Hello list,
on a machine with 2 X UltraSPARC-II, doing ls -l in the openpromfs tree
shows both CPUs, but not quite correctly:
/proc/openprom# ls -l
...
88 dr-xr-xr-x 2 root root 0 May 12 11:04 SUNW,UltraSPARC-II@1c,0
88 dr-xr-xr-x 2 root root 0 May 12 11:04 SUNW,UltraSPARC-II@1c,0
...
What's wrong is obviously that there cannot be two directories with the
same name (and on top, the same inode number).
# uname -a
Linux mason 2.6.13-1.1603sp13smp #1 SMP Mon Apr 10 12:38:32 EDT 2006
sparc64 sparc64 sparc64 GNU/Linux
(Aurora Linux 2.0)
# prtconf -pv
...
Node 0xf0085030
: 00000017
: 00000011
: 000000a0
: 00000009
: 00000001
: 00000040
: 00200000
: 00000040
: 00000001
: 00000020
: 00004000
: 00000040
: 00000002
: 00000020
: 00004000
: 00000000
: 17d78400
: 000001c0.00000000.00000000.00000008
device_type: 'cpu'
name: 'SUNW,UltraSPARC-II'
Node 0xf008539c
: 00000017
: 00000011
: 000000a0
: 00000009
: 00000001
: 00000040
: 00200000
: 00000040
: 00000001
: 00000020
: 00004000
: 00000040
: 00000002
: 00000020
: 00004000
: 00000001
: 17d78400
: 000001c2.00000000.00000000.00000008
device_type: 'cpu'
name: 'SUNW,UltraSPARC-II'
...
Jan Engelhardt
--
From: Jan Engelhardt <[email protected]>
Date: Mon, 15 May 2006 17:06:24 +0200 (MEST)
> /proc/openprom# ls -l
> ...
> 88 dr-xr-xr-x 2 root root 0 May 12 11:04 SUNW,UltraSPARC-II@1c,0
> 88 dr-xr-xr-x 2 root root 0 May 12 11:04 SUNW,UltraSPARC-II@1c,0
> ...
>
> What's wrong is obviously that there cannot be two directories with the
> same name (and on top, the same inode number).
A long standing and known bug, we don't put enough addressing
information into the node name so you get duplicates.
I have no plans to work on a fix, and if anything openpromfs
deserves a rewrite as it's very old code.
Hi list,
in http://lkml.org/lkml/2006/5/15/128 I reported a problem with
openpromfs showing both CPUs under the same node name. As I looked today
into /proc/openpromfs - running 2.6.18-1.2798.al3.1smp now - this issue
is fixed.
Any details about this - can you point me to a linux-sparc archived
message or the git changeset?
Thanks,
-`J'
--