On 2011/12/5 12:39 AM, Mike Stump wrote: > On Dec 4, 2011, at 3:29 AM, Richard Sandiford <rdsandif...@googlemail.com> > wrote: >> The problem is that MIPS has >> native TLS support, but the ABI has not "yet" been extended to MIPS16. >> MIPS16 is supposed to be link-compatible with non-MIPS16, so we can't >> use emultls, and must simply say sorry(). >> >> This patch adds dg-require-profiling to the affected tests. The reason >> I haven't just applied it as obvious is that dg-require-profiling really >> seems to be a test for link-time and runtime support. There are presumably >> targets that can't link profiling code but that are nevertheless happily >> compiling the tests below. So do we want to split the directive into two? >> I ask the question while hoping the answer is "no". :-) > > Hum... I'd rather TLS support be defined and added for MIPS16... I think we > have enough targets with profiling and TLS that coverage won't be lost with > your change. I like simple. If someone feels strongly about splitting, I'll > pre-approve their change. I think your patch is fine. Ok.
We already have a MIPS16 TLS implementation internally, I'll get it ready to post here soon, though I'm afraid it's a next-stage1 kind of modification (unless Richard has the rights to approve it at this stage?). Thanks, Chung-Lin
