brcmfmac not working on ECS LIVA

brcmfmac not working on ECS LIVA

[Felipe Contreras]

Hi,

I just got an ECS LIVA and I haven't been able to get the wireless
working. First I got the error that the brcmfmac43241b4-sdio.txt file
was missing. I got the file from the ECS's beta drivers [1], but it's
still not working.

brcmfmac: brcmf_sdio_drivestrengthinit: No SDIO Drive strength init
done for chip 4324 rev 5 pmurev 17
brcmfmac: brcmf_c_preinit_dcmds: Firmware version = wl0: Jul 17 2013
07:36:07 version 6.10.197.71 (r412987) FWID 01-882d2634
brcmfmac: brcmf_add_if: ERROR: netdev:wlan0 already exists
brcmfmac: brcmf_add_if: ignore IF event

WPA association/authentication failed for interface 'wlan0'

I also tried a nvram file that is stored in EFI variables, but it
didn't work either.

Some things seem to work some times, but not the whole association.

This is a 3.17.2 kernel.

Any ideas?

[1] http://www.ecs.com.tw/LIVA/support.html

-- 
Felipe Contreras

Re: brcmfmac not working on ECS LIVA

[Arend van Spriel]

On 11/08/14 21:39, Felipe Contreras wrote:
> Hi,
>
> I just got an ECS LIVA and I haven't been able to get the wireless
> working. First I got the error that the brcmfmac43241b4-sdio.txt file
> was missing. I got the file from the ECS's beta drivers [1], but it's
> still not working.
>
> brcmfmac: brcmf_sdio_drivestrengthinit: No SDIO Drive strength init
> done for chip 4324 rev 5 pmurev 17
> brcmfmac: brcmf_c_preinit_dcmds: Firmware version = wl0: Jul 17 2013
> 07:36:07 version 6.10.197.71 (r412987) FWID 01-882d2634
> brcmfmac: brcmf_add_if: ERROR: netdev:wlan0 already exists
> brcmfmac: brcmf_add_if: ignore IF event
>
> WPA association/authentication failed for interface 'wlan0'
>
> I also tried a nvram file that is stored in EFI variables, but it
> didn't work either.
>
> Some things seem to work some times, but not the whole association.
>
> This is a 3.17.2 kernel.
>
> Any ideas?

Hi Felipe,

The ideas may come to us with a bit more trace information. The logging 
above is the usual stuff so no problem indication there.

I looked at the nvram file from [1]. At the end there are some entries 
about OOB wakeup, ie. starting with 'sd_'. You may remove those, but 
giving the logging it would not be the first thing to do.

Could you share the nvram obtained from EFI? Also please create a kernel 
log with debug=0xd416.

Regards,
Arend

> [1] http://www.ecs.com.tw/LIVA/support.html
>

Re: brcmfmac not working on ECS LIVA

[Felipe Contreras]

Hi,

On Sun, Nov 9, 2014 at 4:46 AM, Arend van Spriel <arend@broadcom.com> wrote:

> The ideas may come to us with a bit more trace information. The logging
> above is the usual stuff so no problem indication there.
>
> I looked at the nvram file from [1]. At the end there are some entries about
> OOB wakeup, ie. starting with 'sd_'. You may remove those, but giving the
> logging it would not be the first thing to do.

I removed the sd_ lines, it didn't help.

> Could you share the nvram obtained from EFI? Also please create a kernel log
> with debug=0xd416.

Sure, the nvram file is here:

http://pastie.org/9710358

I tried adding this file:

/etc/modprobe.d/brcmfmac.conf
options brcmfmac debug=0xd416

I see nothing extra in the debug log... Am I doing something wrong?

-- 
Felipe Contreras

Re: brcmfmac not working on ECS LIVA

[Arend van Spriel]

On 10-11-14 21:48, Felipe Contreras wrote:
> Hi,
> 
> On Sun, Nov 9, 2014 at 4:46 AM, Arend van Spriel <arend@broadcom.com> wrote:
> 
>> The ideas may come to us with a bit more trace information. The logging
>> above is the usual stuff so no problem indication there.
>>
>> I looked at the nvram file from [1]. At the end there are some entries about
>> OOB wakeup, ie. starting with 'sd_'. You may remove those, but giving the
>> logging it would not be the first thing to do.
> 
> I removed the sd_ lines, it didn't help.
> 
>> Could you share the nvram obtained from EFI? Also please create a kernel log
>> with debug=0xd416.
> 
> Sure, the nvram file is here:
> 
> http://pastie.org/9710358
> 
> I tried adding this file:
> 
> /etc/modprobe.d/brcmfmac.conf
> options brcmfmac debug=0xd416
> 
> I see nothing extra in the debug log... Am I doing something wrong?

