[OmniOS-discuss] Change in uname behavior between r151014 and r151020?

N. Nadine Miller nnmiller at gmail.com
Wed Mar 1 13:56:06 UTC 2017

I installed pkgsrc some time ago and attempted to upgrade this
morning. When I attempted to update the pkg tools, it gave me an error
regarding 32-bit vs 64-bit, so reflexively I checked 'uname' output to
verify I hadn't broken my recent update to r151020.

I discovered that 'uname' incorrectly reports that I have a 32-bit
version of OmniOS:
>root at jarvis:/opt# uname -a
>SunOS jarvis 5.11 omnios-bed3013 i86pc i386 i86pc

My kernel is 64-bit, per 'isainfo -k':
>root at jarvis:/opt# isainfo -k
>root at jarvis:/opt# isalist
>amd64 pentium_pro+mmx pentium_pro pentium+mmx pentium i486 i386 i86

Build version:
>root at jarvis:/opt# uname -v
>root at jarvis:/opt# cat /etc/release
>  OmniOS v11 r151020

Looking at the uname binary, it's 32-bit:
root at jarvis:/# file /sbin/uname
/sbin/uname:    ELF 32-bit LSB executable 80386 Version 1, dynamically
linked, not stripped, no debugging information available

There's another binary in /usr/lib/bash/uname but it is completely broken:
>root at jarvis:/usr/bin# /usr/lib/bash/uname
>uname: Bad entry point


More information about the OmniOS-discuss mailing list