[Bug c/30832] New: [regression] --with-sysroot=foobar wrong handled

[Bug c/30832] New: [regression] --with-sysroot=foobar wrong handled

[mtrudel at gmx dot ch]

Configuring with:
/usr/local/src/gcc/configure --prefix=/home/Marco/Desktop/manual-test/gcc
--with-sysroot=/home/Marco/Desktop/compile-lin-lin/gcc-XYZXYZ-lin/sys-root
--enable-languages=c,c++ --with-gmp=/home/Marco/Desktop/compile-lin-lin/gmp-out
--with-mpfr=/home/Marco/Desktop/compile-lin-lin/mpfr-out

leads to a failing make:
/home/Marco/Desktop/manual-test/build/./gcc/xgcc
-B/home/Marco/Desktop/manual-test/build/./gcc/
-B/home/Marco/Desktop/manual-test/gcc/i686-pc-linux-gnu/bin/
-B/home/Marco/Desktop/manual-test/gcc/i686-pc-linux-gnu/lib/ -isystem
/home/Marco/Desktop/manual-test/gcc/i686-pc-linux-gnu/include -isystem
/home/Marco/Desktop/manual-test/gcc/i686-pc-linux-gnu/sys-include -g
-fkeep-inline-functions -O2  -O2 -g -O2  -DIN_GCC    -W -Wall -Wwrite-strings
-Wstrict-prototypes -Wmissing-prototypes -Wold-style-definition  -isystem
./include  -fPIC -g -DHAVE_GTHR_DEFAULT -DIN_LIBGCC2 -D__GCC_FLOAT_NOT_NEEDED  
-I. -I. -I../.././gcc -I/usr/local/src/gcc/libgcc -I/usr/local/src/gcc/libgcc/.
-I/usr/local/src/gcc/libgcc/../gcc -I/usr/local/src/gcc/libgcc/../include
-I/usr/local/src/gcc/libgcc/../libdecnumber -I../../libdecnumber -o _muldi3.o
-MT _muldi3.o -MD -MP -MF _muldi3.dep -DL_muldi3 -c
/usr/local/src/gcc/libgcc/../gcc/libgcc2.c \
          -fvisibility=hidden -DHIDE_EXPORTS
In file included from
/home/Marco/Desktop/compile-lin-lin/gcc-XYZXYZ-lin/sys-root/usr/include/errno.h:36,
                 from /usr/local/src/gcc/libgcc/../gcc/tsystem.h:96,
                 from /usr/local/src/gcc/libgcc/../gcc/libgcc2.c:33:
/home/Marco/Desktop/compile-lin-lin/gcc-XYZXYZ-lin/sys-root/usr/include/bits/errno.h:25:26:
error: linux/errno.h: No such file or directory
In file included from
/home/Marco/Desktop/compile-lin-lin/gcc-XYZXYZ-lin/sys-root/usr/include/bits/posix1_lim.h:153,
                 from
/home/Marco/Desktop/compile-lin-lin/gcc-XYZXYZ-lin/sys-root/usr/include/limits.h:145,
                 from
/home/Marco/Desktop/manual-test/build/./gcc/include/limits.h:122,
                 from
/home/Marco/Desktop/manual-test/build/./gcc/include/syslimits.h:7,
                 from
/home/Marco/Desktop/manual-test/build/./gcc/include/limits.h:11,
                 from /usr/local/src/gcc/libgcc/../gcc/tsystem.h:108,
                 from /usr/local/src/gcc/libgcc/../gcc/libgcc2.c:33:
/home/Marco/Desktop/compile-lin-lin/gcc-XYZXYZ-lin/sys-root/usr/include/bits/local_lim.h:36:26:
error: linux/limits.h: No such file or directory
make[3]: *** [_muldi3.o] Error 1
make[3]: Leaving directory
`/home/Marco/Desktop/manual-test/build/i686-pc-linux-gnu/libgcc'

In sysroot, I have glibc and glibc-devel. This worked with gcc 4.2 and with gcc
4.3 for every build, host, target combination except all linux-gnu. The most
interesting one is a build=linux host=mingw target=linux where sysroot contains
the exactly same files. This, as said already, works...

This should work at least because I use --with-sysroot and not
--with-build-sysroot.


