[Bug bootstrap/58572] New: [4.9 regression] make install uses -Wno-narrowing with system compiler which does not know about it

[Bug bootstrap/58572] New: [4.9 regression] make install uses -Wno-narrowing with system compiler which does not know about it

[gerald at pfeifer dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58572

            Bug ID: 58572
           Summary: [4.9 regression] make install uses -Wno-narrowing with
                    system compiler which does not know about it
           Product: gcc
           Version: 4.9.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: bootstrap
          Assignee: unassigned at gcc dot gnu.org
          Reporter: gerald at pfeifer dot com

Created attachment 30926
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=30926&action=edit
Build log for a non-parallel build and install that shows the failure

On i386-unknown-freebsd10.0, amd64-unknown-freebsd8.3, and I assume any
system that has a sufficiently old version of GCC as the system compiler
installation now fails as follows:

gmake[2]: Entering directory `/scratch/tmp/gerald/OBJ-0927-1848/gcc'
g++ -c  -DIN_GCC_FRONTEND -g -O2 -DIN_GCC   -fno-exceptions -fno-rtti
-fasynchronous-unwind-tables -W -Wall -Wno-narrowing -Wwrite-strings
-Wcast-qual -Wmissing-format-attribute -pedantic -Wno-long-long
-Wno-variadic-macros -Wno-overlength-strings -Werror -fno-common
-DHAVE_CONFIG_H -I. -Ic -I/scratch/tmp/gerald/gcc-HEAD/gcc ...[-I 
options]...
-o c/c-lang.o -MT c/c-lang.o -MMD -MP -MF c/.deps/c-lang.TPo
/scratch/tmp/gerald/gcc-HEAD/gcc/c/c-lang.c
cc1plus: error: unrecognized command line option "-Wno-narrowing"
gmake[2]: *** [c/c-lang.o] Error 1
gmake[1]: *** [install-gcc] Error 2
gmake: *** [install] Error 2

g++ here is GCC 4.2 which does not know about -Wnarrowing.

And configuring for and building stage 1 we correctly identify that
  checking whether gcc supports -Wnarrowing... no
and this flag was not used during stage 1.

This works until and including r202892 | law | 2013-09-25 15:33:34.

[Bug bootstrap/58572] [4.9 regression] make install uses -Wno-narrowing with system compiler which does not know about it

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58572

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.9.0

[Bug bootstrap/58572] [4.9 regression] make install uses -Wno-narrowing with system compiler which does not know about it

[gary at intrepid dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58572

--- Comment #1 from Gary Funck <gary at intrepid dot com> ---
We're seeing a similar failure on SUSE SLE 11.2. The installed version of gcc
is 4.3.4.

[Bug bootstrap/58572] [4.9 regression] make bootstrap-lean leads to installation failure (doing extra rebuilds and invoking system compiler)

[gerald at pfeifer dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58572

Gerald Pfeifer <gerald at pfeifer dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2013-10-20
                 CC|                            |tromey at redhat dot com
   Target Milestone|4.9.0                       |---
            Summary|[4.9 regression] make       |[4.9 regression] make
                   |install uses -Wno-narrowing |bootstrap-lean leads to
                   |with system compiler which  |installation failure (doing
                   |does not know about it      |extra rebuilds and invoking
                   |                            |system compiler)
     Ever confirmed|0                           |1

--- Comment #2 from Gerald Pfeifer <gerald at pfeifer dot com> ---
In further testing of mine I managed to reduce this to
  make bootstrap   (which works)
and
  make bootstrap-lean   (which works, but makes make install fail).


Also, binary search confirms this is caused by Tom's "automatic
dependencies" patchset.

[Bug bootstrap/58572] [4.9 regression] make bootstrap-lean leads to installation failure (doing extra rebuilds and invoking system compiler)

[gerald at pfeifer dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58572

Gerald Pfeifer <gerald at pfeifer dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.9.0

[Bug bootstrap/58572] [4.9 regression] make bootstrap-lean leads to installation failure (doing extra rebuilds and invoking system compiler)

[tromey at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58572

Tom Tromey <tromey at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |tromey at gcc dot gnu.org

--- Comment #3 from Tom Tromey <tromey at gcc dot gnu.org> ---
I was curious why make had decided to rebuild c/c-lang.o (picking
the recompile that came first for me) so I added a line to the
.c.o rule that printed $?

This yielded:

/home/tromey/Space/Trunk/Git/PFE/prev-x86_64-unknown-linux-gnu/libstdc++-v3/include/cstring
/home/tromey/Space/Trunk/Git/PFE/prev-x86_64-unknown-linux-gnu/libstdc++-v3/include/cstdlib


So I think what is going on is that the automatic tracking scheme
notices a dependency previously omitted by the Makefile.  However, due
to how bootstrap-lean works, this file is deleted before the "install"
step.

I am not sure what to do about this.
All I can think of is that bootstrap-lean probably must leave one
prev- directory around.

[Bug bootstrap/58572] [4.9 regression] make bootstrap-lean leads to installation failure (doing extra rebuilds and invoking system compiler)

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58572

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1

--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> ---
(In reply to Tom Tromey from comment #3)
> I was curious why make had decided to rebuild c/c-lang.o (picking
> the recompile that came first for me) so I added a line to the
> .c.o rule that printed $?
> 
> This yielded:
> 
> /home/tromey/Space/Trunk/Git/PFE/prev-x86_64-unknown-linux-gnu/libstdc++-v3/
> include/cstring
> /home/tromey/Space/Trunk/Git/PFE/prev-x86_64-unknown-linux-gnu/libstdc++-v3/
> include/cstdlib
> 
> 
> So I think what is going on is that the automatic tracking scheme
> notices a dependency previously omitted by the Makefile.  However, due
> to how bootstrap-lean works, this file is deleted before the "install"
> step.
> 
> I am not sure what to do about this.
> All I can think of is that bootstrap-lean probably must leave one
> prev- directory around.

I think it does so anyway during bootstrap, so it's not an issue if
it stays there for install (IMHO).

[Bug bootstrap/58572] [4.9 regression] make bootstrap-lean leads to installation failure (doing extra rebuilds and invoking system compiler)

[tromey at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58572

Tom Tromey <tromey at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |krebbel at gcc dot gnu.org

--- Comment #5 from Tom Tromey <tromey at gcc dot gnu.org> ---
*** Bug 58666 has been marked as a duplicate of this bug. ***

[Bug bootstrap/58572] [4.9 regression] make bootstrap-lean leads to installation failure (doing extra rebuilds and invoking system compiler)

[tromey at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58572

Tom Tromey <tromey at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|unassigned at gcc dot gnu.org      |tromey at gcc dot gnu.org

--- Comment #6 from Tom Tromey <tromey at gcc dot gnu.org> ---
Mine.

[Bug bootstrap/58572] [4.9 regression] make bootstrap-lean leads to installation failure (doing extra rebuilds and invoking system compiler)

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58572

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #7 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Or perhaps if the *prev* directories are missing during install, symlink them
to the latest stage ones for the duration of make install?

[Bug bootstrap/58572] [4.9 regression] make bootstrap-lean leads to installation failure (doing extra rebuilds and invoking system compiler)

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58572

--- Comment #8 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Created attachment 32288
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=32288&action=edit
gcc49-pr58572.patch

That didn't work for some reason (tried that just manually, after make
bootstrap-lean did ln -sf {,prev-}x86_64-unknown-linux-gnu, but make install
still failed when using g++ which doesn't grok -Wnarrowing and -Wno-narrowing
(used a shell script wrapper around /usr/bin/g++ that rejected those options).

But this patch worked, apparently libstdc++-v3 headers already use #pragma GCC
system_header, so this patch shouldn't make a difference warning-wise, but it
makes a difference for -MMD.

[Bug bootstrap/58572] [4.9 regression] make bootstrap-lean leads to installation failure (doing extra rebuilds and invoking system compiler)

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58572

--- Comment #9 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Author: jakub
Date: Fri Mar  7 12:58:27 2014
New Revision: 208400

URL: http://gcc.gnu.org/viewcvs?rev=208400&root=gcc&view=rev
Log:
    PR bootstrap/58572
    * Makefile.tpl (POSTSTAGE1_CXX_EXPORT): Use -isystem instead of
    -I for libstdc++-v3 includes if $(LEAN).
    * Makefile.in: Regenerated.

Modified:
    trunk/ChangeLog
    trunk/Makefile.in
    trunk/Makefile.tpl

[Bug bootstrap/58572] [4.9 regression] make bootstrap-lean leads to installation failure (doing extra rebuilds and invoking system compiler)

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58572

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #10 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Should be fixed now.

[Bug bootstrap/58572] [4.9 regression] make bootstrap-lean leads to installation failure (doing extra rebuilds and invoking system compiler)

[mario.held at de dot ibm.com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58572

--- Comment #11 from Mario Held <mario.held at de dot ibm.com> ---
Checked out revision 208400 and did a "make -j <NUM_CPUS> bootstrap-lean
1>>$OUTPUT 2>&1" on s390x (IBM System z). Success, works as expected.