Unable to compile js shells on mozilla-aurora

RESOLVED FIXED in Firefox 35

Status

()

defect
--
blocker
RESOLVED FIXED
5 years ago
5 years ago

People

(Reporter: gkw, Assigned: sfink)

Tracking

({regression})

Trunk
mozilla38
x86_64
macOS
Points:
---
Bug Flags:
qe-verify -

Firefox Tracking Flags

(firefox35 fixed, firefox36 fixed, firefox37 fixed, firefox38 fixed)

Details

Attachments

(2 attachments)

Reporter

Description

5 years ago
Posted file compilation error
I hit a compile failure when compiling mozilla-aurora rev 3e672f58630b.

CC="clang -Qunused-arguments" CXX="clang++ -Qunused-arguments" AR=ar AUTOCONF=/usr/local/Cellar/autoconf213/2.13/bin/autoconf213 sh /Users/skywalker/trees/mozilla-aurora/js/src/configure --target=x86_64-apple-darwin12.5.0 --enable-debug --enable-optimize --enable-nspr-build --enable-more-deterministic --with-ccache --enable-gczeal --enable-debug-symbols --disable-tests

===

TestingFunctions.o
/usr/local/bin/ccache clang++ -Qunused-arguments -o TestingFunctions.o -c  -fvisibility=hidden -DIMPL_MFBT -DEXPORT_JS_API -DAB_CD= -DNO_NSPR_10_SUPPORT -I/Users/skywalker/trees/mozilla-aurora/js/src -I. -I/Users/skywalker/trees/mozilla-aurora/intl/icu/source/common -I/Users/skywalker/trees/mozilla-aurora/intl/icu/source/i18n -I../../dist/include   -I/Users/skywalker/Desktop/shell-cache/js-dbg-opt-64-dm-nsprBuild-darwin-3e672f58630b/objdir-js/dist/include/nspr        -fPIC  -Qunused-arguments  -DMOZILLA_CLIENT -include ../../js/src/js-confdefs.h -MD -MP -MF .deps/TestingFunctions.o.pp -Qunused-arguments  -Qunused-arguments -Wall -Wsign-compare -Wtype-limits -Wno-invalid-offsetof -Wno-c++0x-extensions -Wno-extended-offsetof -Wno-unknown-warning-option -Wno-return-type-c-linkage -fno-common -fno-rtti -fno-exceptions -fno-math-errno -std=gnu++0x -pthread -pipe  -DDEBUG -DTRACING -g -O3 -fno-stack-protector -fno-omit-frame-pointer      /Users/skywalker/trees/mozilla-aurora/js/src/builtin/TestingFunctions.cpp
/Users/skywalker/trees/mozilla-aurora/js/src/builtin/TestingFunctions.cpp:1819:5: error: use of undeclared identifier 'JS_smprintf_free'
    JS_smprintf_free(buf);
    ^
1 error generated.
make[3]: *** [TestingFunctions.o] Error 1
make[2]: *** [js/src/target] Error 2
make[1]: *** [compile] Error 2
make: *** [default] Error 2

===

autoBisect shows this is probably related to the following changeset:

The first bad revision is:
changeset:   217272:aed50d3edf33
user:        ffxbld
date:        Tue Sep 02 12:54:31 2014 -0400
summary:     Update configs. IGNORE BROKEN CHANGESETS CLOSED TREE NO BUG a=release ba=release

This seems wrong, and I went to investigate more:

http://hg.mozilla.org/releases/mozilla-aurora/rev/549b4ef82544

That line was entered in the following changeset. Setting needinfo? from :sfink as he seemed to have written that patch.
Flags: needinfo?(sphink)
Assignee

Comment 1

5 years ago
GetBacktrace requires JS_smprintf_free to discard the results of FormatStackDump, but it is from jsprf.h which is not included in JS_MORE_DETERMINISTIC builds. I don't think there's anything nondeterministic in the getBacktrace() output, so I believe the fix is to always #include it again.
Attachment #8548538 - Flags: review?(n.nethercote)
Assignee

