User Details
- User Since
- Apr 18 2016, 7:25 AM (388 w, 4 d)
May 31 2023
May 17 2023
Gently ping
Apr 20 2023
Another implementation was merged to upsteam
Mar 27 2023
Gently ping
Mar 15 2023
Feb 28 2023
It is not correct to assume that the mere presence of a function with extra attributes changes the requirement of the module.
Gently ping
Gently ping
Feb 15 2023
- [RISCV] Prepare work to be ready for adding separate Zicsr and Zifencei extensions
- [RISCV] Proper support of extensions Zicsr and Zifencei
- Updated I extension verson
- Fixing after updating
Feb 14 2023
@reames Thank you for fixing documentation, missed this
Feb 2 2023
Gently ping
Gently ping
Gently ping
Could you please answer the question above?
Jan 20 2023
Maybe multi-lib handling should split into another patch?
Target features in a function must not leak outside a function.
I've created one more patch connected with lto https://reviews.llvm.org/D142191 and it also uses attributes of functions as the way to get right TargetMachine in linker.
How does this affect LTO?
Jan 19 2023
How does this affect LTO?
Abandoned in favor of D138550
Jan 18 2023
- Review fixes
- Review fixes
@foad Fixed in https://reviews.llvm.org/D142003
Gently ping
- Updated I extension verson
Jan 17 2023
Thank you for comments!
Jan 16 2023
Thanks! Already look at the problem and waiting just local build for fix
Jan 13 2023
- Review fixes
We are interested in proper support for Zicsr and Zifencei. Could someone look at patch or explain why these extensions were included by default?
Jan 12 2023
- Review fixes
- Review fixes
Dec 18 2022
Test on bge
Dec 16 2022
Gently ping
Gently ping
During fixing this case I had a question why RISCV doesn't use SimplifySetCC from common TargetLowering. It seems that there are some cases of combinations that should work for RISCV. I tried to call it, but found cases that crashes compilations on RISCV. Are there any plans to rework this part and reuse some part of foldings from common lowering? It seems that some optimizations can be lost, doesn't it?
Dec 14 2022
@lewis-revill could you please provide the case where the LTO linker would not receive any information about target features? We met similar problems, it appears on runtime functions in our cases. I made a small fix for this problem https://reviews.llvm.org/D139704. Do you have any other problem cases?
Dec 9 2022
Dec 8 2022
- Review fixes
Updated patch, thank you @kito-cheng
- Ignore unknown extensions
I don't know how I missed this patch. @kito-cheng thanks you, if you aren't against I'll try merge these 2 patches.
Dec 7 2022
I haven't found any reasons written in comments or docs why llvm-objdump couldn't take the attributes from elf file and decode all instructions without extra flags. Could you please have a look is this change can be valid?
Nov 22 2022
- Review fixes
Nov 21 2022
- Review fixes
- Review fixes
- Added test for lto case
Nov 17 2022
Gently ping
Nov 7 2022
@MaskRay thank you for review. I fixed all comments, could you have a look again?
- Review fixes
Nov 6 2022
Oct 31 2022
- Fix merge rules for attributes based on documentation
Gently ping
Oct 20 2022
Oct 17 2022
Gently ping
Gently ping
Gently ping
Sep 20 2022
Sep 19 2022
Gently ping
Gently ping
Gently ping
Aug 29 2022
- Fix format
Please, could someone has a look? Now I just want to add flag to use the current implementation of pass.
- Update after abandoned D127727
Sorry for delay. Looked more on different benchmarks from test-suite during searching a good test case. There are such cases. But a deep exploration shows that SeparateConstOffsetFromGEP pass isn't the main reason, it produces better IR, but in some cases later passes can make it worse and cause worse asssembly code. So hacks I have made in the current pass as workarounds for these particular cases don't seem to be the proper decision. As far as this pass isn't the main reason of regressions we got, I decided to abandon this review.
Hello! Decided to change this reviews stack. Could you please have a look?
Fix format
Update