Re: Build perf with clang, failure with libperf

From: Sedat Dilek
Date: Thu Apr 07 2022 - 13:18:43 EST

Next message: Rob Herring: "Re: [PATCH v2 05/14] dt-bindings: arm: mediatek: document the pcie mirror node on MT7622"
Previous message: Coly Li: "Re: [PATCH v2] md: bcache: check the return value of kzalloc() in detached_dev_do_request()"
In reply to: Nathan Chancellor: "Re: Build perf with clang, failure with libperf"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

On Thu, Apr 7, 2022 at 7:14 PM Nathan Chancellor <nathan@xxxxxxxxxx> wrote:
>
> On Thu, Apr 07, 2022 at 06:25:29PM +0200, Sedat Dilek wrote:
> > On Thu, Apr 7, 2022 at 5:03 PM Nathan Chancellor <nathan@xxxxxxxxxx> wrote:
> > >
> > > On Thu, Apr 07, 2022 at 12:27:14PM +0200, Sedat Dilek wrote:
> > > > On Mon, Apr 4, 2022 at 11:53 PM Nathan Chancellor <nathan@xxxxxxxxxx> wrote:
> > > > >
> > > > > Hi Arnaldo,
> > > > >
> > > > > On Mon, Apr 04, 2022 at 05:43:11PM -0300, Arnaldo Carvalho de Melo wrote:
> > > > > > Hi,
> > > > > >
> > > > > > Trying to apply Sedat's patch something changed in my system,
> > > > > > and that patch wasn't enough, so I had to first apply this one:
> > > > > >
> > > > > > commit 173b552663419f40bcd3cf9df4f68285cac72727
> > > > > > Author: Arnaldo Carvalho de Melo <acme@xxxxxxxxxx>
> > > > > > Date: Mon Apr 4 17:28:48 2022 -0300
> > > > > >
> > > > > > tools build: Use $(shell ) instead of `` to get embedded libperl's ccopts
> > > > > >
> > > > > > Just like its done for ldopts and for both in tools/perf/Makefile.config.
> > > > > >
> > > > > > Using `` to initialize PERL_EMBED_CCOPTS somehow precludes using:
> > > > > >
> > > > > > $(filter-out SOMETHING_TO_FILTER,$(PERL_EMBED_CCOPTS))
> > > > > >
> > > > > > And we need to do it to allow for building with versions of clang where
> > > > > > some gcc options selected by distros are not available.
> > > > > >
> > > > > > Cc: Adrian Hunter <adrian.hunter@xxxxxxxxx>
> > > > > > Cc: Fangrui Song <maskray@xxxxxxxxxx>
> > > > > > Cc: Florian Fainelli <f.fainelli@xxxxxxxxx>
> > > > > > Cc: Ian Rogers <irogers@xxxxxxxxxx>
> > > > > > Cc: Jiri Olsa <jolsa@xxxxxxxxxx>
> > > > > > Cc: John Keeping <john@xxxxxxxxxxxx>
> > > > > > Cc: Leo Yan <leo.yan@xxxxxxxxxx>
> > > > > > Cc: Michael Petlan <mpetlan@xxxxxxxxxx>
> > > > > > Cc: Namhyung Kim <namhyung@xxxxxxxxxx>
> > > > > > Cc: Nathan Chancellor <nathan@xxxxxxxxxx>
> > > > > > Cc: Nick Desaulniers <ndesaulniers@xxxxxxxxxx>
> > > > > > Cc: Sedat Dilek <sedat.dilek@xxxxxxxxx>
> > > > > > Signed-off-by: Arnaldo Carvalho de Melo <acme@xxxxxxxxxx>
> > > > > >
> > > > > > diff --git a/tools/build/feature/Makefile b/tools/build/feature/Makefile
> > > > > > index 1480910c792e2cb3..90774b60d31b2b8e 100644
> > > > > > --- a/tools/build/feature/Makefile
> > > > > > +++ b/tools/build/feature/Makefile
> > > > > > @@ -217,7 +217,7 @@ strip-libs = $(filter-out -l%,$(1))
> > > > > > PERL_EMBED_LDOPTS = $(shell perl -MExtUtils::Embed -e ldopts 2>/dev/null)
> > > > > > PERL_EMBED_LDFLAGS = $(call strip-libs,$(PERL_EMBED_LDOPTS))
> > > > > > PERL_EMBED_LIBADD = $(call grep-libs,$(PERL_EMBED_LDOPTS))
> > > > > > -PERL_EMBED_CCOPTS = `perl -MExtUtils::Embed -e ccopts 2>/dev/null`
> > > > > > +PERL_EMBED_CCOPTS = $(shell perl -MExtUtils::Embed -e ccopts 2>/dev/null)
> > > > > > FLAGS_PERL_EMBED=$(PERL_EMBED_CCOPTS) $(PERL_EMBED_LDOPTS)
> > > > > >
> > > > > > $(OUTPUT)test-libperl.bin:
> > > > > >
> > > > > > ----------------------------------------------------- 8< -------------------
> > > > > >
> > > > > > After this I go on filtering out some of the gcc options that clang
> > > > > > doesn't grok:
> > > > > >
> > > > > > diff --git a/tools/build/feature/Makefile b/tools/build/feature/Makefile
> > > > > > index 90774b60d31b2b8e..bbc5e263e02385ed 100644
> > > > > > --- a/tools/build/feature/Makefile
> > > > > > +++ b/tools/build/feature/Makefile
> > > > > > @@ -215,9 +215,12 @@ grep-libs = $(filter -l%,$(1))
> > > > > > strip-libs = $(filter-out -l%,$(1))
> > > > > >
> > > > > > PERL_EMBED_LDOPTS = $(shell perl -MExtUtils::Embed -e ldopts 2>/dev/null)
