uniq broken with Coreutils-8.22 and i18n patch from Fedora

[Igor Živković]

Can't compile kernel. The latest patch from Fedora git might fix this issue but I haven't tried it yet.

  PASYMS  arch/x86/realmode/rm/pasyms.h
*** Error in `uniq': free(): invalid next size (fast): 0x00000000010f4340 ***
======= Backtrace: =========    
/lib/libc.so.6(+0x72ddf)[0x2b2dc18abddf]
/lib/libc.so.6(+0x785ae)[0x2b2dc18b15ae]
/lib/libc.so.6(+0x79287)[0x2b2dc18b2287]
uniq[0x402d42]                  
uniq[0x4024ab]                  
/lib/libc.so.6(__libc_start_main+0xf5)[0x2b2dc185aad5]
uniq[0x40261a]                  
======= Memory map: ========    
00400000-0040a000 r-xp 00000000 08:01 3419053                            /usr/bin/uniq
00609000-0060a000 rw-p 00009000 08:01 3419053                            /usr/bin/uniq
010cd000-0110f000 rw-p 00000000 00:00 0                                  [heap]
2b2dc1617000-2b2dc1637000 r-xp 00000000 08:01 3801195                    /lib/ld-2.18.so
2b2dc1637000-2b2dc163a000 rw-p 00000000 00:00 0
2b2dc163a000-2b2dc163d000 rw-p 00000000 00:00 0
2b2dc1836000-2b2dc1837000 r--p 0001f000 08:01 3801195                    /lib/ld-2.18.so
2b2dc1837000-2b2dc1838000 rw-p 00020000 08:01 3801195                    /lib/ld-2.18.so
2b2dc1838000-2b2dc1839000 rw-p 00000000 00:00 0
2b2dc1839000-2b2dc19d9000 r-xp 00000000 08:01 3801138                    /lib/libc-2.18.so
2b2dc19d9000-2b2dc1bd8000 ---p 001a0000 08:01 3801138                    /lib/libc-2.18.so
2b2dc1bd8000-2b2dc1bdc000 r--p 0019f000 08:01 3801138                    /lib/libc-2.18.so
2b2dc1bdc000-2b2dc1bde000 rw-p 001a3000 08:01 3801138                    /lib/libc-2.18.so
2b2dc1bde000-2b2dc1be2000 rw-p 00000000 00:00 0
2b2dc1be2000-2b2dc258d000 r--p 00000000 08:01 3412246                    /usr/lib/locale/locale-archive
2b2dc258d000-2b2dc25a2000 r-xp 00000000 08:01 3410870                    /usr/lib/libgcc_s.so.1
2b2dc25a2000-2b2dc27a2000 ---p 00015000 08:01 3410870                    /usr/lib/libgcc_s.so.1
2b2dc27a2000-2b2dc27a3000 rw-p 00015000 08:01 3410870                    /usr/lib/libgcc_s.so.1
7fff7e0e9000-7fff7e10b000 rw-p 00000000 00:00 0                          [stack]
7fff7e1ff000-7fff7e200000 r-xp 00000000 00:00 0                          [vdso]
ffffffffff600000-ffffffffff601000 r-xp 00000000 00:00 0                  [vsyscall]
/bin/sh: line 1:  1475 Done                    nm arch/x86/realmode/rm/header.o arch/x86/realmode/rm/trampoline_64.o arch/x86/realmode/rm/stack.o arch/x86/realmode/rm/reb
oot.o arch/x86/realmode/rm/wakeup_asm.o arch/x86/realmode/rm/wakemain.o arch/x86/realmode/rm/video-mode.o arch/x86/realmode/rm/copy.o arch/x86/realmode/rm/bioscall.o arch
/x86/realmode/rm/regs.o arch/x86/realmode/rm/video-vga.o arch/x86/realmode/rm/video-vesa.o arch/x86/realmode/rm/video-bios.o
      1476                       | sed -n -r -e 's/^([0-9a-fA-F]+) [ABCDGRSTVW] (.+)$/pa_\2 = \2;/p'
      1477                       | sort
      1478 Aborted                 | uniq > arch/x86/realmode/rm/pasyms.h
/sources/LFS/linux-3.12.6/arch/x86/realmode/rm/Makefile:40: recipe for target 'arch/x86/realmode/rm/pasyms.h' failed
make[3]: *** [arch/x86/realmode/rm/pasyms.h] Error 134
/sources/LFS/linux-3.12.6/arch/x86/realmode/Makefile:18: recipe for target 'arch/x86/realmode/rm/realmode.bin' failed
make[2]: *** [arch/x86/realmode/rm/realmode.bin] Error 2
scripts/Makefile.build:455: recipe for target 'arch/x86/realmode' failed
make[1]: *** [arch/x86/realmode] Error 2
Makefile:795: recipe for target 'arch/x86' failed
make: *** [arch/x86] Error 2

[WilliamImm]

I didn't experience this problem when building the kernel, but yes, it does look like the i18n patch has been updated to fix this issue.

[Igor Živković]

I can confirm that the latest patch fixes the issue. I'd upload it but, embarrassingly, couldn't figure out what needs to be changed to prevent autoreconf from running.

[bdubbs@…]

Can you upload the patch here?

[Igor Živković]

Sure

[WilliamImm]

To prevent autoreconf from triggering, at line 3998 in Makefile.in, make a new line and add:

  tests/misc/sort-mb-tests.sh                   \

Run diff again, and then edit the patch. The Makefile.in diff should be at the very bottom - if it isn't, find the section for Makefile.in, and transplant that section to the very bottom of the file. That way, Makefile.in is altered last during the patch, and it makes it look like autoreconf was run.

[Igor Živković]

Thanks William, I need to write that down.

[bdubbs@…]

It seems to me that the autoreconf problem is due to an added test. It may be easier to just skip that test by removing the hunks that modifies Makefile.in and tests/local.mk. Although not needed, the patch could be made shorter by removing tests/misc/sort-mb-tests.sh, but that's only 45 lines.

[Igor Živković]

I've tested and added the new patch to the repository. Thanks again, William.

[bdubbs@…]

Fixed at revision 10423.