Thread (27 messages) 27 messages, 6 authors, 2007-10-29

Re: [PATCH/RFC] powerpc: Pass PID argument to _tlbie (WAS: Apparent kernel bug with GDB on ppc405)

flat view

From: Josh Boyer <hidden>
Date: 2007-10-29 20:15:46

On Mon, 29 Oct 2007 07:08:24 -0500
Josh Boyer [off-list ref] wrote:
On Sat, 27 Oct 2007 17:32:02 +1000
Benjamin Herrenschmidt [off-list ref] wrote:
quoted
Allright, so Matt, let me know if that fixes your problem with gdb, and
I'll clean the patch up a bit and submit it. I want to double check if
something similar may be needed for freescale booke.

Basically, the problem is that things like get_user_pages() can cause
COW operations which in turn call _tlbie() to be called for translations
that aren't in the current address space.

The current 40x and 44x implementation of _tlbie doesn't handle that
properly as it only invalidates entries in the current PID.

This is an attempt at fixing it. Untested, I just checked it builds for
arch/powerpc and arch/ppc. I also haven't looked whether the freescale
BookE _tlbie needs similar treatement (it will get passed the pid in r4
with that patch, so if it needs to do something with it, it can). Kumar,
can you have a look ?
This looks pretty good at first glance.  I'll try to test it out today,
but I'd still like to hear if it solves Matt's problem as I doubt I'll
be able to duplicate it.
Did a boot test on my ebony board with this patch included.  It seems
to be happy about things so far.  If Matt gets around to trying this
out and it works, we should probably look at getting this into 2.6.24.

Oh, and I'd need a signed-off-by for it Ben :)

josh
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help