> Had the discussion above been included in the patch I probably would > have just acked it then :-) Now that I understand what you're doing, > it's fine.
I see, thank you ;). Pan -----Original Message----- From: Jeff Law <jeffreya...@gmail.com> Sent: Tuesday, November 19, 2024 10:57 PM To: Li, Pan2 <pan2...@intel.com>; gcc-patches@gcc.gnu.org Cc: juzhe.zh...@rivai.ai; kito.ch...@gmail.com; rdapp....@gmail.com Subject: Re: [PATCH v1 1/2] RISC-V: Fix incorrect optimization options passing to strided ld/st test On 11/19/24 7:19 AM, Li, Pan2 wrote: > Thanks Jeff for comments. > >> So if this change is the right one to make for the strided subdirectory, >> then shouldn't it also be correct to make for the gather-scatter >> subdirectory as well? > >> And similarly for various other instances where we call dg-runtest in >> that file. > > Yes, all "" "$op" of rvv.exp need to change to "$op" "" if we would like to > test sorts of rvv option combinations. > >> Basically I'd like to see some explanation why this is the right patch >> to make and why this case needs to be handled different from every other >> one that I see in that file. Assuming that explanation makes sense, >> then some kind of comment i this file indicating why this case is >> different seems in order. > > But if we make all those changes together, there will be lots of increased > failure cases. > Thus, I prefer to fix it one by one (like strided, then gather ... etc), to > make sure the patch > could be friendly for review, as well as avoid any new failures of rvv.exp > anonymously. > > Is there any best practice for such kind of changes ? I think adjusting them one by one or in sensible groupings is fine. Had the discussion above been included in the patch I probably would have just acked it then :-) Now that I understand what you're doing, it's fine. Jeff
