r18727 broke fdisk

Bernhard Fischer rep.dot.nop at gmail.com
Fri Sep 21 09:37:47 PDT 2007


On Thu, Jul 12, 2007 at 02:04:38AM +0200, Denis Vlasenko wrote:
>On Sunday 08 July 2007 10:38, Bernhard Fischer wrote:
>> vda,
>> 
>> util-linux/fdisk.c: In function 'seek_sector':
>> util-linux/fdisk.c:641: warning: implicit declaration of function 'lseek64'
>> util-linux/fdisk.c:641: error: 'off64_t' undeclared (first use in this function)
>> util-linux/fdisk.c:641: error: (Each undeclared identifier is reported only once
>> util-linux/fdisk.c:641: error: for each function it appears in.)
>> util-linux/fdisk.c:641: error: expected ')' before 'secno'
>> make[2]: *** [util-linux/fdisk.o] Error 1
>> make[1]: *** [util-linux] Error 2
>> 
>> If LFS is off, then there really is no lseek64. Please revert.
>
>I tested both. On my glibc it works.
>
>I suppose it (maybe incorrectly) provides lseek64 even without
>"#define _LARGEFILE64_SOURCE".
>
>Can you test whether fdisk.c compiles with LFS unset but with
>#define _LARGEFILE64_SOURCE
>added to the top of fdisk.c?

No, this will never help. I do not have and do not want LFS.

b at s39:/tmp/busybox$ /scratch/obj.i686/buildroot_trunk/build_i386/staging_dir/usr/bin/i386-linux-uclibc-objdump -d /scratch/obj.i686/buildroot_trunk/project_build_i386/uclibc/root/lib/libc.so.0 | grep seek64 || echo none
none
b at s39:/tmp/busybox$

I do have lseek, of course:
b at s39:/tmp/busybox$ /scratch/obj.i686/buildroot_trunk/build_i386/staging_dir/usr/bin/i386-linux-uclibc-objdump -d /scratch/obj.i686/buildroot_trunk/project_build_i386/uclibc/root/lib/libc.so.0 | grep lseek || echo none
    8830:       e8 df 08 00 00          call   9114 <__libc_lseek>
00009114 <__libc_lseek>:
    9119:       e8 00 00 00 00          call   911e <__libc_lseek+0xa>
    9141:       76 0c                   jbe    914f <__libc_lseek+0x3b>
    98b1:       e8 5e f8 ff ff          call   9114 <__libc_lseek>
    98c7:       e8 48 f8 ff ff          call   9114 <__libc_lseek>
    9907:       e8 08 f8 ff ff          call   9114 <__libc_lseek>
    b7d2:       e8 3d d9 ff ff          call   9114 <__libc_lseek>
    b99f:       e8 70 d7 ff ff          call   9114 <__libc_lseek>
   15f87:       e8 88 31 ff ff          call   9114 <__libc_lseek>
   160ad:       e8 62 30 ff ff          call   9114 <__libc_lseek>
   160d3:       e8 3c 30 ff ff          call   9114 <__libc_lseek>
   19126:       e8 e9 ff fe ff          call   9114 <__libc_lseek>

This issue is still present on trunk, r19927 (and also in 1.7.1).


More information about the busybox mailing list