This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] Properly detect working jobserver in gcc driver.
- From: Jan Hubicka <hubicka at ucw dot cz>
- To: Richard Biener <richard dot guenther at gmail dot com>
- Cc: Jakub Jelinek <jakub at redhat dot com>, Martin Liška <mliska at suse dot cz>, Jeff Law <law at redhat dot com>, GCC Patches <gcc-patches at gcc dot gnu dot org>, Michael Matz <matz at suse dot de>
- Date: Fri, 2 Aug 2019 11:08:48 +0200
- Subject: Re: [PATCH] Properly detect working jobserver in gcc driver.
- References: <f7274b96-34fd-7baf-4c3e-84caf60861f9@suse.cz> <fa270dc0-e449-ddef-5712-2dc4dab8b6cb@suse.cz> <20190801131932.GA2726@tucnak> <089f4b9d-a29a-5031-a272-e005cb5ce78c@suse.cz> <20190801144126.GC2726@tucnak> <b974de1b-3618-6b1e-2215-5f42bb341613@suse.cz> <20190802074450.GG2726@tucnak> <cdb8b176-acf7-4530-8e94-17474f598520@suse.cz> <20190802085001.GK2726@tucnak> <CAFiYyc2Qu_n=tDHiV98Hkf3HP6jFTw7n0K2R2OVwk67s2qSsnw@mail.gmail.com>
> On Fri, Aug 2, 2019 at 10:50 AM Jakub Jelinek <jakub@redhat.com> wrote:
> >
> > On Fri, Aug 02, 2019 at 10:47:10AM +0200, Martin Liška wrote:
> > > > Can you strace if other fds are opened and not closed in the spot you had it
> > > > before? Advantage of doing it there is that it will not be done for all the
> > > > -E/-S/-c compilations when the linker is not spawned.
> > >
> > > I've used the same trick which you used and I'm attaching the output.
> > > I believe it's fine, I can't see any opened fd by GCC.
> >
> > LGTM.
>
> Btw, we discussed yesterday on the phone and the conclusion was to
> make -flto auto-detect a job-server (but not fall back to # of threads)
> and add -flto=auto to auto-detect a job-server and fall back to # of threads.
> That basically makes -flto=jobserver the default behavior which means
> we should document -flto=1 as a way to override jobserver detection.
>
> We also discussed carrying distribution-local patches to make GNU make
> expose the jobserver FD to all jobs, not just those marked to make the
> new default effective. Does anybody have an idea if there's another
> common enough make that would benefit from support? That is,
> do other "jobserver" implementations/APIs exist we could support
> or is another make used in more than 1% of build systems?
Ideal solution would be to provide a simple jobserver library for
multithreaded tools to use (GCC is not the only one - llvm, compression
programs etc) and patch make and its replacements to use it.
I think Ninja is used by some larger projects now.
Honza
>
> Richard.
>
> > > gcc/ChangeLog:
> > >
> > > 2019-08-02 Martin Liska <mliska@suse.cz>
> > >
> > > * gcc.c (driver::maybe_run_linker): Call detect_jobserver
> > > to detect working job server.
> > > (driver::detect_jobserver): Test whether jobserver
> > > is active from GCC driver. That will prevent situation where
> > > GCC is invoked from a LD plugin and the linker already uses
> > > file descriptors suggested by make. That leads to a wrong
> > > detection.
> > > * gcc.h (driver): Add detect_jobserver.
> > > * lto-wrapper.c (jobserver_active_p): Simplify sscanf by
> > > not scanning for --jobserver-auth prefix.
> > > ---
> > > gcc/gcc.c | 42 ++++++++++++++++++++++++++++++++++++++++++
> > > gcc/gcc.h | 1 +
> > > gcc/lto-wrapper.c | 2 +-
> > > 3 files changed, 44 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/gcc/gcc.c b/gcc/gcc.c
> > > index a4323eb146e..18a07426290 100644
> > > --- a/gcc/gcc.c
> > > +++ b/gcc/gcc.c
> > > @@ -8268,6 +8268,8 @@ driver::maybe_run_linker (const char *argv0) const
> > > {
> > > int tmp = execution_count;
> > >
> > > + detect_jobserver ();
> > > +
> > > if (! have_c)
> > > {
> > > #if HAVE_LTO_PLUGIN > 0
> > > @@ -8357,6 +8359,46 @@ driver::final_actions () const
> > > }
> > > }
> > >
> > > +/* Detect whether jobserver is active and working. If not drop
> > > + --jobserver-auth from MAKEFLAGS. */
> > > +
> > > +void
> > > +driver::detect_jobserver () const
> > > +{
> > > + /* Detect jobserver and drop it if it's not working. */
> > > + const char *makeflags = env.get ("MAKEFLAGS");
> > > + if (makeflags != NULL)
> > > + {
> > > + const char *needle = "--jobserver-auth=";
> > > + const char *n = strstr (makeflags, needle);
> > > + if (n != NULL)
> > > + {
> > > + int rfd = -1;
> > > + int wfd = -1;
> > > +
> > > + bool jobserver
> > > + = (sscanf (n + strlen (needle), "%d,%d", &rfd, &wfd) == 2
> > > + && rfd > 0
> > > + && wfd > 0
> > > + && fcntl (rfd, F_GETFD) >= 0
> > > + && fcntl (wfd, F_GETFD) >= 0);
> > > +
> > > + /* Drop the jobserver if it's not working now. */
> > > + if (!jobserver)
> > > + {
> > > + unsigned offset = n - makeflags;
> > > + char *dup = xstrdup (makeflags);
> > > + dup[offset] = '\0';
> > > +
> > > + const char *space = strchr (makeflags + offset, ' ');
> > > + if (space != NULL)
> > > + strcpy (dup + offset, space);
> > > + xputenv (concat ("MAKEFLAGS=", dup, NULL));
> > > + }
> > > + }
> > > + }
> > > +}
> > > +
> > > /* Determine what the exit code of the driver should be. */
> > >
> > > int
> > > diff --git a/gcc/gcc.h b/gcc/gcc.h
> > > index a0a1d94c6e6..dc77dba67fb 100644
> > > --- a/gcc/gcc.h
> > > +++ b/gcc/gcc.h
> > > @@ -51,6 +51,7 @@ class driver
> > > void do_spec_on_infiles () const;
> > > void maybe_run_linker (const char *argv0) const;
> > > void final_actions () const;
> > > + void detect_jobserver () const;
> > > int get_exit_code () const;
> > >
> > > private:
> > > diff --git a/gcc/lto-wrapper.c b/gcc/lto-wrapper.c
> > > index 353187c6043..3414adedd26 100644
> > > --- a/gcc/lto-wrapper.c
> > > +++ b/gcc/lto-wrapper.c
> > > @@ -1234,7 +1234,7 @@ jobserver_active_p (void)
> > > int rfd = -1;
> > > int wfd = -1;
> > >
> > > - return ((sscanf(n, "--jobserver-auth=%d,%d", &rfd, &wfd) == 2)
> > > + return (sscanf (n + strlen (needle), "%d,%d", &rfd, &wfd) == 2
> > > && rfd > 0
> > > && wfd > 0
> > > && fcntl (rfd, F_GETFD) >= 0
> > > --
> > > 2.22.0
> > >
> >
> >
> > Jakub