carlosgalvezp added a comment. In D116750#3240680 <https://reviews.llvm.org/D116750#3240680>, @jansvoboda11 wrote:
> In D116750#3240363 <https://reviews.llvm.org/D116750#3240363>, @carlosgalvezp > wrote: > >> Hi @jansvoboda11 ! >> >> I believe this commit is the root cause of this issue: >> https://github.com/llvm/llvm-project/issues/53161 >> >> Believe it or not, I was (un)lucky enough to run into it. I have bisected >> the repo and my tests indicate that this commit is the offender. Do you have >> any obvious idea what could be the problem? >> >> I'm suspecting this is the offender, since it's related to `#include_next` >> which is where I get the error: >> https://github.com/llvm/llvm-project/commit/8503c688d555014b88849e933bf096035a351586#diff-fb96fa41fef79b66ab53ac1d845ab467defc41c2639e9b0cd60b471b04f898cdR981 >> To summarize, clang cannot find `/usr/include` only if I include exactly 511 >> directories via `-I`. If I include 510 or 512, then it works fine. I will >> test if the same happens with smaller powers of two. I have a test script >> that I can share if you want to test it yourself. >> >> EDIT: the same happen when including exactly 255 directories. It does not >> happen for 127 and below. It does happen for 1023 and above. > > Hi, thanks for reporting this. The reproducer script would be helpful. > Looking at the code, I don't see anything obviously wrong. Sure, I attach it do this message. It assumes you run from a place from which you can run `./build/bin/clang`.F21622105: test.py <https://reviews.llvm.org/F21622105> Let me know if I can provide anything else to help! It might not be this commit in particular, perhaps the problem is in some allocator class merged earlier. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D116750/new/ https://reviews.llvm.org/D116750 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