Updated

5 years ago
Assignee: nobody → sphink
Status: NEW → ASSIGNED
Attachment #8548538 - Flags: review?(n.nethercote) → review+
Assignee

Comment 3

5 years ago
Comment on attachment 8548538 [details] [diff] [review]
GetBacktrace needs to be able to free the results buffer

Approval Request Comment
[Feature/regressing bug #]: bug 1015339
[User impact if declined]: none, this is relevant to fuzz testing only
[Describe test coverage new/current, TBPL]: on inbound
[Risks and why]: "no" risk. JS_MORE_DETERMINISTIC is only defined for fuzzing, so this does not change shipped builds.
[String/UUID change made/needed]: none
Flags: needinfo?(sphink)
Attachment #8548538 - Flags: approval-mozilla-aurora?
Reporter

Comment 4

5 years ago
We should also backport this to mozilla-beta (or maybe mozilla-release?).

The regressor landed in: https://bugzilla.mozilla.org/show_bug.cgi?id=1015339#c30

(and we didn't notice earlier because fuzzing other repositories besides mozilla-central was disabled for awhile)
Flags: needinfo?(sphink)
https://hg.mozilla.org/mozilla-central/rev/30796eecd360
Status: ASSIGNED → RESOLVED
Closed: 5 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla38
Assignee

Comment 6

5 years ago
(In reply to Gary Kwong [:gkw] [:nth10sd] from comment #4)
> We should also backport this to mozilla-beta (or maybe mozilla-release?).
> 
> The regressor landed in:
> https://bugzilla.mozilla.org/show_bug.cgi?id=1015339#c30

Uh, and that bug is marked milestone 37, which seems to be aurora now. Why would we need to backport further? Am I getting versions mixed up again?
Flags: needinfo?(sphink) → needinfo?(gary)
Reporter

Comment 7

5 years ago
(In reply to Steve Fink [:sfink, :s:] from comment #6)
> Uh, and that bug is marked milestone 37, which seems to be aurora now. Why
> would we need to backport further? Am I getting versions mixed up again?

That's because the follow-up only landed in v37, but the regressor landed before that:

http://hg.mozilla.org/releases/mozilla-aurora/rev/549b4ef82544
http://hg.mozilla.org/releases/mozilla-beta/rev/549b4ef82544
http://hg.mozilla.org/releases/mozilla-release/rev/549b4ef82544

If you check the date that the regressor landed, it was approximately 14 Aug 2014, which means it would have been in Fx34 as per:

https://wiki.mozilla.org/RapidRelease/Calendar
Flags: needinfo?(gary) → needinfo?(sphink)
Assignee

Comment 8

5 years ago
Ah, ok. Can you request approval for whatever branches you would like to run fuzzing on, then? I don't know which of the esr and other branches you might want.
Flags: needinfo?(sphink) → needinfo?(gary)
Assignee

Comment 9

5 years ago
Comment on attachment 8548538 [details] [diff] [review]
GetBacktrace needs to be able to free the results buffer

Approval Request Comment
See above comment for the aurora request.

On irc:

<gkw> sfink: just mozilla-beta and mozilla-release
<gkw> sfink: regressor didn't land on esr, and I'm not sure if it will land on -release, but we should need at least -beta
Flags: needinfo?(gary)
Attachment #8548538 - Flags: approval-mozilla-release?
Attachment #8548538 - Flags: approval-mozilla-beta?
Comment on attachment 8548538 [details] [diff] [review]
GetBacktrace needs to be able to free the results buffer

approving for branches so fuzzing can get running again.
Attachment #8548538 - Flags: approval-mozilla-release?
Attachment #8548538 - Flags: approval-mozilla-release+
Attachment #8548538 - Flags: approval-mozilla-beta?
Attachment #8548538 - Flags: approval-mozilla-beta+
Attachment #8548538 - Flags: approval-mozilla-aurora?
Attachment #8548538 - Flags: approval-mozilla-aurora+
Flags: qe-verify-
You need to log in before you can comment on or make changes to this bug.