> I have been reading about kprobes and one thing particularly bothers me
> in the case of mmio-trace. The probe will actually service the page
> fault, therefore it should be able force do_page_fault() to return at
> the probe point. I could not figure out a way to do that.
>
> Is it possible
On Wed, 9 Jan 2008 11:41:49 +0100
Ingo Molnar <[EMAIL PROTECTED]> wrote:
> i agree. There a few practical complication on x86: the
> do_page_fault() function is currently excluded from kprobe probing,
> for recursion reasons. handle_mm_fault() can be probed OTOH - but
> that does not catch
> Probing vmalloc faults is _really_ tricky : it also implies that the
> handler (let's call it probe) connected to the probe point (marker or
> kprobe) should _never_ cause a vmalloc page fault,
That is why vmalloc_sync_all() was invented. It might make sense
to just call that on kprobe
* Ingo Molnar ([EMAIL PROTECTED]) wrote:
>
> (kprobes folks Cc:-ed)
>
> * David Miller <[EMAIL PROTECTED]> wrote:
>
> > From: Christoph Hellwig <[EMAIL PROTECTED]>
> > Date: Wed, 9 Jan 2008 08:19:45 +0100
> >
> > > On Wed, Jan 09, 2008 at 03:55:20AM +, Dave Airlie wrote:
> > > > now
* Andi Kleen <[EMAIL PROTECTED]> wrote:
> On Wed, Jan 09, 2008 at 07:18:46AM -0800, Arjan van de Ven wrote:
> > On Wed, 9 Jan 2008 03:17:37 + (GMT)
> > Dave Airlie <[EMAIL PROTECTED]> wrote:
> >
> > > So all distros with 2.6.24 kernels are useless to mmiotrace I don't
> > > see why leaving
On Wed, Jan 09, 2008 at 07:18:46AM -0800, Arjan van de Ven wrote:
> On Wed, 9 Jan 2008 03:17:37 + (GMT)
> Dave Airlie <[EMAIL PROTECTED]> wrote:
>
> > So all distros with 2.6.24 kernels are useless to mmiotrace I don't
> > see why leaving things as is until a suitable replacement mechanism
>
On Wed, 9 Jan 2008 03:17:37 + (GMT)
Dave Airlie <[EMAIL PROTECTED]> wrote:
> So all distros with 2.6.24 kernels are useless to mmiotrace I don't
> see why leaving things as is until a suitable replacement mechanism
> can be used..
you work for a distro.. surely you can convince your own
> You can set a kprobe on the x86 fault handler to do things like
> mmiotrace.
That would mean that if the kprobe faults it goes into an endless loop.
Most of do_page_fault() is not really safe for kprobes.
-Andi
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the
From: Jiri Kosina <[EMAIL PROTECTED]>
Date: Wed, 9 Jan 2008 14:19:58 +0100 (CET)
> On Tue, 8 Jan 2008, David Miller wrote:
>
> > You can set a kprobe on the x86 fault handler to do things like
> > mmiotrace.
>
> Currently, on x86, you can not, because:
>
> fastcall void __kprobes
On Tue, 8 Jan 2008, David Miller wrote:
> You can set a kprobe on the x86 fault handler to do things like
> mmiotrace.
Currently, on x86, you can not, because:
fastcall void __kprobes do_page_fault( ... );
--
Jiri Kosina
--
To unsubscribe from this list: send the line "unsubscribe
(kprobes folks Cc:-ed)
* David Miller <[EMAIL PROTECTED]> wrote:
> From: Christoph Hellwig <[EMAIL PROTECTED]>
> Date: Wed, 9 Jan 2008 08:19:45 +0100
>
> > On Wed, Jan 09, 2008 at 03:55:20AM +, Dave Airlie wrote:
> > > now because Linus said send him a patch to revert regressions rather
>That change has been in the mainline tree for nearly three months. All
>these affected parties have left it until the eve of 2.6.24 to actually
>tell us about it. This is causing me sympathy problems :(
Not true - I complained about this on Dec 3rd (attached), with the result of
not getting a
That change has been in the mainline tree for nearly three months. All
these affected parties have left it until the eve of 2.6.24 to actually
tell us about it. This is causing me sympathy problems :(
Not true - I complained about this on Dec 3rd (attached), with the result of
not getting a
(kprobes folks Cc:-ed)
* David Miller [EMAIL PROTECTED] wrote:
From: Christoph Hellwig [EMAIL PROTECTED]
Date: Wed, 9 Jan 2008 08:19:45 +0100
On Wed, Jan 09, 2008 at 03:55:20AM +, Dave Airlie wrote:
now because Linus said send him a patch to revert regressions rather than
just
On Tue, 8 Jan 2008, David Miller wrote:
You can set a kprobe on the x86 fault handler to do things like
mmiotrace.
