This is the mail archive of the gcc-bugs@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[Bug target/70232] [6 regression] excessive stack usage with -O2


https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70232

--- Comment #14 from Jeffrey A. Law <law at redhat dot com> ---
WRT c#8, the FSM threader is much better at finding jump threading
opportunities which involve multiple blocks in the thread path.  I'm still not
entirely happy with the heuristics around that.

Closely related, the FSM threader is particularly poor at identifying jump
threads which use identical paths (starting from different incoming edges to
the path).  The FSM threader makes a unique copy of the path for each jump
thread.  THe old threader will create one jump thread path for that case and
redirect all appropriate edges to that common path -- this can significantly
reduce unnecessary block copying.  I suspect (but have not confirmed) that
limitation of the FSM threader is playing a role here.  Addressing that is near
the top of the gcc-7 queue.

ANyway, I'll try to take a deeper look tomorrow.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]