Date:   Fri, 10 Dec 2021 16:51:35 +0100
From:   Heiko Carstens <hca@linux.ibm.com>
To:     Miroslav Benes <mbenes@suse.cz>
Cc:     Jerome Marchand <jmarchan@redhat.com>,
        Steven Rostedt <rostedt@goodmis.org>,
        linux-s390@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] recordmcount.pl: look for jgnop instruction as well as
 bcrl on s390
Message-ID: <YbN3h7cyC01adiwr@osiris>
References: <20211210093827.1623286-1-jmarchan@redhat.com>
 <alpine.LSU.2.21.2112101054070.5704@pobox.suse.cz>
 <YbMyvJ6gz7LvoXgE@osiris>
 <alpine.LSU.2.21.2112101310080.5704@pobox.suse.cz>
 <alpine.LSU.2.21.2112101321010.5704@pobox.suse.cz>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <alpine.LSU.2.21.2112101321010.5704@pobox.suse.cz>
Precedence: bulk

On Fri, Dec 10, 2021 at 01:31:23PM +0100, Miroslav Benes wrote:
> > > Otherwise I would be surprised that this would make any difference.
> > 
> > How come? I mean, s390 does not define HAVE_C_RECORDMCOUNT which implies 
> > that recordmcount.pl is used (see scripts/Makefile.build and 
> > BUILD_C_RECORDMCOUNT definition in Makefile).
> 
> Ah, sorry, that was too quick.
> 
> I guess you want to say that recordmcount is not used at all and GCC 
> should do it. Because CONFIG_FTRACE_MCOUNT_USE_RECORDMCOUNT is not set, 
> since FTRACE_MCOUNT_USE_CC should be set.
> 
> What a maze.
> 
> But isn't -mrecord-mcount x86_64-only option?

No, it is (at least) also supported on s390. IIRC that was added with
gcc9. Anyway, this is not too important since the fix makes sense
anyway.