Currently, on x86, you can not, because:
fastcall void __kprobes do_page_fault( ... );
--
Jiri Kosina
--
To unsubscribe from this list: send the line unsubscribe
From: Jiri Kosina [EMAIL PROTECTED]
Date: Wed, 9 Jan 2008 14:19:58 +0100 (CET)
On Tue, 8 Jan 2008, David Miller wrote:
You can set a kprobe on the x86 fault handler to do things like
mmiotrace.
Currently, on x86, you can not, because:
fastcall void __kprobes do_page_fault( ...
You can set a kprobe on the x86 fault handler to do things like
mmiotrace.
That would mean that if the kprobe faults it goes into an endless loop.
Most of do_page_fault() is not really safe for kprobes.
-Andi
--
To unsubscribe from this list: send the line unsubscribe linux-kernel in
the
On Wed, Jan 09, 2008 at 07:18:46AM -0800, Arjan van de Ven wrote:
On Wed, 9 Jan 2008 03:17:37 + (GMT)
Dave Airlie [EMAIL PROTECTED] wrote:
So all distros with 2.6.24 kernels are useless to mmiotrace I don't
see why leaving things as is until a suitable replacement mechanism
can be
* Andi Kleen [EMAIL PROTECTED] wrote:
On Wed, Jan 09, 2008 at 07:18:46AM -0800, Arjan van de Ven wrote:
On Wed, 9 Jan 2008 03:17:37 + (GMT)
Dave Airlie [EMAIL PROTECTED] wrote:
So all distros with 2.6.24 kernels are useless to mmiotrace I don't
see why leaving things as is
On Wed, 9 Jan 2008 03:17:37 + (GMT)
Dave Airlie [EMAIL PROTECTED] wrote:
So all distros with 2.6.24 kernels are useless to mmiotrace I don't
see why leaving things as is until a suitable replacement mechanism
can be used..
you work for a distro.. surely you can convince your own distro
* Ingo Molnar ([EMAIL PROTECTED]) wrote:
(kprobes folks Cc:-ed)
* David Miller [EMAIL PROTECTED] wrote:
From: Christoph Hellwig [EMAIL PROTECTED]
Date: Wed, 9 Jan 2008 08:19:45 +0100
On Wed, Jan 09, 2008 at 03:55:20AM +, Dave Airlie wrote:
now because Linus said send him
On Wed, 9 Jan 2008 11:41:49 +0100
Ingo Molnar [EMAIL PROTECTED] wrote:
i agree. There a few practical complication on x86: the
do_page_fault() function is currently excluded from kprobe probing,
for recursion reasons. handle_mm_fault() can be probed OTOH - but
that does not catch vmalloc()-ed
I have been reading about kprobes and one thing particularly bothers me
in the case of mmio-trace. The probe will actually service the page
fault, therefore it should be able force do_page_fault() to return at
the probe point. I could not figure out a way to do that.
Is it possible to do
From: Christoph Hellwig <[EMAIL PROTECTED]>
Date: Wed, 9 Jan 2008 08:19:45 +0100
> On Wed, Jan 09, 2008 at 03:55:20AM +, Dave Airlie wrote:
> > now because Linus said send him a patch to revert regressions rather than
> > just complain,
>
> this is not a regression by any definition. You
From: Christoph Hellwig <[EMAIL PROTECTED]>
Date: Wed, 9 Jan 2008 08:17:27 +0100
> NACK. If you want to do it you'll need a much better reason and an
> in-tree user. And if you want to redo it it should be available for
> all platforms with a consistant API.
I majorly NACK this as well, we
On Wed, Jan 09, 2008 at 03:55:20AM +, Dave Airlie wrote:
> now because Linus said send him a patch to revert regressions rather than
> just complain,
this is not a regression by any definition. You were abusing exported
symbols for out of tree junk, so you'll lose.
--
To unsubscribe from
On Wed, Jan 09, 2008 at 02:34:46AM +, Dave Airlie wrote:
>
> [This an initial RFC but I'd like to have this patch in before 2.6.24 goes
> final as it really breaks this useful feature]
>
> mmiotrace the MMIO access tracer used to reverse engineer binary blobs
> used this notifier interface
> An alternative might be to come up with something decent and target 2.6.24.x
If you want zero cache line cost the only way is to handle that using Mathieu's
inline patch infrastructure. Having a generic notifier type based on that would
be
probably a good idea.
