1807652 - ProfilerThreadRegistration.h:217:38: error: declaration of ‘mozilla::profiler::ThreadRegistration::OnThreadRef::RWOnThreadWithLock mozilla::profiler::ThreadRegistration::OnThreadRef::LockedRWOnThread()’ changes meaning of ‘LockedRWOnThread’ [-fpermissive]

Status: RESOLVED FIXED

(Core :: Gecko Profiler, defect, P2)

Description

[Martin Liška]

User Agent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/108.0.0.0 Safari/537.36

Steps to reproduce:

Build Firefox with GCC 13

Actual results:

[ 109s] 0:44.92 | ^
[ 109s] 0:44.97 In file included from /home/abuild/rpmbuild/BUILD/obj/dist/include/mozilla/ProfilerThreadState.h:14,
[ 109s] 0:44.97 from /home/abuild/rpmbuild/BUILD/obj/dist/include/mozilla/ProfilerLabels.h:14,
[ 109s] 0:44.97 from /home/abuild/rpmbuild/BUILD/obj/ipc/ipdl/PDocAccessiblePlatformExtChild.cpp:8,
[ 109s] 0:44.97 from Unified_cpp_ipc_extension_other0.cpp:11:
[ 109s] 0:44.97 /home/abuild/rpmbuild/BUILD/obj/dist/include/mozilla/ProfilerThreadRegistration.h: At global scope:
[ 109s] 0:44.97 /home/abuild/rpmbuild/BUILD/obj/dist/include/mozilla/ProfilerThreadRegistration.h:217:38: error: declaration of ‘mozilla::profiler::ThreadRegistration::OnThreadRef::RWOnThreadWithLock mozilla::profiler::ThreadRegistration::OnThreadRef::LockedRWOnThread()’ changes meaning of ‘LockedRWOnThread’ [-fpermissive]
[ 109s] 0:44.97 217 | [[nodiscard]] RWOnThreadWithLock LockedRWOnThread() {
[ 109s] 0:44.97 | ^~~~~~~~~~~~~~~~
[ 109s] 0:44.97 /home/abuild/rpmbuild/BUILD/obj/dist/include/mozilla/ProfilerThreadRegistration.h:213:7: note: used here to mean ‘using mozilla::profiler::ThreadRegistration::LockedRWOnThread = class mozilla::profiler::ThreadRegistrationLockedRWOnThread’
[ 109s] 0:44.97 213 | LockedRWOnThread& mLockedRWOnThread;
[ 109s] 0:44.97 | ^~~~~~~~~~~~~~~~
[ 109s] 0:44.97 /home/abuild/rpmbuild/BUILD/obj/dist/include/mozilla/ProfilerThreadRegistration.h:63:9: note: declared here
[ 109s] 0:44.97 63 | using LockedRWOnThread = ThreadRegistrationLockedRWOnThread;
[ 109s] 0:44.97 | ^~~~~~~~~~~~~~~~
[ 109s] 0:44.97 In file included from /home/abuild/rpmbuild/BUILD/obj/dist/include/mozilla/ProfilerThreadState.h:15:

Comment 1

[BugBot [:suhaib / :marco/ :calixte]]

The Bugbug bot thinks this bug should belong to the 'Core::Gecko Profiler' component, and is moving the bug to that component. Please correct in case you think the bot is wrong.

Comment 2

[BugBot [:suhaib / :marco/ :calixte]]

The severity field is not set for this bug.
:canova, could you have a look please?

For more information, please visit auto_nag documentation.

Comment 3

[Sam James]

Attachment: build.log.xz (GCC 13)

Reproduced with GCC 13.0.0_pre20230115. Attaching my build log.

I can probably hack up a Dockerfile or similar to make it easier to get GCC 13, if required.

Comment 4

[Nazım Can Altınova [:canova][:canaltinova on phabricator]]

Attachment: Bug 1807652 - Rename some methods to not conflict with aliases r?mstange

We have some aliases with the names of LockedRWFromAnyThread and
LockedRWOnThread. We also had some methods with the same name. clang and gcc
that were before version 12 weren't complaining about that. But apparently
starting from gcc 13, it became an error. This patch renames the methods so we
don't conflict with these aliases anymore.

Comment 5

[Nazım Can Altınova [:canova][:canaltinova on phabricator]]

Thanks for the bug report and sorry for looking at this late.

I had some problems finding gcc 13, that's why I compiled my own from the master branch. I got some more errors that were not related to profiler though. That's why I'm exactly sure what's happening there. It would be great if you can verify that this patch fixes your problem.

Comment 6

[Martin Liška]

Thanks for the patch. I can confirm it fixed the reported problem. Note I see still some problems related to:
https://gcc.gnu.org/gcc-13/porting_to.html#header-dep-changes

/home/abuild/rpmbuild/BUILD/obj/dist/include/mozilla/gfx/Rect.h:328:47: error: 'int32_t' is not a member of 'std'; did you mean 'int32_t'?
/home/abuild/rpmbuild/BUILD/firefox-109.0.1/toolkit/components/telemetry/pingsender/pingsender.cpp:30:7: error: 'uint32_t' does not name a type

and there's one GCC issue that I'm reducing right now:

lto1: internal compiler error: in force_decl_die, at dwarf2out.cc:26770

Comment 7

[Nazım Can Altınova [:canova][:canaltinova on phabricator]]

Awesome, thanks for checking!
It looks like the other issues are related to Graphics and Telemetry. Would you mind filing separate bugs for them?

Comment 8

[Martin Liška]

and there's one GCC issue that I'm reducing right now:

lto1: internal compiler error: in force_decl_die, at dwarf2out.cc:26770

It's reported as:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108772

Comment 9

[Martin Liška]

It looks like the other issues are related to Graphics and Telemetry. Would you mind filing separate bugs for them?

Sure.

Comment 10

[Martin Liška]

Reported as:
https://bugzilla.mozilla.org/show_bug.cgi?id=1816407

Comment 11

[Nazım Can Altınova [:canova][:canaltinova on phabricator]]

Thanks!

Comment 12

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/8d310a8b6460
Rename some methods to not conflict with aliases r=mstange

Comment 13

[Serban Stanca [:SerbanS]]

https://hg.mozilla.org/mozilla-central/rev/8d310a8b6460