> > > > > > +PERL_EMBED_LDOPTS := $(filter-out -specs=%,$(PERL_EMBED_LDOPTS))
> > > > > > PERL_EMBED_LDFLAGS = $(call strip-libs,$(PERL_EMBED_LDOPTS))
> > > > > > PERL_EMBED_LIBADD = $(call grep-libs,$(PERL_EMBED_LDOPTS))
> > > > > > PERL_EMBED_CCOPTS = $(shell perl -MExtUtils::Embed -e ccopts 2>/dev/null)
> > > > > > +PERL_EMBED_CCOPTS := $(filter-out -ffat-lto-objects, $(PERL_EMBED_CCOPTS))
> > > > > > +PERL_EMBED_CCOPTS := $(filter-out -specs=%,$(PERL_EMBED_CCOPTS))
> > > > > > FLAGS_PERL_EMBED=$(PERL_EMBED_CCOPTS) $(PERL_EMBED_LDOPTS)
> > > > > >
> > > > > > $(OUTPUT)test-libperl.bin:
> > > > > >
> > > > > > ----------------------------------------------------- 8< -------------------
> > > > > >
> > > > > > And then get to the problems at the end of this message, which seem
> > > > > > similar to the problem described here:
> > > > > >
> > > > > > From Nathan Chancellor <>
> > > > > > Subject [PATCH] mwifiex: Remove unnecessary braces from HostCmd_SET_SEQ_NO_BSS_INFO
> > > > > >
> > > > > > https://lkml.org/lkml/2020/9/1/135
> > > > > >
> > > > > > So perhaps in this case its better to disable that
> > > > > > -Werror,-Wcompound-token-split-by-macro when building with clang?
> > > > >
> > > > > Yes, I think that is probably the best solution. As far as I can tell,
> > > > > at least in this file and context, the warning appears harmless, as the
> > > > > "create a GNU C statement expression from two different macros" is very
> > > > > much intentional, based on the presence of PERL_USE_GCC_BRACE_GROUPS.
> > > > > The warning is fixed in upstream Perl by just avoiding creating GNU C
> > > > > statement expressions using STMT_START and STMT_END:
> > > > >
> > > > > https://github.com/Perl/perl5/issues/18780
> > > > > https://github.com/Perl/perl5/pull/18984
> > > > >
> > > > > If I am reading the source code correctly, an alternative to disabling
> > > > > the warning would be specifying -DPERL_GCC_BRACE_GROUPS_FORBIDDEN but it
> > > > > seems like that might end up impacting more than just this site,
> > > > > according to the issue discussion above.
> > > > >
> > > >
> > > > Thanks for the pointer Nathan.
> > > >
> > > > As said I hit the problem with Debian's perl v5.34.
> > > >
> > > > Checking perl5 Git reveals:
> > > >
> > > > "skip using gcc brace groups for STMT_START/END"
> > > > https://github.com/Perl/perl5/commit/7169efc77525df70484a824bff4ceebd1fafc760
> > >
> > > GitHub says this is in 5.35.2, so it would make sense that 5.34 still
> > > shows the issue.
> > >
> > > > "Partially Revert "skip using gcc brace groups for STMT_START/END""
> > > > https://github.com/Perl/perl5/commit/e08ee3cb66f362c4901846a46014cfdfcd60326c
> > > >
> > > > Perl v5.34.x seems not to have these changes:
> > > > https://github.com/Perl/perl5/compare/v5.34.0...v5.34.1
> > > >
> > > > Unsure if there exists a real fix for perl5.
> > >
> > > Perhaps those two changes could be cherry-picked into Debian's 5.34. I
> > > have no idea if that is possible though.
> > >
> >
> > What perl5 version has Arch Linux?
>
> $ pacman -Q perl
> perl 5.34.1-1
>
> It appears that 5.35 is the development version, which I suppose means
> we won't see this fix widely available for some time:
>
> https://www.nntp.perl.org/group/perl.perl5.porters/2022/03/msg263388.html
>
> > Do you see the issue and need -Wno-compound-token-split-by-macro?
>
> I don't build perf but given the above information, I suspect I would
> need that flag as well.
>

Can you be so kind and apply my 2 diffs to perl.h and acme patchset
(or Git tree) and test?

- Sedat -

[1] https://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git/log/?h=tmp.perf/urgent

Next message: Rob Herring: "Re: [PATCH v2 05/14] dt-bindings: arm: mediatek: document the pcie mirror node on MT7622"
Previous message: Coly Li: "Re: [PATCH v2] md: bcache: check the return value of kzalloc() in detached_dev_do_request()"
In reply to: Nathan Chancellor: "Re: Build perf with clang, failure with libperf"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]