[Patch] MIPS: Factor atomic memory operations into their own .md file.
Richard Sandiford
rdsandiford@googlemail.com
Sun Aug 24 09:40:00 GMT 2008
David Daney <ddaney@avtrex.com> writes:
> This is a very mechanical patch that moves 527 lines of code from
> mips.md to a new file sync.md. I discussed this with Richard when
> working on the atomic memory operations, and finally got around to doing it.
Thanks for doing this.
> * config/mips/mips.md (fetchop_bit, immediate_insn, atomic_hiqi_op
> memory_barrier, sync_compare_and_swap<mode>,
> compare_and_swap_12, sync_add<mode>, sync_<optab><mode>,
> sync_<optab>_12, sync_old_<optab><mode>, sync_old_<optab>_12,
> sync_new_<optab><mode>, sync_new_<optab>_12, sync_nand<mode>,
> sync_nand_12, sync_old_nand<mode>, sync_old_nand_12,
> sync_new_nand<mode>, sync_new_nand_12, sync_sub<mode>,
> sync_old_add<mode>, sync_old_sub<mode>, sync_new_add<mode>,
> sync_new_sub<mode>, sync_<optab><mode>, sync_old_<optab><mode>,
> sync_new_<optab><mode>, sync_nand<mode>, sync_old_nand<mode>,
> sync_new_nand<mode>, sync_lock_test_and_set<mode>,
> sync_lock_test_and_set<mode>, test_and_set_12): Moved to sync.md.
> (sync.md): Include.
> * config/mips/sync.md: New file.
OK for trunk, but I've slight preference for keeping the immediate_insn
definition in mips.md, perhaps straight after the "insn" definition.
immediate_insn is just a mapping between rtx codes and MIPS insns,
which is the kind of thing we might need anywhere in the .md files.
Doesn't need a full retest. ;)
Richard
More information about the Gcc-patches
mailing list