-- 
           Summary: [regression] --with-sysroot=foobar wrong handled
           Product: gcc
           Version: 4.3.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: mtrudel at gmx dot ch
 GCC build triplet: i686-pc-linux-gnu
  GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30832

[Bug c/30832] [regression] --with-sysroot=foobar wrong handled

[schwab at suse dot de]

------- Comment #1 from schwab at suse dot de  2007-02-17 10:16 -------
Your sysroot is missing the kernel headers.  There should be a dependency of
glibc-devel on the appropriate package that contains them.


-- 

schwab at suse dot de changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30832

[Bug c/30832] [regression] --with-sysroot=foobar wrong handled

[mtrudel at gmx dot ch]

------- Comment #2 from mtrudel at gmx dot ch  2007-02-17 11:30 -------
No, for these reasons:
- It worked for 4.2
- I would agree if I used --with-build-sysroot
- It works with a cross host=mingw target=linux
- It doesn't work if I copy the missing headers to sysroot


-- 

mtrudel at gmx dot ch changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30832

[Bug c/30832] [regression] --with-sysroot=foobar wrong handled

[schwab at suse dot de]

------- Comment #3 from schwab at suse dot de  2007-02-17 12:12 -------
Please show the output of "ls
/home/Marco/Desktop/compile-lin-lin/gcc-XYZXYZ-lin/sys-root/usr/include/linux/errno.h".


-- 

schwab at suse dot de changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |WAITING


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30832

[Bug c/30832] [regression] --with-sysroot=foobar wrong handled

[mtrudel at gmx dot ch]

------- Comment #4 from mtrudel at gmx dot ch  2007-02-17 12:23 -------
Update:
It works if I copy the kernel-headers into $SYS_ROOT/usr/include instead of
$SYS_ROOT/include.
However, I really think this should not be required since it worked with gcc
4.2 and everything else is the same. So either it was a catastrophic build
error on 4.2 and miracle that the compiled gcc worked (without linux/limits.h)
or it's a regression now.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30832

[Bug c/30832] [regression] --with-sysroot=foobar wrong handled

[schwab at suse dot de]

------- Comment #5 from schwab at suse dot de  2007-02-17 14:31 -------
$SYS_ROOT/include has no significance.  The standard include directory is
/usr/include, and sysroot mirrors exactly the standard directory layout.


-- 

schwab at suse dot de changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |RESOLVED
         Resolution|                            |INVALID


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30832

[Bug bootstrap/30832] [regression] --with-sysroot=foobar wrong handled

[drow at false dot org]

------- Comment #6 from drow at gcc dot gnu dot org  2007-02-17 21:24 -------
Subject: Re:  [regression] --with-sysroot=foobar wrong handled

On Sat, Feb 17, 2007 at 12:23:51PM -0000, mtrudel at gmx dot ch wrote:
> However, I really think this should not be required since it worked with gcc
> 4.2 and everything else is the same. So either it was a catastrophic build
> error on 4.2 and miracle that the compiled gcc worked (without linux/limits.h)
> or it's a regression now.

Sounds more like the former.  4.2 may have been using your system's
headers in this case.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30832

[Bug bootstrap/30832] [regression] --with-sysroot=foobar wrong handled

[mtrudel at gmx dot ch]

------- Comment #7 from mtrudel at gmx dot ch  2007-02-19 08:25 -------
Well, it makes sense that the kernel-source is also needed. But then again, we
have four setups:
1. GCC 4.2, build=linux, host=linux, target=linux
2. GCC 4.2, build=linux, host=mingw, target=linux
3. GCC 4.3, build=linux, host=linux, target=linux
4. GCC 4.3, build=linux, host=mingw, target=linux

For all, I specify --with-sysroot but not --with-build-sysroot and put glibc
and glibc-devel into the specified sysroot. For setup 1, 2 and 4 this works.
Only setup 3 is missing "linux/limits.h". So, how probable is that we have 3
builds that work although they shouldn't and only one that does the right thing
by failing?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30832

[Bug bootstrap/30832] [regression] --with-sysroot=foobar wrong handled

[mtrudel at gmx dot ch]

------- Comment #8 from mtrudel at gmx dot ch  2007-02-20 10:42 -------
Since there seems to be an issue, I would like to reopen this bug but change
the summary to:
gcc 4.2/4.3: different --with-sysroot handling

Although I'm not sure if it actually is a bug, I think it should be attended
to.
So if there are no objections until tomorrow evening, I'll reopen it (once
more).


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30832