Not sure. You can verify and set the log level through sysfs, ie.:

$ cat /sys/module/brcmfmac/parameters/debug

To set it:

$ echo 0xd416 > /sys/module/brcmfmac/parameters/debug

Regards,
Arend

Re: brcmfmac not working on ECS LIVA

[Felipe Contreras]

On Tue, Nov 11, 2014 at 5:08 AM, Arend van Spriel <arend@broadcom.com> wrote:
> On 10-11-14 21:48, Felipe Contreras wrote:

>> /etc/modprobe.d/brcmfmac.conf
>> options brcmfmac debug=0xd416
>>
>> I see nothing extra in the debug log... Am I doing something wrong?
>
> Not sure. You can verify and set the log level through sysfs, ie.:
>
> $ cat /sys/module/brcmfmac/parameters/debug

54294

I guess ArchLinux's kernel is missing some configuration option needed
for the debugging.

-- 
Felipe Contreras

Re: brcmfmac not working on ECS LIVA

[Arend van Spriel]

On 11-11-14 19:32, Felipe Contreras wrote:
> On Tue, Nov 11, 2014 at 5:08 AM, Arend van Spriel <arend@broadcom.com> wrote:
>> On 10-11-14 21:48, Felipe Contreras wrote:
> 
>>> /etc/modprobe.d/brcmfmac.conf
>>> options brcmfmac debug=0xd416
>>>
>>> I see nothing extra in the debug log... Am I doing something wrong?
>>
>> Not sure. You can verify and set the log level through sysfs, ie.:
>>
>> $ cat /sys/module/brcmfmac/parameters/debug
> 
> 54294
> 
> I guess ArchLinux's kernel is missing some configuration option needed
> for the debugging.

Can you boot the kernel with 'debug ignore_loglevel' on the kernel
command line.

Regards,
Arend

Re: brcmfmac not working on ECS LIVA

[Felipe Contreras]

On Tue, Nov 11, 2014 at 12:35 PM, Arend van Spriel <arend@broadcom.com> wrote:
> On 11-11-14 19:32, Felipe Contreras wrote:
>> On Tue, Nov 11, 2014 at 5:08 AM, Arend van Spriel <arend@broadcom.com> wrote:
>>> On 10-11-14 21:48, Felipe Contreras wrote:
>>
>>>> /etc/modprobe.d/brcmfmac.conf
>>>> options brcmfmac debug=0xd416
>>>>
>>>> I see nothing extra in the debug log... Am I doing something wrong?
>>>
>>> Not sure. You can verify and set the log level through sysfs, ie.:
>>>
>>> $ cat /sys/module/brcmfmac/parameters/debug
>>
>> 54294
>>
>> I guess ArchLinux's kernel is missing some configuration option needed
>> for the debugging.
>
> Can you boot the kernel with 'debug ignore_loglevel' on the kernel
> command line.

Yes. I see more systemd messages, nothing extra from brcmfmac.

-- 
Felipe Contreras

Re: brcmfmac not working on ECS LIVA

[Arend van Spriel]

On 11/11/14 19:49, Felipe Contreras wrote:
> On Tue, Nov 11, 2014 at 12:35 PM, Arend van Spriel<arend@broadcom.com>  wrote:
>> On 11-11-14 19:32, Felipe Contreras wrote:
>>> On Tue, Nov 11, 2014 at 5:08 AM, Arend van Spriel<arend@broadcom.com>  wrote:
>>>> On 10-11-14 21:48, Felipe Contreras wrote:
>>>
>>>>> /etc/modprobe.d/brcmfmac.conf
>>>>> options brcmfmac debug=0xd416
>>>>>
>>>>> I see nothing extra in the debug log... Am I doing something wrong?
>>>>
>>>> Not sure. You can verify and set the log level through sysfs, ie.:
>>>>
>>>> $ cat /sys/module/brcmfmac/parameters/debug
>>>
>>> 54294
>>>
>>> I guess ArchLinux's kernel is missing some configuration option needed
>>> for the debugging.
>>
>> Can you boot the kernel with 'debug ignore_loglevel' on the kernel
>> command line.
>
> Yes. I see more systemd messages, nothing extra from brcmfmac.

