Martin Liška
Martin Liška
@sdkrystian Thanks for the PR!
I isolated that deeper a bit (a single rayon thread is sufficient) for the demonstration: ``` [2024-01-15T20:26:52.574Z INFO debuginfod_rs] walking 1917 RPM files (5.6 GB) [20459] mapping 7f26248e0000..7f26248e3000 (12 KiB)...
Got it! Oh, it's smart: ``` [pid 24127] mmap(NULL, 134217728, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7f6d743ff000 [pid 24127] munmap(0x7f6d743ff000, 62918656) = 0 [pid 24127] munmap(0x7f6d7c000000, 4190208) = 0 [pid 24127]...
glibc implementation proves my theory: https://github.com/bminor/glibc/blob/master/malloc/arena.c#L404-L441
Sure, I've just created a pull request for it.
The suggested change looks fine! However, it seems the additional changes to the renaming pass will be needed in order to fix the 2 failures spotted with LLVM 19. @sdkrystian...
The current LLVM master seems promising, but we will likely need a conditional build for the existing releases. What do you think 🤔?
The CI should be happy after 16a34b273966e32843b3f193590198e2a769b1b0.
Again, thanks for the report, it's great to have somebody who runs C-Vise on Windows target which is experimental right now. I would start here: https://github.com/marxin/cvise/blob/16a34b273966e32843b3f193590198e2a769b1b0/cvise/utils/testing.py#L35-L39 where I would remove...
Yes, I granted the write access to @emaxx-google as well, so he can hopefully properly maintain the project. Speaking about the repo move, if it helps, I am open to...