum)]

Jesse, I'm not sure what RelEng can do here, it looks like a compilation issue that a developer needs to take a look at.

Comment 2

[Luke Wagner [:luke]]

This seems like a pretty kosher use of inlining.  Perhaps Julian or Nick know something on the valgrind end?

Comment 3

[Nicholas Nethercote [inactive]]

The full log is not available and comment 0 doesn't have enough context to know what's going on.

How are these valgrind-enabled builds built differently from normal builds?

Comment 4

[Chris AtLee [:catlee]]

They're using this script to build:
http://hg.mozilla.org/build/tools/file/default/scripts/valgrind/valgrind.sh

Comment 5

[Phil Ringnalda (:philor)]

https://tbpl.mozilla.org/php/getParsedLog.php?id=7703424&tree=Firefox&full=1 for Linux32, https://tbpl.mozilla.org/php/getParsedLog.php?id=7703438&tree=Firefox for Linux64, https://tbpl.mozilla.org/?noignore=1&jobname=Valgrind for finding logs.

Comment 6

[Nicholas Nethercote [inactive]]

(In reply to Chris AtLee [:catlee] from comment #4)
> They're using this script to build:
> http://hg.mozilla.org/build/tools/file/default/scripts/valgrind/valgrind.sh

The only vaguely unusual things I can see in there are --disable-elfhack (but that seems like it would avoid problems rather than cause them) and --freorder-blocks (but I don't see how that would cause problems).

Unrelatedly, you can use --smc-check=all-non-file instead of --smc-check=all and it'll run faster.

Comment 7

[Gary Kwong [:gkw] [:nth10sd] (NOT official MoCo now)]

(In reply to Nicholas Nethercote [:njn] from comment #6)
> Unrelatedly, you can use --smc-check=all-non-file instead of --smc-check=all
> and it'll run faster.

This requires at least Valgrind 3.7.0.

Comment 8

[(not currently active) Ted Mielczarek]

Er, that valgrind script appears to be using GCC 4.3.3, whereas the rest of our builds use GCC 4.5.2. We should probably fix that.

Comment 9

[(not currently active) Ted Mielczarek]

Still broken, I think the valgrind.sh script got broken along the way, it doesn't seem to be conveying its mozconfig options to configure. I have patches to (hopefully) fix that in bug 735309.

Comment 10

[(not currently active) Ted Mielczarek]

Switching to GCC 4.5 made these builds work again:
https://tbpl.mozilla.org/php/getParsedLog.php?id=10369320&tree=Firefox

They're still showing up as red on TBPL, but that's because Valgrind is apparently exiting with a non-zero status (and these jobs are run in a slightly unusual way).