Ok. Probably brcmfmac was not built with CONFIG_BRCMDBG. Can you 
checkout kernel config.

Regards,
Arend

Re: brcmfmac not working on ECS LIVA

[Felipe Contreras]

On Tue, Nov 11, 2014 at 1:10 PM, Arend van Spriel <arend@broadcom.com> wrote:
> On 11/11/14 19:49, Felipe Contreras wrote:
>>
>> On Tue, Nov 11, 2014 at 12:35 PM, Arend van Spriel<arend@broadcom.com>
>> wrote:
>>>
>>> On 11-11-14 19:32, Felipe Contreras wrote:
>>>>
>>>> On Tue, Nov 11, 2014 at 5:08 AM, Arend van Spriel<arend@broadcom.com>
>>>> wrote:
>>>>>
>>>>> On 10-11-14 21:48, Felipe Contreras wrote:
>>>>
>>>>
>>>>>> /etc/modprobe.d/brcmfmac.conf
>>>>>> options brcmfmac debug=0xd416
>>>>>>
>>>>>> I see nothing extra in the debug log... Am I doing something wrong?
>>>>>
>>>>>
>>>>> Not sure. You can verify and set the log level through sysfs, ie.:
>>>>>
>>>>> $ cat /sys/module/brcmfmac/parameters/debug
>>>>
>>>>
>>>> 54294
>>>>
>>>> I guess ArchLinux's kernel is missing some configuration option needed
>>>> for the debugging.
>>>
>>>
>>> Can you boot the kernel with 'debug ignore_loglevel' on the kernel
>>> command line.
>>
>>
>> Yes. I see more systemd messages, nothing extra from brcmfmac.
>
>
> Ok. Probably brcmfmac was not built with CONFIG_BRCMDBG. Can you checkout
> kernel config.

That's right; it's not set. I will enable it and rebuild.

-- 
Felipe Contreras

Re: brcmfmac not working on ECS LIVA

[Felipe Contreras]

On Sun, Nov 9, 2014 at 4:46 AM, Arend van Spriel <arend@broadcom.com> wrote:

> Could you share the nvram obtained from EFI? Also please create a kernel log
> with debug=0xd416.

There you go:
http://people.freedesktop.org/~felipec/dmesg-brcm.txt

-- 
Felipe Contreras

Re: brcmfmac not working on ECS LIVA

[Arend van Spriel]

On 11-11-14 23:32, Felipe Contreras wrote:
> On Sun, Nov 9, 2014 at 4:46 AM, Arend van Spriel <arend@broadcom.com> wrote:
> 
>> Could you share the nvram obtained from EFI? Also please create a kernel log
>> with debug=0xd416.
> 
> There you go:
> http://people.freedesktop.org/~felipec/dmesg-brcm.txt
> 

It is using the sdhci-acpi driver for the sdio host controller. If I am
not mistaken that one does runtime-pm. This has caused issues with our
device. This is what happens:

1. driver sends control message and waits for firmware response.
 2. runtime pm kicks in and suspends host controller.
  3. host controller suspend disable sdio clock.
   4. host controller does not detect firmware interrupt for response.
    5. control message times out.

You can disable it through sysfs, but the patch below may do it.

Regards,
Arend

diff --git a/drivers/net/wireless/brcm80211/brcmfmac/bcmsdh.c
b/drivers/net/wire
index f754ffc..d231e65 100644
--- a/drivers/net/wireless/brcm80211/brcmfmac/bcmsdh.c
+++ b/drivers/net/wireless/brcm80211/brcmfmac/bcmsdh.c
@@ -27,6 +27,7 @@
 #include <linux/mmc/sdio_func.h>
 #include <linux/mmc/card.h>
 #include <linux/mmc/host.h>
+#include <linux/pm_runtime.h>
 #include <linux/platform_device.h>
 #include <linux/platform_data/brcmfmac-sdio.h>
 #include <linux/suspend.h>
@@ -976,6 +977,12 @@ static int brcmf_sdiod_probe(struct brcmf_sdio_dev
*sdiodev
                                           SG_MAX_SINGLE_ALLOC);
        sdiodev->max_segment_size = host->max_seg_size;

