Larry Finger wrote:
merely triggered by some interaction with ACPI and/or the BIOS.
From what I found in looking back through the DMA error reports,
most (if not all) people with the problem have netbook computers
with Intel ATOM processors.
Gábor Stefanik wrote:
Linus has also reported this
2009/11/14 Matthew Garrett mj...@srcf.ucam.org:
On Sat, Nov 14, 2009 at 12:41:49AM +, Andrew Benton wrote:
And it seems to be working well. No errors so far. I've just downloaded
a kernel, browsed slashdot a bit. I'll test it some more tomorrow but
this is a BIG step in the right
On 14/11/09 07:29, Matthew Garrett wrote:
I should emphasise that this patch works by effectively disabling deep C
states on your CPU, which in turn will increase your power consumption.
It's very much either a workaround for broken hardware or something that
covers up a more subtle bug
On Sat, 14 Nov 2009 09:52:15 +
Andrew Benton b3n...@gmail.com wrote:
It was working fine last night but I can't get that kernel to connect
today. Nothing has changed, I can see no reason why it was working
and isn't working now. It feels like a hardware problem.
I'm recompiling with a
On 14/11/09 11:24, Chris Vine wrote:
Be aware that if you have been using the proprietary wl driver to send
on your bug reports, you must do a cold boot before testing b43, as if
you warm boot after having initialised the wireless device with the wl
driver then the DMA bug disappears.
That
Larry Finger wrote:
On 11/13/2009 06:15 PM, William Bourque wrote:
Larry Finger wrote:
Based on a suggestion by Matthew Garrett, please try the patch below.
Thanks,
Larry
=
Index: wireless-testing/drivers/net/wireless/b43/main.c
On 11/14/2009 12:51 PM, William Bourque wrote:
Ok, I tried with value of 150, 100 already and I'm recompiling to see
the result with 175 and 125.
150us seems to give me the best result. As with 200 I could use the
wireless for several minutes and insert/remove the module. However it
On Friday 13 November 2009 01:02:44 Larry Finger wrote:
On 11/12/2009 05:57 PM, Michael Buesch wrote:
On Friday 13 November 2009 00:23:59 Larry Finger wrote:
No, then was a 14e4:4311. I have now installed that same card and it seems
to be
working without the workaround. When I had that
-phy0: Loading firmware version 410.2160
(2007-05-26 15:32:10)
Nov 13 12:14:14 doughnut kernel: b43-phy0: Controller restarted
Nov 13 12:14:14 doughnut kernel: b43-phy0 ERROR: Fatal DMA error: 0x0400,
0x, 0x, 0x, 0x, 0x
Nov 13 12:14:14 doughnut kernel: b43
Nov 13 13:21:28 doughnut kernel: b43-phy0 ERROR: Fatal DMA error: 0x0400,
0x, 0x, 0x, 0x, 0x
Nov 13 13:21:28 doughnut kernel: b43-phy0: Controller RESET (DMA error) ...
Nov 13 13:21:28 doughnut kernel: b43-phy0: Loading firmware version 410.2160
(2007
On 11/13/2009 05:16 AM, Michael Buesch wrote:
Ok, so my guess is that the DMA allocator simply returned high memory
that was unusable to the device. My new code explicitly checks for that (and a
few other things) and retries with GFP_DMA in case the address has illegal
bits set.
That's the
On Friday 13 November 2009 17:05:30 Larry Finger wrote:
(3) When a DMA descriptor error occurs, a dump of the descriptors does not
reveal any obvious problems.
I was going to write a patch that dumps the whole affected ring. But I think we
don't
see something suspicious there, either. So I
On Fri, Nov 13, 2009 at 5:05 PM, Larry Finger larry.fin...@lwfinger.net wrote:
On 11/13/2009 05:16 AM, Michael Buesch wrote:
Ok, so my guess is that the DMA allocator simply returned high memory
that was unusable to the device. My new code explicitly checks for that (and
a
few other things)
Please test the following patch. It changes more stuff related to the
descriptor ring handling (remove the old patch first before applying this one).
http://bu3sch.de/patches/wireless-testing/20091113-1834/patches/001-b43-rewrite-dma-ring-alloc.patch
--
Greetings, Michael.
