Added new patch set.
https://chromiumcodereview.appspot.com/10443052/diff/6001/src/arm/lithium-arm.cc File src/arm/lithium-arm.cc (right): https://chromiumcodereview.appspot.com/10443052/diff/6001/src/arm/lithium-arm.cc#newcode2115 src/arm/lithium-arm.cc:2115: : UseTempRegister(instr->object()); On 2012/05/29 15:09:26, Vyacheslav Egorov (Google) wrote:
I don't think it has to be TempRegister. WriteBarrier does not destroy
object
register anymore.
Yes it does, the third of the three write-barriers uses "object" as a scratch register after it set the value in the properties backing store. https://chromiumcodereview.appspot.com/10443052/diff/6001/src/x64/lithium-x64.cc File src/x64/lithium-x64.cc (right): https://chromiumcodereview.appspot.com/10443052/diff/6001/src/x64/lithium-x64.cc#newcode2046 src/x64/lithium-x64.cc:2046: : UseTempRegister(instr->object()); On 2012/05/29 15:09:26, Vyacheslav Egorov (Google) wrote:
I don't think it has to be temp anymore. RecordWriteField should not
clobber it. See response to previous comment. https://chromiumcodereview.appspot.com/10443052/diff/6001/src/x64/lithium-x64.cc#newcode2059 src/x64/lithium-x64.cc:2059: LOperand* temp = (!instr->is_in_object() || needs_write_barrier || On 2012/05/29 15:09:26, Vyacheslav Egorov (Google) wrote:
I wonder if we can use designated scratch kScratchRegister to minimize
number of
allocated temps?
Done. https://chromiumcodereview.appspot.com/10443052/ -- v8-dev mailing list [email protected] http://groups.google.com/group/v8-dev
