[Bug rtl-optimization/69570] [6 Regression] if-conversion bug on i?86

[tom at compton dot nu]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69570

--- Comment #11 from Tom Hughes  ---
This is C++ so -fexcess-precision=standard is no help as that is C only.

Likewise -ffloat-store is, as I understand it, not much help in real world code
because you need to make sure that you force stores in order to trigger it?

Using -mpc64 seems very scary as I believe it alters the global state of the
program.

So short of -mfpmath=sse I suspect the only solution is to replace the equality
with a comparisin that allow some variation.

I'll just slink off and go back to hating x87 FP math I think...

[Bug rtl-optimization/69570] [6 Regression] if-conversion bug on i?86

[tom at compton dot nu]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69570

--- Comment #14 from Tom Hughes  ---
Yes upstream took my fix to avoid the equality
(https://github.com/mapnik/node-mapnik/pull/589) but have also now noticed that
most of the FP can be one away with completely.

[Bug target/79197] [5/6 Regression] ICE in extract_insn in gcc/recog.c:2311

[tom at compton dot nu]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79197

Tom Hughes  changed:

   What|Removed |Added

 CC||tom at compton dot nu

--- Comment #14 from Tom Hughes  ---
That build worked because the tests were disabled on ppc64le. The first attempt
(https://koji.fedoraproject.org/koji/taskinfo?taskID=17540752) is the one which
failed.

It looks like the test crashed running a test called "Testing hard unsigned
long -> float conversions" which sounded like it might be related to this.

[Bug tree-optimization/105545] [12/13 Regression] Warning for string assignment with _GLIBCXX_ASSERTIONS since r12-3347-g8af8abfbbace49e6

[tom at compton dot nu via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105545

--- Comment #4 from Tom Hughes  ---
You don't need -D_GLIBCXX_ASSERTIONS in C++20 mode but you do in C++17 mode it
seems.

[Bug tree-optimization/105545] [12/13 Regression] Warning for string assignment with _GLIBCXX_ASSERTIONS since r12-3347-g8af8abfbbace49e6

[tom at compton dot nu via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105545

--- Comment #5 from Tom Hughes  ---
On top of -O1 you seem to need all of -fexpensive-optimizations -ftree-vrp
-fipa-sra to trigger it.

[Bug tree-optimization/105545] [12/13 Regression] Warning for string assignment with _GLIBCXX_ASSERTIONS since r12-3347-g8af8abfbbace49e6

[tom at compton dot nu via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105545

--- Comment #6 from Tom Hughes  ---
The reason it only happens with -D_GLIBCXX_ASSERTIONS or in C++20 mode is that
both of those stop the use of the explicit instantiations for basic_string and
cause them to be implicitly instantiated.