-Andi
--
To unsubscribe from
>
> An alternative might be to come up with something decent and target 2.6.24.x
I don't see mmiotrace getting merged into a stable kernel... how do
however see it getting cleaned up for 2.6.25 now that people know how
fragile the kernel hooks for it are..
> We put the crappy code back in
On Wed, 9 Jan 2008 03:17:37 + (GMT) Dave Airlie <[EMAIL PROTECTED]> wrote:
>
> > On Wed, 9 Jan 2008 02:34:46 + (GMT) Dave Airlie <[EMAIL PROTECTED]>
> > wrote:
> >
> > >
> > > [This an initial RFC but I'd like to have this patch in before 2.6.24
> > > goes
> > > final as it really
> On Wed, 9 Jan 2008 02:34:46 + (GMT) Dave Airlie <[EMAIL PROTECTED]> wrote:
>
> >
> > [This an initial RFC but I'd like to have this patch in before 2.6.24 goes
> > final as it really breaks this useful feature]
> >
> > mmiotrace the MMIO access tracer used to reverse engineer binary
On Wed, 9 Jan 2008 02:34:46 + (GMT) Dave Airlie <[EMAIL PROTECTED]> wrote:
>
> [This an initial RFC but I'd like to have this patch in before 2.6.24 goes
> final as it really breaks this useful feature]
>
> mmiotrace the MMIO access tracer used to reverse engineer binary blobs
> used this
On Wed, Jan 09, 2008 at 02:34:46AM +, Dave Airlie wrote:
>
> [This an initial RFC but I'd like to have this patch in before 2.6.24 goes
> final as it really breaks this useful feature]
>
> mmiotrace the MMIO access tracer used to reverse engineer binary blobs
> used this notifier interface
On Wed, Jan 09, 2008 at 02:34:46AM +, Dave Airlie wrote:
[This an initial RFC but I'd like to have this patch in before 2.6.24 goes
final as it really breaks this useful feature]
mmiotrace the MMIO access tracer used to reverse engineer binary blobs
used this notifier interface and is
On Wed, 9 Jan 2008 02:34:46 + (GMT) Dave Airlie [EMAIL PROTECTED] wrote:
[This an initial RFC but I'd like to have this patch in before 2.6.24 goes
final as it really breaks this useful feature]
mmiotrace the MMIO access tracer used to reverse engineer binary blobs
used this notifier
On Wed, 9 Jan 2008 03:17:37 + (GMT) Dave Airlie [EMAIL PROTECTED] wrote:
On Wed, 9 Jan 2008 02:34:46 + (GMT) Dave Airlie [EMAIL PROTECTED]
wrote:
[This an initial RFC but I'd like to have this patch in before 2.6.24
goes
final as it really breaks this useful
An alternative might be to come up with something decent and target 2.6.24.x
I don't see mmiotrace getting merged into a stable kernel... how do
however see it getting cleaned up for 2.6.25 now that people know how
fragile the kernel hooks for it are..
We put the crappy code back in for
An alternative might be to come up with something decent and target 2.6.24.x
If you want zero cache line cost the only way is to handle that using Mathieu's
inline patch infrastructure. Having a generic notifier type based on that would
be
probably a good idea.
-Andi
--
To unsubscribe from
On Wed, Jan 09, 2008 at 02:34:46AM +, Dave Airlie wrote:
[This an initial RFC but I'd like to have this patch in before 2.6.24 goes
final as it really breaks this useful feature]
mmiotrace the MMIO access tracer used to reverse engineer binary blobs
used this notifier interface and is
From: Christoph Hellwig [EMAIL PROTECTED]
Date: Wed, 9 Jan 2008 08:19:45 +0100
On Wed, Jan 09, 2008 at 03:55:20AM +, Dave Airlie wrote:
now because Linus said send him a patch to revert regressions rather than
just complain,
this is not a regression by any definition. You were
From: Christoph Hellwig [EMAIL PROTECTED]
Date: Wed, 9 Jan 2008 08:17:27 +0100
NACK. If you want to do it you'll need a much better reason and an
in-tree user. And if you want to redo it it should be available for
all platforms with a consistant API.
I majorly NACK this as well, we don't
On Wed, Jan 09, 2008 at 03:55:20AM +, Dave Airlie wrote:
now because Linus said send him a patch to revert regressions rather than
just complain,
this is not a regression by any definition. You were abusing exported
symbols for out of tree junk, so you'll lose.
--
To unsubscribe from
42 matches
Mail list logo
