Problems compiling Mozilla with GCC 4.5

Problems compiling Mozilla with GCC 4.5

[Ehren Metcalfe]

I've recently come across a couple of issues trying to compile Firefox
trunk with 4.5 (I have a very simple plugin that I need to run on
mozilla-central). I've posted two bugs here
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42139 and here
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42171

I''m not particularly concerned about the first, which can be bypassed
by disabling optimization on the problem file, but the second, a link
error, is a show stopper. Can anyone think of a way to sidestep the
problem? I'm not even sure I've addressed the bug to the right place,
but it would seem that if the same version of ld chokes during a 4.5
compile, but does fine with an earlier version of gcc, then it's
really not a binutils issue. I'm a bit out of my element here though
so any suggestions would be welcome.

Ehren

Re: Problems compiling Mozilla with GCC 4.5

[Andrew Haley]

Ehren Metcalfe wrote:
> I've recently come across a couple of issues trying to compile Firefox
> trunk with 4.5 (I have a very simple plugin that I need to run on
> mozilla-central). I've posted two bugs here
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42139 and here
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42171
> 
> I''m not particularly concerned about the first, which can be bypassed
> by disabling optimization on the problem file, but the second, a link
> error, is a show stopper. Can anyone think of a way to sidestep the
> problem? I'm not even sure I've addressed the bug to the right place,
> but it would seem that if the same version of ld chokes during a 4.5
> compile, but does fine with an earlier version of gcc, then it's
> really not a binutils issue. I'm a bit out of my element here though
> so any suggestions would be welcome.

Is the symbol `nsAccessibleWrap::ReturnString(nsAString_internal&)::returnedString'
defined anywhere?

Andrew.

Re: Problems compiling Mozilla with GCC 4.5

[Andrew Pinski]

On Wed, Nov 25, 2009 at 1:17 AM, Andrew Haley <aph@redhat.com> wrote:
>
> Is the symbol `nsAccessibleWrap::ReturnString(nsAString_internal&)::returnedString'
> defined anywhere?

This sounds like bug 41611.  Is
nsAccessibleWrap::ReturnString(nsAString_internal&) an inline function
(which causes that variable to be vague linkage)?

Thanks,
Andrew Pinski

Re: Problems compiling Mozilla with GCC 4.5

[Ehren Metcalfe]

Yup, ReturnString is inline:
http://hg.mozilla.org/mozilla-central/file/d76583175408/accessible/src/atk/nsAccessibleWrap.h#l114

I'll change it and try a build,

Ehren

On Wed, Nov 25, 2009 at 4:24 AM, Andrew Pinski <pinskia@gmail.com> wrote:
> On Wed, Nov 25, 2009 at 1:17 AM, Andrew Haley <aph@redhat.com> wrote:
>>
>> Is the symbol `nsAccessibleWrap::ReturnString(nsAString_internal&)::returnedString'
>> defined anywhere?
>
> This sounds like bug 41611.  Is
> nsAccessibleWrap::ReturnString(nsAString_internal&) an inline function
> (which causes that variable to be vague linkage)?
>
> Thanks,
> Andrew Pinski
>

Re: Problems compiling Mozilla with GCC 4.5

[Ehren Metcalfe]

I just changed it,

Firefox now compiles with 4.5!

Thanks for your help,

Ehren

On Wed, Nov 25, 2009 at 8:11 AM, Ehren Metcalfe <ehren.m@gmail.com> wrote:
> Yup, ReturnString is inline:
> http://hg.mozilla.org/mozilla-central/file/d76583175408/accessible/src/atk/nsAccessibleWrap.h#l114
>
> I'll change it and try a build,
>
> Ehren
>
> On Wed, Nov 25, 2009 at 4:24 AM, Andrew Pinski <pinskia@gmail.com> wrote:
>> On Wed, Nov 25, 2009 at 1:17 AM, Andrew Haley <aph@redhat.com> wrote:
>>>
>>> Is the symbol `nsAccessibleWrap::ReturnString(nsAString_internal&)::returnedString'
>>> defined anywhere?
>>
>> This sounds like bug 41611.  Is
>> nsAccessibleWrap::ReturnString(nsAString_internal&) an inline function
>> (which causes that variable to be vague linkage)?
>>
>> Thanks,
>> Andrew Pinski
>>
>