This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
Other format: | [Raw text] |
Joern RENNECKE wrote:
There is another drawback to doing this late inheritance in emit_input_reload_insns: Since we've alread started emitting reloads, it is hard to cancel any reloads that beome unnecessary - even if we don't want to do this right now, we might want to later. I think we should factor this code out into a separate function that is called right after merge_assigned_reloads. Should we call this reconsider_reload_regs ?
Maybe it's something reload_cse will do for us anyway? We might want to experiment with deleting that code entirely.
If you like, you could try your original patch on the reload-branch, where it shouldn't run into this kind of problem.
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |