[Bug bootstrap/56227] New: Bootstrap failure on MinGW building ggc-page.c

[Bug bootstrap/56227] New: Bootstrap failure on MinGW building ggc-page.c

[craig.powers at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56227

             Bug #: 56227
           Summary: Bootstrap failure on MinGW building ggc-page.c
    Classification: Unclassified
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: bootstrap
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: craig.powers@gmail.com


Bootstrap is failing in stage 3 with the following error:
../../gcc-4.8-20130203/gcc/ggc-page.c: In function 'void
ggc_print_statistics()'
:
../../gcc-4.8-20130203/gcc/ggc-page.c:2174:31: error: unknown conversion type
ch
aracter 'l' in format [-Werror=format=]
         G.stats.total_overhead);
                               ^
../../gcc-4.8-20130203/gcc/ggc-page.c:2174:31: error: too many arguments for
for
mat [-Werror=format-extra-args]

(and several more identical errors on the subsequent lines)


File was attempted to compile as:
/mingw/src/gcc-build/./prev-gcc/xg++ -B/mingw/src/gcc-build/./prev-gcc/
-B/usr/l
ocal/i686-pc-mingw32/bin/ -nostdinc++
-B/mingw/src/gcc-build/prev-i686-pc-mingw3
2/libstdc++-v3/src/.libs
-B/mingw/src/gcc-build/prev-i686-pc-mingw32/libstdc++-v
3/libsupc++/.libs
-I/mingw/src/gcc-build/prev-i686-pc-mingw32/libstdc++-v3/inclu
de/i686-pc-mingw32
-I/mingw/src/gcc-build/prev-i686-pc-mingw32/libstdc++-v3/incl
ude -I/mingw/src/gcc-4.8-20130203/libstdc++-v3/libsupc++
-L/mingw/src/gcc-build/
prev-i686-pc-mingw32/libstdc++-v3/src/.libs
-L/mingw/src/gcc-build/prev-i686-pc-
mingw32/libstdc++-v3/libsupc++/.libs -c   -g -O2 -D__USE_MINGW_ACCESS
-Wno-pedan
tic-ms-format -gtoggle -DIN_GCC   -fno-exceptions -fno-rtti
-fasynchronous-unwin
d-tables -W -Wall -Wno-narrowing -Wwrite-strings -Wcast-qual
-Wmissing-format-at
tribute -pedantic -Wno-long-long -Wno-variadic-macros -Wno-overlength-strings
-W
error   -DHAVE_CONFIG_H -I. -I. -I../../gcc-4.8-20130203/gcc
-I../../gcc-4.8-201
30203/gcc/. -I../../gcc-4.8-20130203/gcc/../include
-I../../gcc-4.8-20130203/gcc
/../libcpp/include -I/mingw/src/gcc-build/./gmp
-I/mingw/src/gcc-4.8-20130203/gm
p -I/mingw/src/gcc-build/./mpfr -I/mingw/src/gcc-4.8-20130203/mpfr
-I/mingw/src/
gcc-4.8-20130203/mpc/src  -I../../gcc-4.8-20130203/gcc/../libdecnumber
-I../../g
cc-4.8-20130203/gcc/../libdecnumber/bid -I../libdecnumber
-I../../gcc-4.8-201302
03/gcc/../libbacktrace    ../../gcc-4.8-20130203/gcc/ggc-page.c -o ggc-page.o

Operating system is Windows 7 / x64.  System compiler is gcc 4.7.2, target
mingw32.

prev-gcc/xg++ version is,
$ ./prev-gcc/xg++ -v
Using built-in specs.
COLLECT_GCC=C:\MinGW\src\gcc-build\prev-gcc\xg++.exe
Target: i686-pc-mingw32
Configured with: ../gcc-4.8-20130203/configure --enable-languages=c,fortran,c++
--enable-checking=release
Thread model: win32
gcc version 4.8.0 20130203 (experimental) (GCC)

[Bug bootstrap/56227] Bootstrap failure on MinGW building ggc-page.c

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56227

--- Comment #1 from Uros Bizjak <ubizjak at gmail dot com> 2013-02-06 16:54:49 UTC ---
Created attachment 29374
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=29374
proposed patch to use HOST_LONG_LONG_FORMAT

Please try to bootstrap with the attached patch.

[Bug bootstrap/56227] Bootstrap failure on MinGW building ggc-page.c

[craig.powers at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56227

--- Comment #2 from Craig Powers <craig.powers at gmail dot com> 2013-02-06 18:42:58 UTC ---
(In reply to comment #1)
> Created attachment 29374 [details]
> proposed patch to use HOST_LONG_LONG_FORMAT
> 
> Please try to bootstrap with the attached patch.

That fixes the error I had, thanks!

It's still going right now, I'll add another update when it's done.

[Bug bootstrap/56227] Bootstrap failure on MinGW building ggc-page.c

[craig.powers at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56227

--- Comment #3 from Craig Powers <craig.powers at gmail dot com> 2013-02-06 18:50:34 UTC ---
There's another one on line 14622 of gcc/config/i386/i386.c.

[Bug bootstrap/56227] Bootstrap failure on MinGW building ggc-page.c

[craig.powers at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56227

--- Comment #4 from Craig Powers <craig.powers at gmail dot com> 2013-02-06 18:54:39 UTC ---
(In reply to comment #3)
> There's another one on line 14622 of gcc/config/i386/i386.c.

Doing the same substitution as in ggc-page.c fixes it.

[Bug bootstrap/56227] Bootstrap failure on MinGW building ggc-page.c

[craig.powers at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56227

--- Comment #5 from Craig Powers <craig.powers at gmail dot com> 2013-02-06 19:03:10 UTC ---
../../gcc-4.8-20130203/gcc/lto/lto.c: In function 'void
lto_resolution_read(spla
y_tree, FILE*, lto_file*)':
../../gcc-4.8-20130203/gcc/lto/lto.c:2229:33: error: unknown conversion type
cha
racter 'I' in format [-Werror=format=]
        " not in object file", id);
                                 ^

The full line is,
        internal_error ("resolution sub id " HOST_WIDE_INT_PRINT_HEX_PURE
                " not in object file", id);

Given that it's an internal error, I'm going to do a workaround so it will
still build for me.

[Bug bootstrap/56227] Bootstrap failure on MinGW building ggc-page.c

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56227

--- Comment #6 from Uros Bizjak <ubizjak at gmail dot com> 2013-02-06 19:40:33 UTC ---
(In reply to comment #5)
> ../../gcc-4.8-20130203/gcc/lto/lto.c: In function 'void
> lto_resolution_read(spla
> y_tree, FILE*, lto_file*)':
> ../../gcc-4.8-20130203/gcc/lto/lto.c:2229:33: error: unknown conversion type
> cha
> racter 'I' in format [-Werror=format=]
>         " not in object file", id);
>                                  ^
> 
> The full line is,
>         internal_error ("resolution sub id " HOST_WIDE_INT_PRINT_HEX_PURE
>                 " not in object file", id);
> 
> Given that it's an internal error, I'm going to do a workaround so it will
> still build for me.

I don't see anything wrong with the above HOST_WIDE_INT_PRINT_HEX_PURE
definition. Does this hints and some MinGW specific problem in type checking?

(I don't have access to MinGW platform, someone else should look a this
problem.)

[Bug bootstrap/56227] Bootstrap failure on MinGW building ggc-page.c

[craig.powers at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56227

--- Comment #7 from Craig Powers <craig.powers at gmail dot com> 2013-02-06 20:19:15 UTC ---
(In reply to comment #6)
> (In reply to comment #5)
> > ../../gcc-4.8-20130203/gcc/lto/lto.c: In function 'void
> > lto_resolution_read(spla
> > y_tree, FILE*, lto_file*)':
> > ../../gcc-4.8-20130203/gcc/lto/lto.c:2229:33: error: unknown conversion type
> > cha
> > racter 'I' in format [-Werror=format=]
> >         " not in object file", id);
> >                                  ^
> > 
> > The full line is,
> >         internal_error ("resolution sub id " HOST_WIDE_INT_PRINT_HEX_PURE
> >                 " not in object file", id);
> > 
> > Given that it's an internal error, I'm going to do a workaround so it will
> > still build for me.
> 
> I don't see anything wrong with the above HOST_WIDE_INT_PRINT_HEX_PURE
> definition. Does this hints and some MinGW specific problem in type checking?
> 
> (I don't have access to MinGW platform, someone else should look a this
> problem.)

I have no idea, personally.  I'm willing to try out proposed fixes for any
issues, as long as I can run it in the background and not eat up time I should
be spending getting other work done.

[Bug bootstrap/56227] Bootstrap failure on MinGW building ggc-page.c

[joseph at codesourcery dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56227

--- Comment #8 from joseph at codesourcery dot com <joseph at codesourcery dot com> 2013-02-07 01:48:56 UTC ---
On Wed, 6 Feb 2013, ubizjak at gmail dot com wrote:

> > The full line is,
> >         internal_error ("resolution sub id " HOST_WIDE_INT_PRINT_HEX_PURE
> >                 " not in object file", id);
> > 
> > Given that it's an internal error, I'm going to do a workaround so it will
> > still build for me.
> 
> I don't see anything wrong with the above HOST_WIDE_INT_PRINT_HEX_PURE
> definition. Does this hints and some MinGW specific problem in type checking?

internal_error takes GCC formats, not printf formats.  
HOST_WIDE_INT_PRINT_HEX_PURE is a printf format and may not be valid for 
direct use in GCC diagnostic functions.  You need to use a GCC format such 
as %wx here.

[Bug bootstrap/56227] Bootstrap failure on MinGW building ggc-page.c

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56227

Uros Bizjak <ubizjak at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |joseph at codesourcery dot
                   |                            |com, ktietz at gcc dot
                   |                            |gnu.org

--- Comment #9 from Uros Bizjak <ubizjak at gmail dot com> 2013-02-07 10:06:12 UTC ---
Adding some CCs.

[Bug bootstrap/56227] Bootstrap failure on MinGW building ggc-page.c

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56227

--- Comment #10 from Uros Bizjak <ubizjak at gmail dot com> 2013-02-07 10:07:43 UTC ---
(In reply to comment #8)

> > I don't see anything wrong with the above HOST_WIDE_INT_PRINT_HEX_PURE
> > definition. Does this hints and some MinGW specific problem in type checking?
> 
> internal_error takes GCC formats, not printf formats.  
> HOST_WIDE_INT_PRINT_HEX_PURE is a printf format and may not be valid for 
> direct use in GCC diagnostic functions.  You need to use a GCC format such 
> as %wx here.

Ah, sorry for not noticing this comment. I will prepare a complete patch.

[Bug bootstrap/56227] Bootstrap failure on MinGW building ggc-page.c

[ktietz at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56227

--- Comment #11 from Kai Tietz <ktietz at gcc dot gnu.org> 2013-02-07 10:15:42 UTC ---
(In reply to comment #9)
> Adding some CCs.

The two changes about using HOST_LONG_LONG_FORMAT are fine.
The use of HOST_WIDE_INT_PRINT_HEX_PURE in lto/lto.c is indeed wrong. The
internal_error function uses %wx instead (AFAICS).

[Bug bootstrap/56227] Bootstrap failure on MinGW building ggc-page.c

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56227

Uros Bizjak <ubizjak at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|                            |http://gcc.gnu.org/ml/gcc-p
                   |                            |atches/2013-02/msg00298.htm
                   |                            |l
   Target Milestone|---                         |4.7.3

--- Comment #12 from Uros Bizjak <ubizjak at gmail dot com> 2013-02-07 12:59:43 UTC ---
Patch at [1].

[1] http://gcc.gnu.org/ml/gcc-patches/2013-02/msg00298.html

[Bug bootstrap/56227] Bootstrap failure on MinGW building ggc-page.c

[uros at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56227

--- Comment #13 from uros at gcc dot gnu.org 2013-02-07 18:02:00 UTC ---
Author: uros
Date: Thu Feb  7 18:01:40 2013
New Revision: 195860

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=195860
Log:
    PR bootstrap/56227
    * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
    instead of "ll".
    * config/i386/i386.c (ix86_print_operand): Ditto.

lto/ChangeLog:

    PR bootstrap/56227
    * lto.c (lto_resolution_ready): Use %wx instead of
    HOST_WIDE_INT_PRINT_HEX_PURE in the argument to internal_error.


Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/i386/i386.c
    trunk/gcc/ggc-page.c
    trunk/gcc/lto/ChangeLog
    trunk/gcc/lto/lto.c

[Bug bootstrap/56227] Bootstrap failure on MinGW building ggc-page.c

[uros at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56227

--- Comment #14 from uros at gcc dot gnu.org 2013-02-07 19:18:40 UTC ---
Author: uros
Date: Thu Feb  7 19:18:27 2013
New Revision: 195865

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=195865
Log:
    Backport from mainline
    2013-02-07  Uros Bizjak  <ubizjak@gmail.com>

    PR bootstrap/56227
    * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
    instead of "ll".
    * config/i386/i386.c (ix86_print_operand): Ditto.

lto/ChangeLog:

    Backport from mainline
    2013-02-07  Uros Bizjak  <ubizjak@gmail.com>

    PR bootstrap/56227
    * lto.c (lto_resolution_ready): Use %wx instead of
    HOST_WIDE_INT_PRINT_HEX_PURE in the argument to internal_error.


Modified:
    branches/gcc-4_7-branch/gcc/ChangeLog
    branches/gcc-4_7-branch/gcc/config/i386/i386.c
    branches/gcc-4_7-branch/gcc/ggc-page.c
    branches/gcc-4_7-branch/gcc/lto/ChangeLog
    branches/gcc-4_7-branch/gcc/lto/lto.c

[Bug bootstrap/56227] Bootstrap failure on MinGW building ggc-page.c

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56227

Uros Bizjak <ubizjak at gmail dot com> changed:

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

--- Comment #15 from Uros Bizjak <ubizjak at gmail dot com> 2013-02-07 19:20:07 UTC ---
Fixed.

[Bug bootstrap/56227] Bootstrap failure on MinGW building ggc-page.c

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56227

--- Comment #16 from Uros Bizjak <ubizjak at gmail dot com> 2013-02-09 08:02:22 UTC ---
I didn't notice that my backport to 4.7 caused:

../../gcc-svn/branches/gcc-4_7-branch/gcc/lto/lto.c:1060:68: warning: format
‘%wx’ expects argument of type ‘unsigned int’, but argument 2 has type ‘long
unsigned int’ [-Wformat]

However, "%wx" _is_ correct in this case, and it works for 4.8 without
problems. 

Joseph, is there some additional magic that has to be added in lto/lto.c for
%wx to pass the check?

[Bug bootstrap/56227] Bootstrap failure on MinGW building ggc-page.c

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56227

--- Comment #17 from Uros Bizjak <ubizjak at gmail dot com> 2013-02-09 08:17:41 UTC ---
(In reply to comment #16)
> I didn't notice that my backport to 4.7 caused:
> 
> ../../gcc-svn/branches/gcc-4_7-branch/gcc/lto/lto.c:1060:68: warning: format
> ‘%wx’ expects argument of type ‘unsigned int’, but argument 2 has type ‘long
> unsigned int’ [-Wformat]
> 
> However, "%wx" _is_ correct in this case, and it works for 4.8 without
> problems. 
> 
> Joseph, is there some additional magic that has to be added in lto/lto.c for
> %wx to pass the check?

To answer my own question - we have this typedef in output.h:

/* This is a magic identifier which allows GCC to figure out the type
   of HOST_WIDE_INT for %wd specifier checks.  You must issue this
   typedef before using the __asm_fprintf__ format attribute.  */
typedef HOST_WIDE_INT __gcc_host_wide_int__;

I am testing following additional patch:

--cut here--
Index: lto.c
===================================================================
--- lto.c       (revision 195911)
+++ lto.c       (working copy)
@@ -25,6 +25,7 @@
 #include "toplev.h"
 #include "tree.h"
 #include "tree-flow.h"
+#include "output.h"
 #include "diagnostic-core.h"
 #include "tm.h"
 #include "cgraph.h"
--cut here--

[Bug bootstrap/56227] Bootstrap failure on MinGW building ggc-page.c

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56227

--- Comment #18 from Uros Bizjak <ubizjak at gmail dot com> 2013-02-09 09:47:51 UTC ---
Fixed again on 4.7 branch by #including "output.h" in lto.c.