OP_GREVLUT is needed, replacing OP_GREV, same process as for bug #755 and likely morphing that code. formal correctness proof is needed, unit tests etc. TODO list (edit this comment):
lkcl, the commit where you disabled grev's tests should probably be reverted, because grev* will be an assembler alias for grevlut* with the appropriate immediate -- we want to test that that works. https://git.libre-soc.org/?p=openpower-isa.git;a=commitdiff;h=657a78c6fb89eddf49feb70905801c683c2155f3
(In reply to Jacob Lifshay from comment #1) > lkcl, the commit where you disabled grev's tests should probably be > reverted, because grev* will be an assembler alias for grevlut* with the > appropriate immediate -- we want to test that that works. there are no assembler aliases in sv/trans/svp64.py's translator (and none are going to be added). when grevlut has been implemented it will become appropriate to rewrite the grev unit tests to explicitly use the correct assembler instruction. until then the tests remain disabled so as not to have to spend time hunting through old commits for removed code.
Hi folks, any updates on this task? I've been checking cold case of bug #836 which strives to add support for grevlut/grevluti for binutils.
(In reply to Dmitry Selyutin from comment #3) > Hi folks, any updates on this task? I've been checking cold case of bug #836 > which strives to add support for grevlut/grevluti for binutils. you're welcome to implement it, if you like. i will help.