https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98982
Bug ID: 98982
Summary: Optimizing loop variants of fixed-byte-order functions
Product: gcc
Version: unknown
Status: UNCONFIRMED
Severity: normal
Priority: P3
Com
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98982
--- Comment #1 from Julian Andres Klode ---
Verified with gcc version 11.0.0 20210203 (experimental) [master revision
e8c87bc07b5:def4749fcba:84110515b93a6709de24240d6658ac207db5129f] (Ubuntu
11-20210203-0ubuntu1)
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98982
--- Comment #2 from Julian Andres Klode ---
clang works perfectly, fwiw. I was told it has a separate pass dedicated to
just recognizing loop idioms and optimizing them. I have no idea if gcc has
that, but it seems useful.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118537
--- Comment #8 from Julian Andres Klode ---
(In reply to Andrew Pinski from comment #6)
> If you provide the preprocessed source a lot of the time we can do the rest
> of getting a runtime Testcase and reducing it but we need the preprocessed
>
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118537
--- Comment #10 from Julian Andres Klode ---
Created attachment 60197
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=60197&action=edit
Partially minimized by cvise to 48k lines
I made it exit 42 because I could not insert an assert into t
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118537
--- Comment #14 from Julian Andres Klode ---
(In reply to Andrew Pinski from comment #11)
> >These may be related (do q31 and v31 overlap?)
>
> Q31 and V31 are the same register just different views into it. That is q31
> is the full 128bit vie
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118537
--- Comment #15 from Julian Andres Klode ---
I have the answer(In reply to Andrew Pinski from comment #12)
> A few things to test out first:
> Is it -fsanitize=address clean?
I managed to recompile it with that option and annoyingly, that also
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118537
--- Comment #18 from Julian Andres Klode ---
(In reply to Sam James from comment #17)
> (In reply to Julian Andres Klode from comment #16)
>
> I built apt and couldn't reproduce it using this yet with tip of 14 release
> branch and trunk too.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118537
--- Comment #20 from Julian Andres Klode ---
OK sorry folks, further debugging on that hunch on d15 from the minimized code
led me to build libcrypto without assembler code and now apt works correctly;
so my guess here really is that the hand-wr
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118537
Julian Andres Klode changed:
What|Removed |Added
See Also||https://github.com/openssl/
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118537
Bug ID: 118537
Summary: [14 Regression; 20241102 -> 20241128] wrong
initialization of member variable on Graviton [related
to tree-slp-vectorize]
Product: gcc
Ve
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118537
--- Comment #1 from Julian Andres Klode ---
Created attachment 60194
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=60194&action=edit
diff between the bad and good cc1plus versions
Attaching the diff between the bad and good cc1plus versi
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118537
--- Comment #4 from Julian Andres Klode ---
I only have a huge ton of code so far, I can't minimize it easily, as it
requires actually getting to the point where the constructor is called in the
loop of the https downloader which happens _after_
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118537
--- Comment #16 from Julian Andres Klode ---
Created attachment 60200
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=60200&action=edit
Reduced to 55k
The test case is now reduced to 55k bytes from 2.9M; attaching it if someone
wants to lo
14 matches
Mail list logo
