423913 - -fshort-wchar does not work on gcc 4.2.1 in arm scratchbox

Status: RESOLVED FIXED

(Firefox Build System :: General, defect)

Description

[Doug Turner (:dougt)]

currently configure.in passes -fshort-wchar to gcc.  this doesn't work when building against the maemo sdk:

[sbox-arm3: ~] > cat t.c
int main()
{
        return 0;
}
[sbox-arm3: ~] > gcc -fshort-wchar t.c
/scratchbox/compilers/arm-linux-cs2007q3-51sb1/bin/../lib/gcc/arm-none-linux-gnueabi/4.2.1/../../../../arm-none-linux-gnueabi/bin/ld: ERROR: /tmp/ccoJUOWj.o: Conflicting definitions of wchar_t


The work around is to set an env flag:

export SBOX_BLOCK_CROSS_COMPILER_ARGS="-fshort-wchar"

Comment 1

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

The configure check is somewhat bogus; it's assuming that -fshort-wchar works even though binaries that result with it set don't run.  It should probably be skipped (possibly entirely -- what platforms is it needed on?), or at least made into an optional --configure setting (--with-short-wchar or something).

Comment 2

[Doug Turner (:dougt)]

see bug 139211

Comment 3

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

I dunno if it's 139211 -- that's separate breakage for when it's available for g++ but not gcc or somesuch, no?

Comment 4

[Mike Hommey [:glandium]]

This could just be a toolchain problem. Both "compiler -fshort-wchar option" and "correct temporary object destruction order" were passing, before... See http://buildd.debian.org/fetch.cgi?pkg=xulrunner;ver=1.8.1.12-5;arch=armel;stamp=1206621388 for example.

Comment 5

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Could be, but either way we should check instead of always setting it, no?

Comment 6

[Mike Hommey [:glandium]]

A proposed patch on the debian bug for the same issue is to turn AC_TRY_COMPILE into a AC_TRY_LINK for the -fshort-wchar option test, which makes it fail, thus making -fshort-wchar not used.

Comment 7

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Yep, that would work.

Comment 8

[Brad Lassey [:blassey] (use needinfo?)]

Yup, AC_TRY_LINK works for me.

Comment 9

[Mike Hommey [:glandium]]

Attachment: patch

Comment 10

[Benjamin Smedberg]

Comment on attachment 317321 [details] [diff] [review]
patch

Please re-indent the remaining code

Comment 11

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Checked in, with reindentation.

Comment 12

[Mike Hommey [:glandium]]

Where was that checked in ?

Comment 13

[:Gavin Sharp [email: [e-mail]]]

mozilla-central: http://hg.mozilla.org/index.cgi/mozilla-central/log/378707d75b14