2016-02-09 18:13 GMT+03:00 Bin.Cheng <amker.ch...@gmail.com>: > On Tue, Feb 9, 2016 at 2:56 PM, Ilya Enkovich <enkovich....@gmail.com> wrote: >> >> I agree it looks inconsistent. But I don't think PIC_OFFSET_TABLE_REGNUM is >> supposed to return pseudo regno. Using EBX_REG value for this macro was a >> workaround for problem of NULL pic_offset_table_rtx usage in cost >> functions. I think >> we should try to initialize pic_offset_table_rtx with some pseudo >> register in i386 >> target for cost estimation purposes and always return INVALID_REG for >> PIC_OFFSET_TABLE_REGNUM. > Hi IIya, > Could you please test a patch in this way? You are more experienced > than me here.
Not now I'm afraid. It doesn't seem safe enough for stage4 anyway. So let's get back to it on stage1. Thanks, Ilya > > Thanks, > bin >>
