]> git.karo-electronics.de Git - karo-tx-linux.git/commit
powerpc/ftrace: Do not trace restore_interrupts()
authorSteven Rostedt <rostedt@goodmis.org>
Mon, 4 Jun 2012 16:27:54 +0000 (16:27 +0000)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Mon, 16 Jul 2012 16:03:21 +0000 (09:03 -0700)
commit3f8d5752c3926dab96e00bfb5d0c8e2f015a975e
tree32ca440fda746c596e5099eefe6960904e984c64
parent9fe9217b874769f0b42bc4f8de4fc3df785e9ef3
powerpc/ftrace: Do not trace restore_interrupts()

commit 2d773aa4810d4a612d1c879faacc38594cc3f841 upstream.

As I was adding code that affects all archs, I started testing function
tracer against PPC64 and found that it currently locks up with 3.4
kernel. I figured it was due to tracing a function that shouldn't be, so
I went through the following process to bisect to find the culprit:

 cat /debug/tracing/available_filter_functions > t
 num=`wc -l t`
 sed -ne "1,${num}p" t > t1
 let num=num+1
 sed -ne "${num},$p" t > t2
 cat t1 > /debug/tracing/set_ftrace_filter
 echo function /debug/tracing/current_tracer
 <failed? bisect t1, if not bisect t2>

It finally came down to this function: restore_interrupts()

I'm not sure why this locks up the system. It just seems to prevent
scheduling from occurring. Interrupts seem to still work, as I can ping
the box. But all user processes freeze.

When restore_interrupts() is not traced, function tracing works fine.

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
arch/powerpc/kernel/irq.c