Re: cc1plus invoked oom-killer: gfp_mask=0x280da, order=0, oom_adj=0

[Justin P. Mattock]

Dave Korn wrote:
> Justin Mattock wrote:
>
>
>    
> > ==1830== Command: c++ -o jsxml.o -c -DOSTYPE="Linux2.6" -DOSARCH=Linux
> >      
>
>    Ah, you're running it on the "c++" utility and it's reporting the stats for
> that, but how it works is that "c++" (and "gcc", "g++", et al) is just a
> driver, that parses the command line arguments and shells out to the actual
> compiler ("cc1plus"), assembler and linker to get them to do all the work.
>
>    If you add "-v --save-temps" to the c++ invocation, it'll show you the
> separate command lines it executes for the subprograms; the first invocation
> will be of cc1plus, using the -E flag to generate the preprocessed source into
> a .ii file, it's the second invocation you want, the one which uses the
> "-fpreprocessed" flag and names the .ii file as input, which is the one that
> actually then compiles the pre-processed source into assembly.  For fuller
> explanation, see the GCC wiki:
>
> http://gcc.gnu.org/wiki/DebuggingGCC
>
>      cheers,
>        DaveK
>
>
>
>    
o.k. after some reading and some experimenting,
I finally got firefox to compile all the way without a
memory leak..
here's what I found:
using the default .mozconfig I see during compiling there
is a -O3 in the CFLAGS(mozilla default optimization)
When using my optimization I use -O2
by mistake I accidentally forgot to add the -O2 while
tweaking my .mozconfig for a right CFLAGS to use for
debugging, in doing so firefox compiled all the way.

So using a custom CFLAGS in mozconfig and leaving out
-O* works..

Any ideas what might be happening here?

Justin P. Mattock
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/