trans: Always lower to frem
#33508
trans: Always lower to frem
#33508
Conversation
|
(rust_highfive has picked a reviewer for you, use r? to override) |
|
Theoretically, we check for LLVM >=3.6 in configure, so this would break people using external LLVM. In practice, I don't think there are such people who also use 32-bit MSVC, so I guess it's okay. |
|
Huh, I was under the impression we didn't do |
Long ago LLVM unfortunately didn't handle the 32-bit MSVC case of `frem` where it can't be lowered to `fmodf` because that symbol doesn't exist. That was since fixed in http://reviews.llvm.org/D12099 (landed as r246615) and was released in what appears to be LLVM 3.8. Now that we're using that branch of LLVM let's remove our own hacks and help LLVM optimize a little better by giving it knowledge about what we're doing.
alexcrichton
force-pushed
the
always-lower-frem
branch
from
d48ef24
to
96b2288
Compare
|
@bors r+ |
|
📌 Commit 96b2288 has been approved by |
…ikomatsakis trans: Always lower to `frem` Long ago LLVM unfortunately didn't handle the 32-bit MSVC case of `frem` where it can't be lowered to `fmodf` because that symbol doesn't exist. That was since fixed in http://reviews.llvm.org/D12099 (landed as r246615) and was released in what appears to be LLVM 3.8. Now that we're using that branch of LLVM let's remove our own hacks and help LLVM optimize a little better by giving it knowledge about what we're doing.
|
⌛ Testing commit 96b2288 with merge 7f8fa4b... |
|
💔 Test failed - auto-win-msvc-32-opt |
|
@bors: retry
On Fri, May 13, 2016 at 6:26 AM, bors [email protected] wrote:
|
trans: Always lower to `frem` Long ago LLVM unfortunately didn't handle the 32-bit MSVC case of `frem` where it can't be lowered to `fmodf` because that symbol doesn't exist. That was since fixed in http://reviews.llvm.org/D12099 (landed as r246615) and was released in what appears to be LLVM 3.8. Now that we're using that branch of LLVM let's remove our own hacks and help LLVM optimize a little better by giving it knowledge about what we're doing.
Long ago LLVM unfortunately didn't handle the 32-bit MSVC case of
fremwhereit can't be lowered to
fmodfbecause that symbol doesn't exist. That was sincefixed in http://reviews.llvm.org/D12099 (landed as r246615) and was released in
what appears to be LLVM 3.8. Now that we're using that branch of LLVM let's
remove our own hacks and help LLVM optimize a little better by giving it
knowledge about what we're doing.