[llvm-dev] Where and how to report an optimisation issue that doesn't cause a crash

David Blaikie via llvm-dev llvm-dev at lists.llvm.org
Thu Oct 24 11:41:30 PDT 2019

/looks/ necessary to me. As one of the replies said: "The event_t member of
queue_ptr could point at the same memory as event_ptr."

This means that this line "queue_ptr->queue[queue_ptr->wr_idx++] =
*event_ptr;" invalidates all the loads through queue_ptr before it, and
they have to be loaded again after this (because queue_ptr->queue might
point into the same memory that queue_ptr points into, so the assignment
might've modified the things queue_ptr points to)

‪On Thu, Oct 24, 2019 at 11:23 AM ‫אלכס לופ' via llvm-dev‬‎ <
llvm-dev at lists.llvm.org> wrote:‬

> Hi,
> Sorry for disturbind all of you through this mailing list.
> I am looking for a way to report an optimization issue which I inspected
> when compiled a C code with clang using -Oz optimization option (-O2/3 also
> reproduces it).
> I posted this to SO (
> https://stackoverflow.com/questions/58407841/does-clang-misunderstand-the-const-pointer-specifier)
> but still got no satisfying answer.
> Thank you for your help,
> Alex.
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20191024/209b4e52/attachment.html>

More information about the llvm-dev mailing list