[llvm-bugs] [Bug 49473] New: 800x compile time regression for AVR.cpp on sparcv9

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=49473

            Bug ID: 49473
           Summary: 800x compile time regression for AVR.cpp on sparcv9
           Product: new-bugs
           Version: trunk
          Hardware: Sun
                OS: Solaris
            Status: NEW
          Severity: normal
          Priority: P
         Component: new bugs
          Assignee: unassignedbugs at nondot.org
          Reporter: ro at gcc.gnu.org
                CC: aeubanks at google.com, htmldeveloper at gmail.com,
                    llvm-bugs at lists.llvm.org

Created attachment 24601
  --> https://bugs.llvm.org/attachment.cgi?id=24601&action=edit
-ftime report output

The compile time for clang/lib/Driver/ToolChains/AVR.cpp on Solaris/sparcv9 has
exploded recently: it grew from 4.92s to 1h 3m 50.16s, almost a factor of 800
and doubling the wall clock time of ninja -j96 on a Netra SPARC S7-2.

A reghunt identified the new PM as the culprit

commit 669ddd1e9b1226432b003dbba05b99f8e992285b
Author: Arthur Eubanks <aeubanks at google.com>
Date:   Mon Jan 25 11:00:56 2021 -0800

    Turn on the new pass manager by default

and indeed compile time is back to normal with -flegacy-pass-manager.

Unfortunately, a build with -DENABLE_EXPERIMENTAL_NEW_PASS_MANAGER=OFF fails:

FAILED:
tools/clang/tools/extra/clangd/CMakeFiles/obj.clangDaemon.dir/ClangdLSPServer.cpp.o
[...]
Did not see access type in access path!
  store i64 %Handler.coerce1, i64* %.fca.1.gep, align 8, !tbaa !372
!372 = !{!370, !6, i64 8}

I'm attaching the -ftime-report output for reference.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20210308/659279a7/attachment.html>