------- Comment #3 from anemo at mba dot ocn dot ne dot jp 2006-07-06 16:20 ------- Subject: Re: gcc moves an expensive instruction outside of a conditional
One note: I think "rdhwr $v1, $29" should not be placed in delay slot anyway. The instruction always generate an exception, so if it was in delay slot the kernel must calculate the target address of the preceding branch/jump instruction. Is it OK to add (set_attr "can_delay" "no") for "tls_get_tp_<mode>" definition? -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=28126