On Tue, 22 Nov 2016 11:46:33 +0000 James Cowgill <[email protected]> wrote: > Control: tags -1 patch > > Hi, > > On 18/11/16 23:39, James Cowgill wrote: > > On 16/11/16 17:15, James Cowgill wrote: > >> Control: forwarded -1 https://sourceware.org/bugzilla/show_bug.cgi?id=20828 > >> > >> Hi, > >> > >> I've submitted the bug upstream with a reduced testcase and a bisection. > >> > >> As the bug requires -Wl,--gc-sections to occur, it may be possible to > >> workaround in mesa by recompiling without it. > > > > I have attached two patches: > > > > binutils-pr20828-v1.patch attempts to fix the underlying issue in > > binutils where local symbols appear intermixed with global symbols. It > > does this by applying another pass over the symbol table to assign all > > the local symbol indexes first, then ignoring local symbols in the > > second pass. This seems to fix my testcase and mesa, although I am not > > 100% sure of it and I would like to get some feedback from upstream as > > well since I don't really do much work with binutils. > > I think the breakage is too much to wait any longer, so please can you > apply this patch to binutils so we can start recovering from all this.
Can any other mips* porter review this? Bonus points if you comment on the upstream bug so that it can (eventually) get merged upstream. Thanks, Emilio
