Broken big-endian SMP /proc/irq/prof_cpu_mask (2.6.0-mm1)?
Are you running 2.6.0-mm1 or later on a big-endian SMP box?
If so, I probably broke your file /proc/irq/prof_cpu_mask file (and /proc/irq/<pid>/smp_affinity files as well ...).
Could you cat out /proc/irq/prof_cpu_mask for me, if you have such a box, and tell me if the bytes are backward? Please also indicate your hardware architecture and number of CPUs, just so I can be sure I am on track here.
If say you have 4 CPUs, then seeing something in the range of "1" to "f" would be good news, but seeing some multiple of 1000000 would be bad news.
For further details, see the lkml thread started yesterday by [email protected]:
Subject: seperator error in __mask_snprintf_len
If I have broken this as I suspect, I will prepare a patch for Andrew shortly that fixes it.
Thanks for you assistance.
Signature · 3 lines
I won't rest till it's the best ...
Programmer, Linux Scalability
Paul Jackson <[email protected]> 1.650.933.1373