On 13/11/09 17:36, Michael Buesch wrote:
Please test the following patch. It changes more stuff related to the
descriptor ring handling (remove the old patch first before applying this
one).
http://bu3sch.de/patches/wireless-testing/20091113-1834/patches/001-b43-rewrite-dma-ring-alloc.patch
[ 134.520799] b43-phy0 debug: QoS enabled
[ 134.563410] b43-phy0 debug: Wireless interface started
[ 134.563424] b43-phy0: Controller restarted
[ 134.583034] b43-phy0 ERROR: Fatal DMA error: 0x0400, 0x,
0x, 0x, 0x, 0x
#2 :
[7.834359] b43-phy0: Broadcom
On 11/13/2009 11:36 AM, Michael Buesch wrote:
Please test the following patch. It changes more stuff related to the
descriptor ring handling (remove the old patch first before applying this
one).
Based on a suggestion by Matthew Garrett, please try the patch below.
Thanks,
Larry
=
Index: wireless-testing/drivers/net/wireless/b43/main.c
===
--- wireless-testing.orig/drivers/net/wireless/b43/main.c
+++
Larry Finger wrote:
Based on a suggestion by Matthew Garrett, please try the patch below.
Thanks,
Larry
=
Index: wireless-testing/drivers/net/wireless/b43/main.c
===
---
On 13/11/09 21:38, Larry Finger wrote:
Based on a suggestion by Matthew Garrett, please try the patch below.
I've only been using it for a few minutes but this looks very good. I
compile the kernel with lots of ACPI
CONFIG_ACPI=y
CONFIG_ACPI_SYSFS_POWER=y
CONFIG_ACPI_AC=y
CONFIG_ACPI_BATTERY=y
On 11/13/2009 06:15 PM, William Bourque wrote:
Larry Finger wrote:
Based on a suggestion by Matthew Garrett, please try the patch below.
Thanks,
Larry
=
Index: wireless-testing/drivers/net/wireless/b43/main.c
On Sat, Nov 14, 2009 at 12:41:49AM +, Andrew Benton wrote:
And it seems to be working well. No errors so far. I've just downloaded
a kernel, browsed slashdot a bit. I'll test it some more tomorrow but
this is a BIG step in the right direction. This is the first kernel
that's worked for
On Thursday 12 November 2009 12:42:47 Andrew Benton wrote:
On 11/11/09 19:12, Larry Finger wrote:
Such an error in SLUB handling could be arising from a DMA problem in b43,
but
it could also arise from a memory error. Please run memtest86+ for an
extended
period so that a hardware
On 11/11/09 19:12, Larry Finger wrote:
Such an error in SLUB handling could be arising from a DMA problem in b43, but
it could also arise from a memory error. Please run memtest86+ for an extended
period so that a hardware error can be ruled out. A 24 hour run would be good.
If that is not
On 12/11/09 00:37, Larry Finger wrote:
Andy,
Please try the patch below to see what we can learn from the DMA descriptor
errors. Some of this code is temporary, but there are also some statements
that
will probably become permanent.
Please post any messages that result.
The patch failed
From: Larry Finger larry.fin...@lwfinger.net
User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US;
rv:1.9.1.4pre) Gecko/20090915 SUSE/3.0b4-3.6 Thunderbird/3.0b4
MIME-Version: 1.0
To: Andrew Benton b3n...@gmail.com
Subject: Re: b43-phy0 ERROR: Fatal DMA error: 0x0400
References
On 11/12/2009 06:16 AM, Andrew Benton wrote:
On 12/11/09 00:37, Larry Finger wrote:
Andy,
Please try the patch below to see what we can learn from the DMA descriptor
errors. Some of this code is temporary, but there are also some statements
that
will probably become permanent.
