[patch][gold] Fix PR11604

Cary Coutant ccoutant@google.com
Tue May 25 00:06:00 GMT 2010

>> In the test case, plugin_test_7_1.syms should be derived automatically
>> from plugin_test_7_1.o, rather than checked in as a source file.
> So, I added it as a source, because it is *not* what you will get if
> you run readelf on the .o. The idea of the test is to simulate what
> LTO can do. The .syms file corresponds to what the original file
> looked like. The .c file corresponds to what optimizations LTO can do.

Oh, right. Sorry, I glossed right over those comments.

> I could check two .c files if you prefer.

I think that would be better. I'm just worried about what would happen
if checking in raw .syms files becomes a regular practice. If the
readelf output format changes, we'd have to regenerate all the
checked-in .syms files; if we avoid checking in .syms files, we'd have
only one thing to change (parse_readelf_line in plugin_test.c).


More information about the Binutils mailing list