On 14/08/13 17:33, Richard Henderson wrote: > On 08/14/2013 05:44 AM, Torvald Riegel wrote: >> Can we instead move the successful path of the HTM fastpath to the >> _ITM_beginTransaction asm function, and just do the fallback and retry >> code in beginend.cc? > > Certainly, but this seems like a larger re-design, and I was thinking of > a shorter-term solution. >
Before http://gcc.gnu.org/ml/gcc-patches/2013-05/msg00820.html no FPR appears to be used as GPR save slot. So it should be safe for 4.8 branch. Skimming through the EC12 compiled libitm.so disassembly all FPR uses appear to be limited to the ITM_* accessor functions. I think 4.8 should be ok as is and we perhaps could go with the patches from Andi for mainline? Bye, -Andreas-