Please
On 12/11/09 15:50, Larry Finger wrote:
Sorry about the kernel mismatch. I developed that patch while offline and
waiting at an auto repair place and forgot to refresh my sources before
sending
it. The revised version that Michael sent should work. If not, please let me
know and I will send
On Thursday 12 November 2009 17:42:07 Andrew Benton wrote:
On 12/11/09 15:50, Larry Finger wrote:
Sorry about the kernel mismatch. I developed that patch while offline and
waiting at an auto repair place and forgot to refresh my sources before
sending
it. The revised version that Michael
On 11/12/2009 10:53 AM, Michael Buesch wrote:
On Thursday 12 November 2009 17:42:07 Andrew Benton wrote:
On 12/11/09 15:50, Larry Finger wrote:
Sorry about the kernel mismatch. I developed that patch while offline and
waiting at an auto repair place and forgot to refresh my sources before
(2007-05-26 15:32:10)
Nov 12 18:40:49 doughnut kernel: b43-phy0: Controller restarted
Nov 12 18:40:49 doughnut kernel: b43-phy0 ERROR: Fatal DMA error: 0x0400,
0x, 0x, 0x, 0x, 0x
Nov 12 18:40:49 doughnut kernel: b43-phy0: Controller RESET (DMA error
On Thursday 12 November 2009 19:45:45 Andrew Benton wrote:
On 12/11/09 17:14, Larry Finger wrote:
I guess I'm a failure at writing diagnostic patches. Until there is a DMA
error,
the only effect of the patch is to add a little extra time to the routine
that
fills in the descriptor
On 11/12/2009 12:45 PM, Andrew Benton wrote:
On 12/11/09 17:14, Larry Finger wrote:
I guess I'm a failure at writing diagnostic patches. Until there is a DMA
error,
the only effect of the patch is to add a little extra time to the routine
that
fills in the descriptor structure, and it adds
Michael Buesch wrote:
On Thursday 12 November 2009 19:45:45 Andrew Benton wrote:
On 12/11/09 17:14, Larry Finger wrote:
I guess I'm a failure at writing diagnostic patches. Until there is a DMA
error,
the only effect of the patch is to add a little extra time to the routine
that
fills in
On Thursday 12 November 2009 20:33:54 Larry Finger wrote:
Nov 12 18:40:43 doughnut kernel: b43: Descr. 0: 0x0 0x930 0x364BD020
0x8000
Nov 12 18:40:43 doughnut kernel: b43: Descr. 1: 0x0 0x930 0x364BF020
0x8000
Nov 12 18:40:43 doughnut kernel: b43: Descr. 2: 0x0 0x930
On Thursday 12 November 2009 20:32:32 William Bourque wrote:
Sorry for the late reply... I seem to have the exact same bug here. Do
you need more people to run the diagnostic patch?
Well, it doesn't hurt.
--
Greetings, Michael.
___
Bcm43xx-dev
On 11/12/2009 01:48 PM, Michael Buesch wrote:
Now we have some progress. You will note the difference in the control words
(first 2 columns) for descriptors 8 9. They are wrong.
What do you think is wrong here? I think the control words are OK.
At the point where I captured them, I didn't
On Thursday 12 November 2009 21:10:59 Larry Finger wrote:
Do the address_low values for 8 and 9 look right? They
should be aligned on a 4K boundary.
Is this really a requirement? I think the 4k alignment is only required
for the descriptor memory. We never guaranteed any alignment for the skbs.
On Thursday 12 November 2009 20:33:54 Larry Finger wrote:
While Michael is coming up with a test patch,
Here you go:
http://bu3sch.de/patches/wireless-testing/20091112-2213/patches/001-b43-rewrite-dma-ring-alloc.patch
Please test this patch (also on 64bit-DMA devices that currently work).
It
] b43-phy0 debug: 64-bit DMA initialized
[ 109.213169] b43-phy0 debug: QoS enabled
[ 109.253562] b43-phy0 debug: Wireless interface started
[ 109.272964] b43-phy0 ERROR: Fatal DMA error: 0x0400, 0x,
0x, 0x, 0x, 0x
[ 109.272990] b43-phy0: Controller
On Thursday 12 November 2009 22:34:00 William Bourque wrote:
Michael Buesch wrote:
On Thursday 12 November 2009 20:32:32 William Bourque wrote:
Sorry for the late reply... I seem to have the exact same bug here. Do
you need more people to run the diagnostic patch?
