[clang] [Cygwin] Reduced number of inline elements of CallArgList. (PR #74977)

[徐持恒 Xu Chiheng via cfe-commits]

xu-chiheng wrote:

> > I have build scripts and patches at: https://github.com/xu-chiheng/Note
> > […](#)
> > On Mon, Dec 11, 2023 at 11:25 PM Yingchi Long _**@**_._**> wrote: When bootstrapping Clang on Cygwin, there are various weird problems. What are those "weired problems" Reducing the size of CallArgList solve them all. And why reduce the size of CallArgList solve them all? Could you kindly elaborate? — Reply to this email directly, view it on GitHub <[#74977 (comment)](https://github.com/llvm/llvm-project/pull/74977#issuecomment-1850302699)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AA5RCUGAE5BNGG5HBCLW66DYI4QXJAVCNFSM6AAAAABAOLH3Y6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQNJQGMYDENRZHE . You are receiving this because you authored the thread.Message ID: _**@**_.**_>
> 
> I see, can you just paste logs(e.g. compiler warnings) here or provide a small reproducer? (e.g step-by-step instructions to reproduce the issue).
> 
> I think it will be nice if the patch really improves llvm portability, but it is hard to reproduce bugs via a set of undocumented scripts & unknown build environments.

No, currently. It was several months ago.  I don't remember the details. Basically speaking,  it seems that some fields of CallArgList or other types are overwritten, because CallArgList size exceed an unknown limit, leading to various weird errors and failed bootstrap.  This need to be invesgated, but I'm not very familiar with the code. 

This was discovered by bisect. Clang 16 can bootstrap without this patch.


https://github.com/llvm/llvm-project/pull/74977