mpers.sh test uses the wrong readelf

[Dmitry V. Levin]

On Fri, Nov 24, 2017 at 01:26:27PM +0100, Rolf Eike Beer wrote:
> I cross-compile from a Debian amd64 bases system to a custom amd64 system. 
> During configure I get these warnings:
> 
> checking whether mpers.sh -m32 works... readelf: Warning: skipping invalid relocation offset 0x22e in section .debug_info
> readelf: Warning: skipping invalid relocation offset 0x23a in section .debug_info
> readelf: Warning: skipping invalid relocation offset 0x247 in section .debug_info
> readelf: Warning: Invalid pointer size (0) in compunit header, using 4 instead
> readelf: Warning: CU at offset 0 contains corrupt or unsupported version number: 588.
> readelf: Warning: Invalid pointer size (82) in compunit header, using 4 instead
> readelf: Warning: Debug info is corrupted, length of CU at 5 extends beyond end of section (length = 1000002)
> cmp: EOF on mpers-m32/sample_struct.h
> no
> 
> The same happens for -mx32 later. I decided to use the target readelf and 
> things went smoother then:
> 
> sed "s,^[[:space:]]*readelf ,${target_platform}-readelf ," -i strace/mpers.sh
> 
> […]
> checking whether mpers.sh -m32 works... yes
> checking for struct stat... yes
> checking for struct stat.st_mtime_nsec... yes
> checking for struct stat64... yes
> checking for struct stat64.st_mtime_nsec... yes
> 
> I'm unsure how "correct" either behavior is, but for me it looks sensible to 
> use the target readelf on a target binary.

I've pushed a change to this effect, thanks for reporting!


-- 
ldv
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 801 bytes
Desc: not available
URL: <http://lists.strace.io/pipermail/strace-devel/attachments/20171127/6181ac8d/attachment.bin>