Well, it doesn't
debug: 64-bit DMA initialized
[ 181.210719] b43-phy0 debug: QoS enabled
[ 181.251102] b43-phy0 debug: Wireless interface started
[ 181.251113] b43-phy0 debug: Adding Interface type 2
[ 181.270285] b43-phy0 ERROR: Fatal DMA error: 0x0400, 0x,
0x, 0x, 0x
On Thursday 12 November 2009 22:59:26 William Bourque wrote:
Michael Buesch wrote:
On Thursday 12 November 2009 22:34:00 William Bourque wrote:
Michael Buesch wrote:
On Thursday 12 November 2009 20:32:32 William Bourque wrote:
Sorry for the late reply... I seem to have the exact same bug
On 11/12/2009 03:16 PM, Michael Buesch wrote:
On Thursday 12 November 2009 20:33:54 Larry Finger wrote:
While Michael is coming up with a test patch,
Here you go:
http://bu3sch.de/patches/wireless-testing/20091112-2213/patches/001-b43-rewrite-dma-ring-alloc.patch
Please test this patch
On Friday 13 November 2009 00:04:50 Larry Finger wrote:
On 11/12/2009 03:16 PM, Michael Buesch wrote:
On Thursday 12 November 2009 20:33:54 Larry Finger wrote:
While Michael is coming up with a test patch,
Here you go:
On 11/12/2009 05:08 PM, Michael Buesch wrote:
On Friday 13 November 2009 00:04:50 Larry Finger wrote:
On 11/12/2009 03:16 PM, Michael Buesch wrote:
On Thursday 12 November 2009 20:33:54 Larry Finger wrote:
While Michael is coming up with a test patch,
Here you go:
On Friday 13 November 2009 00:23:59 Larry Finger wrote:
No, then was a 14e4:4311. I have now installed that same card and it seems to
be
working without the workaround. When I had that problem, I had a different
laptop than I do now, thus it is not possible to reduplicate the setup. I am
On 11/12/2009 05:57 PM, Michael Buesch wrote:
On Friday 13 November 2009 00:23:59 Larry Finger wrote:
No, then was a 14e4:4311. I have now installed that same card and it seems
to be
working without the workaround. When I had that problem, I had a different
laptop than I do now, thus it is
On 11/11/09 02:58, Larry Finger wrote:
On 11/10/2009 06:48 PM, Andrew Benton wrote:
Nov 11 00:04:37 doughnut kernel: b43-phy0 ERROR: Fatal DMA error:
0x0400, 0x, 0x, 0x, 0x, 0x
Nov 11 00:04:37 doughnut kernel: b43-phy0: Controller RESET (DMA error
I recompiled my kernel today (current wireless-testing) and disabled ACPI
entirely.
It worked fine for 10 minutes and then the internet connection died and left
this
in /var/log/sys.log
Nov 11 14:31:31 doughnut ntpd[398]: kernel time sync status change 2001
Nov 11 14:36:57 doughnut ntpd[398]:
On 11/11/2009 08:52 AM, Andrew Benton wrote:
I recompiled my kernel today (current wireless-testing) and disabled ACPI
entirely.
It worked fine for 10 minutes and then the internet connection died and left
this
in /var/log/sys.log
Nov 11 14:31:31 doughnut ntpd[398]: kernel time sync
On Wednesday 11 November 2009 20:12:59 Larry Finger wrote:
On 11/11/2009 08:52 AM, Andrew Benton wrote:
I recompiled my kernel today (current wireless-testing) and disabled ACPI
entirely.
It worked fine for 10 minutes and then the internet connection died and
left this
in
Andy,
Please try the patch below to see what we can learn from the DMA descriptor
errors. Some of this code is temporary, but there are also some statements that
will probably become permanent.
Please post any messages that result.
Larry
Index: wireless-testing/drivers/net/wireless/b43/dma.c
: Controller RESET (DMA error) ...
Nov 11 00:04:31 doughnut kernel: b43-phy0: Loading firmware version
410.2160 (2007-05-26 15:32:10)
Nov 11 00:04:37 doughnut kernel: b43-phy0: Controller restarted
Nov 11 00:04:37 doughnut kernel: b43-phy0 ERROR: Fatal DMA error:
0x0400, 0x, 0x
-phy0 ERROR: Fatal DMA error:
0x0400, 0x, 0x, 0x, 0x, 0x
Nov 11 00:04:37 doughnut kernel: b43-phy0: Controller RESET (DMA error) ...
The 0x0800 indicates a descriptor problem. Why it should happen after 15
minutes is perplexing. If I write a diagnostic
55 matches
Mail list logo
