I think we need to reserve an additional register for other uses, like the smi tag. I thought it wouldn't hurt to do it now, but I guess we shouldn't do it until we have a definition, like kSmiMaskRegister. I definitely would use R12, in that case, since it is not used in any memory operands. Removed.
On Fri, Jun 12, 2009 at 11:24 AM, <[email protected]> wrote: > LGTM > > > http://codereview.chromium.org/126043/diff/1/2 > File src/x64/register-allocator-x64-inl.h (right): > > http://codereview.chromium.org/126043/diff/1/2#newcode41 > Line 41: reg.is(kScratchRegister) || reg.is(r12); > See below for r12. > > http://codereview.chromium.org/126043/diff/1/2#newcode49 > Line 49: static int numbers[] = { > Would it help compiler optimization to make this "const int"? I don't know. I hope so. I'll do it. > > > http://codereview.chromium.org/126043/diff/1/2#newcode62 > Line 62: -1, // r12 > No need to exclude r12 (unless you reserve it for something yourself). > We currently disallow r12 as index register, but we shouldn't - it works > fine. I'll post a CL to remove the restriction. > > If anything, if we can make r12 and r13 low priority registers (last in > line to get used), it might make code a little smaller, since opcodes > using r13 (with no displacement) or r12 (with no index) as base > registers cost an extra byte. > > > http://codereview.chromium.org/126043 > -- William Hesse Software Engineer [email protected] Google Denmark ApS Frederiksborggade 20B, 1 sal 1360 København K Denmark CVR nr. 28 86 69 84 If you received this communication by mistake, please don't forward it to anyone else (it may contain confidential or privileged information), please erase all copies of it, including all attachments, and please let the sender know it went to the wrong person. Thanks. --~--~---------~--~----~------------~-------~--~----~ v8-dev mailing list [email protected] http://groups.google.com/group/v8-dev -~----------~----~----~----~------~----~------~--~---
