uname -p and uname -i reporting `unknown'

This has been discussed upstream; see:

* http://lists.gnu.org/archive/html/bug-coreutils/2005-09/msg00053.html for a more discussed thread, or
* http://lists.gnu.org/archive/html/bug-coreutils/2007-04/msg00089.html for a to-the-point one.

There are other hits, BTW.

Summary, as far as I can understand: coreutils will not provide 'uname -p -i' on Linux platforms since the 'sysinfo()' call does not support it, and /proc is er, considered too Linux-specific.

I would tend to mark this one INVALID, but I agree this should be available somewhere...

Also see bug 111863.

uh. Doko, you mention "apparently on other distributions". Do you have a reference to give me?

On 03.11.2009 18:03, C de-Avillez wrote:
> uh. Doko, you mention "apparently on other distributions". Do you have a
> reference to give me?

asked somebody to check on Fedora, which apparently sets these.

uname -i in fedora 12 gave me i386.

Thanks. I will download the sources for coreutils on Fedora 12, and see what goes on there.

I cherry-picked the uname patch from Fedora 12; tested it locally, on Karmic, Lucid, and on coreutils GIT. All seem to work without a glitch.

I am publishing coreutils 7.4-2ubuntu2 on my PPA, and attaching the debdiff for Lucid here. Nevertheless, I would really appreciate some more tests.

that ppa works fine uname -i and uname -p worked

Hi all,

Have the problem Ubuntu 9.10 :

ubuntu:/usr$ uname -i
unknown
ubuntu:/usr$

But i've found another uname installed :

ubuntu:/usr$ /usr/lib/klibc/bin/uname -i
i386
ubuntu:/usr$

This is working well :

ubuntu:/usr$ /usr/lib/klibc/bin/uname -a
Linux ubuntu 2.6.31-15-generic #50-Ubuntu SMP Tue Nov 10 14:54:29 UTC 2009 i686 i386
ubuntu:/usr$ /usr/lib/klibc/bin/uname -s
Linux
ubuntu:/usr$ /usr/lib/klibc/bin/uname -n
ubuntu
ubuntu:/usr$ /usr/lib/klibc/bin/uname -r
2.6.31-15-generic
ubuntu:/usr$ /usr/lib/klibc/bin/uname -v
#50-Ubuntu SMP Tue Nov 10 14:54:29 UTC 2009
ubuntu:/usr$ /usr/lib/klibc/bin/uname -m
i686
ubuntu:/usr$ /usr/lib/klibc/bin/uname -i
i386
ubuntu:/usr$

Can anyone explain me why two version of uname please ?
And how to use /usr/lib/klibc/bin/uname instead of /bin/uname

Txs all

i686 is your processor's and i386 is the kernel's

@linflaas:
/usr/lib/klibc/bin is provided by the libklibc package, and is for use with initramfs. This is a work-in-progress, and does *not* replace coreutils.

I really do not think you should use it in lieu of the standard utilities in coreutils.

OK txs a lot.

i then will wait for release ;)

For the record, my PPA has a GIT coreutils package for Lucid with this patch in (https://edge.launchpad.net/~hggdh2/+archive/ppa).

Milestoning: this breaks some 3rd party installer scripts.

un-assigning myself, marking as Triaged, waiting for Sponsorship.

We should have dealt with this before, but I'm afraid that it seems that at this point this is far too risky to do for Lucid. Won't-Fixing for Lucid, and we can reconsider this for Maverick.

Could somebody get that change in maverick now?

I tested this patch and it works. Waiting for ACK from ubuntu-release team.

No, I think once again it is too late for this. It should go in pre-feature freeze.

Now would be the time in the cycle to get such change in, Colin, Mathias do you think it makes sense? Could you sponsor it if that's the case?

the merge request comment suggest upstream didn't ack the patch so unsubscribing sponsors

Upstream coreutils will not ack the patch; this has been discussed over and over upstream, and their position is this is a service that should be provided by the kernel (via sysinfo() or uname(), probably).

As such we either carry this delta, or close this bug WONTFIX.

so lets add 'linux' to this bug, if it's fixed in either that or coreutils depending on the outcome.

For the record, I have added the Debian bug about this -- http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=193170.

This bug was fixed in the package coreutils - 8.5-1ubuntu5

---------------
coreutils (8.5-1ubuntu5) natty; urgency=low

  * debian/patches/80_fedora_sysinfo.dpatch: make 'uname -i -p' return the
    real processor/hardware, instead of unknown. Patch cherry-picked from
    Fedora 12 (original: coreutils-4.5.3-sysinfo.patch, from the
    coreutils-7.6-5.src.rpm). LP: #470550.
 -- C de-Avillez <email address hidden> Tue, 10 Nov 2009 12:38:24 -0600

While this might be a valid request for a change to the kernel, it will have to be come from upstream in order for us to consider it. In any case, the change wouldn't meed our requirements to be applied as an SRU to Lucid:

https://wiki.ubuntu.com/KernelTeam/KernelUpdates

Set to wontfix