Re: 2024Q2: firefox-esr failing on NetBSD 10 i386

[Izumi Tsutsui <tsutsui%ceres.dti.ne.jp@localhost>]

> > https://releng.netbsd.org/bulktracker/pkg/27161360
> >
> > Is anyone with a netbsd-10 i386 system (with 2024Q2) able to build this?
> >
> > If there are no reports of success I will conclude that it is generally
> > not buildable on the branch.
> >
> > Is there anyone using NetBSD 10 i386 and trying to use firefox-esr?  On
> > one hand that seems hard to believe but this is a retrocomputing crowd!
> 
> I have no NetBSD/i386 10 build environment yet.
> I will set it up and try to build www/firefox115 tonight.

I've just tried it on NetBSD/i386 10.0 + pkgsrc-2024Q2 and
it complains as the following:

---
In file included from /s/obj.i386/pkgsrc/www/firefox115/work.i386/build/dist/include/js/AllocPolicy.h:18,
                 from /s/obj.i386/pkgsrc/www/firefox115/work.i386/build/dist/include/js/HeapAPI.h:16,
                 from /s/obj.i386/pkgsrc/www/firefox115/work.i386/build/dist/include/js/TracingAPI.h:11,
                 from /s/obj.i386/pkgsrc/www/firefox115/work.i386/build/dist/include/js/GCPolicyAPI.h:52,
                 from /s/obj.i386/pkgsrc/www/firefox115/work.i386/build/dist/include/js/RootingAPI.h:23,
                 from /s/obj.i386/pkgsrc/www/firefox115/work.i386/build/dist/include/js/CallArgs.h:73,
                 from /s/obj.i386/pkgsrc/www/firefox115/work.i386/build/dist/include/js/CallNonGenericMethod.h:12,
                 from /s/obj.i386/pkgsrc/www/firefox115/work.i386/firefox-115.12.0/js/src/NamespaceImports.h:15,
                 from /s/obj.i386/pkgsrc/www/firefox115/work.i386/firefox-115.12.0/js/src/gc/Barrier.h:12,
                 from /s/obj.i386/pkgsrc/www/firefox115/work.i386/firefox-115.12.0/js/src/gc/WeakMap.h:12,
                 from /s/obj.i386/pkgsrc/www/firefox115/work.i386/firefox-115.12.0/js/src/debugger/DebugScript.h:16,
                 from /s/obj.i386/pkgsrc/www/firefox115/work.i386/firefox-115.12.0/js/src/debugger/DebugScript.cpp:7,
                 from Unified_cpp_js_src_debugger0.cpp:2:
/s/obj.i386/pkgsrc/www/firefox115/work.i386/firefox-115.12.0/js/src/vm/MallocProvider.h: In instantiation of 'T* js::MallocProvider<Client>::new_(Args&& ...) [with T = js::DebuggerFrame::GeneratorInfo; Args = {JS::Handle<js::AbstractGen
eratorObject*>&, JS::Rooted<JSScript*>&}; Client = JSContext]':
/s/obj.i386/pkgsrc/www/firefox115/work.i386/firefox-115.12.0/js/src/vm/MallocProvider.h:242:3:   required from 'mozilla::UniquePtr<T, JS::DeletePolicy<U> > js::MallocProvider<Client>::make_unique(Args&& ...) [with T = js::DebuggerFrame:
:GeneratorInfo; Args = {JS::Handle<js::AbstractGeneratorObject*>&, JS::Rooted<JSScript*>&}; Client = JSContext]'
/s/obj.i386/pkgsrc/www/firefox115/work.i386/firefox-115.12.0/js/src/debugger/Frame.cpp:374:56:   required from here
/s/obj.i386/pkgsrc/www/firefox115/work.i386/build/dist/include/js/Utility.h:482:20: error: static assertion failed: over-aligned type is not supported by JS_DECLARE_NEW_METHODS
  482 |         alignof(T) <= alignof(max_align_t),                                 \
      |         ~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~
/s/obj.i386/pkgsrc/www/firefox115/work.i386/firefox-115.12.0/js/src/vm/MallocProvider.h:234:3: note: in expansion of macro 'JS_DECLARE_NEW_METHODS'
  234 |   JS_DECLARE_NEW_METHODS(new_, pod_malloc<uint8_t>, MOZ_ALWAYS_INLINE)
      |   ^~~~~~~~~~~~~~~~~~~~~~
---

Disabling this static assertion works around at least on
firefox 128.0 builds, though (The similar error happend since 120.0).

I was told that there was a related bug report on glibc malloc():
>> Bug 21120 - glibc malloc is incompatible with GCC 7 (needs 16 byte alignment) 
https://sourceware.org/bugzilla/show_bug.cgi?id=21120

---
Izumi Tsutsui