On Fri, 22 Dec 2017, Jakub Jelinek wrote:
> Hi!
>
> The recent change to clear TREE_SIDE_EFFECTS on STATEMENT_LIST containing
> DEBUG_BEGIN_STMTs and a single other statement without TREE_SIDE_EFFECTS on
> it breaks the C stmt expr handling. The problem is that it assumes if
> TREE_SIDE_EFFECTS is clear on a STATEMENT_LIST then that means the
> STATEMENT_LIST is empty, i.e. ({ }), which is no longer the case.
>
> Fixed by skipping over DEBUG_BEGIN_STMTs from the tail to find the last
> non-DEBUG_BEGIN_STMT stmt (if any, and if none, treat it like ({ }) ),
> and also not processing the last non-DEBUG_BEGIN_STMT for -Wunused-value
> if it is followed by DEBUG_BEGIN_STMTs.
>
> In addition to this, the patch includes a fix from my earlier patch, where
> alloc_stmt_list TREE_SIDE_EFFECTS bit is inconsistent, depending on whether
> it is reused from the cache (then TREE_SIDE_EFFECTS was clear), or newly
> allocated (then it was set).
>
> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
OK. (This looks the issue causing glibc builds with current mainline to
fall over
<https://sourceware.org/ml/libc-testresults/2017-q4/msg00538.html>.)
--
Joseph S. Myers
jos...@codesourcery.com
