On 05/21/2013 07:28 AM, Torvald Riegel wrote: >> > The additional prologue/epilogue FPR backups for TXs can only be >> > avoided if the transaction is fully contained in the function body >> > (and does not use the FPRs). I call these non-escaping transactions. > That's what __transaction_atomic etc. give you. I believe we already > check whether we need to save/restore vector registers, but I guess > we're not checking for FPRs. >
No, we have bits reserved for optimizing the vector/fpr register save/restore case, but we never actually set them. We don't have enough feedback from the register allocator in order to get those bits set. r~
