[llvm-bugs] [Bug 43263] New: MIPS/C++ excessive compile time

via llvm-bugs llvm-bugs at lists.llvm.org
Tue Sep 10 07:18:37 PDT 2019


https://bugs.llvm.org/show_bug.cgi?id=43263

            Bug ID: 43263
           Summary: MIPS/C++ excessive compile time
           Product: new-bugs
           Version: unspecified
          Hardware: PC
                OS: FreeBSD
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: new bugs
          Assignee: unassignedbugs at nondot.org
          Reporter: kevans at FreeBSD.org
                CC: htmldeveloper at gmail.com, llvm-bugs at lists.llvm.org

Created attachment 22480
  --> https://bugs.llvm.org/attachment.cgi?id=22480&action=edit
Reproducer (amd64 target, mips target) and sample backtrace

Attaching a reproducer. Building on FreeBSD-CURRENT, amd64 host. Tried with
both:

$ c++ -v
FreeBSD clang version 8.0.1 (tags/RELEASE_801/final 366581) (based on LLVM
8.0.1)
Target: x86_64-unknown-freebsd13.0
Thread model: posix
InstalledDir: /usr/bin

$ clang++90 -v
clang version 9.0.0 (tags/RELEASE_900/rc2)
Target: x86_64-portbld-freebsd13.0
Thread model: posix
InstalledDir: /usr/local/llvm90/bin


Attached tar contains two repros:
- -triple x86_64-unknown-freebsd13.0
- -triple mips64-unknown-freebsd13.0


The build for the mips repro takes about 1500x longer than the build for the
amd64 repro:

$ time sh utils-mips.sh
sh utils-mips.sh   6480.70s user 9.90s system 99% cpu 1:48:59.79 total

$ time sh utils-amd64.sh
sh utils-amd64.sh  4.50s user 0.18s system 99% cpu 4.682 total

As far as I can be patient, every backtrace (one of which is also included in
the attached tar) I produce seems to be roughly the same up to visitCallSite. 

The originating file[0] is from our fusefs tests and uses googletest (headers
at [1]). 

[0]
https://svnweb.freebsd.org/base/head/tests/sys/fs/fusefs/utils.cc?view=markup
[1]
https://svnweb.freebsd.org/base/head/contrib/googletest/googletest/include/gtest/

-- 
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/20190910/30604d91/attachment-0001.html>


More information about the llvm-bugs mailing list