+       /*
+        * Several issues with the driver seem to occur when the SDIO host
+        * is using runtime-pm. Prevent host from doing runtime-pm here.
+        */
+       __pm_runtime_use_autosuspend(&host->class_dev, false);
+
        /* allocate scatter-gather table. sg support
         * will be disabled upon allocation failure.
         */

Re: brcmfmac not working on ECS LIVA

[Felipe Contreras]

On Wed, Nov 12, 2014 at 4:40 AM, Arend van Spriel <arend@broadcom.com> wrote:
> On 11-11-14 23:32, Felipe Contreras wrote:
>> On Sun, Nov 9, 2014 at 4:46 AM, Arend van Spriel <arend@broadcom.com> wrote:
>>
>>> Could you share the nvram obtained from EFI? Also please create a kernel log
>>> with debug=0xd416.
>>
>> There you go:
>> http://people.freedesktop.org/~felipec/dmesg-brcm.txt
>>
>
> It is using the sdhci-acpi driver for the sdio host controller. If I am
> not mistaken that one does runtime-pm. This has caused issues with our
> device. This is what happens:
>
> 1. driver sends control message and waits for firmware response.
>  2. runtime pm kicks in and suspends host controller.
>   3. host controller suspend disable sdio clock.
>    4. host controller does not detect firmware interrupt for response.
>     5. control message times out.
>
> You can disable it through sysfs, but the patch below may do it.

I tried this patch, and although it seems to make a difference, it
still doesn't work.

http://people.freedesktop.org/~felipec/dmesg-brcm-2.txt

-- 
Felipe Contreras

Re: brcmfmac not working on ECS LIVA

[Arend van Spriel]

On 12-11-14 16:23, Felipe Contreras wrote:
> On Wed, Nov 12, 2014 at 4:40 AM, Arend van Spriel <arend@broadcom.com> wrote:
>> On 11-11-14 23:32, Felipe Contreras wrote:
>>> On Sun, Nov 9, 2014 at 4:46 AM, Arend van Spriel <arend@broadcom.com> wrote:
>>>
>>>> Could you share the nvram obtained from EFI? Also please create a kernel log
>>>> with debug=0xd416.
>>>
>>> There you go:
>>> http://people.freedesktop.org/~felipec/dmesg-brcm.txt
>>>
>>
>> It is using the sdhci-acpi driver for the sdio host controller. If I am
>> not mistaken that one does runtime-pm. This has caused issues with our
>> device. This is what happens:
>>
>> 1. driver sends control message and waits for firmware response.
>>  2. runtime pm kicks in and suspends host controller.
>>   3. host controller suspend disable sdio clock.
>>    4. host controller does not detect firmware interrupt for response.
>>     5. control message times out.
>>
>> You can disable it through sysfs, but the patch below may do it.
> 
> I tried this patch, and although it seems to make a difference, it
> still doesn't work.
> 
> http://people.freedesktop.org/~felipec/dmesg-brcm-2.txt

Hi Felipe,

All the messages to the firmware seem ok. If connection would succeed I
normally see a LINK event followed by SET_SSID event. In your log I only
see the SET_SSID event with failed status, hence the connect is
unsuccessful. Do you have any control over the AP? Could you try without
WPA security to quickly confirm connection works? May help in root
causing the issue.

Regards,
Arend

Re: brcmfmac not working on ECS LIVA

[Felipe Contreras]

On Thu, Nov 13, 2014 at 4:47 AM, Arend van Spriel <arend@broadcom.com> wrote:
> On 12-11-14 16:23, Felipe Contreras wrote:
>> On Wed, Nov 12, 2014 at 4:40 AM, Arend van Spriel <arend@broadcom.com> wrote:

>> I tried this patch, and although it seems to make a difference, it
>> still doesn't work.
>>
>> http://people.freedesktop.org/~felipec/dmesg-brcm-2.txt
>
> Hi Felipe,
>
> All the messages to the firmware seem ok. If connection would succeed I
> normally see a LINK event followed by SET_SSID event. In your log I only
> see the SET_SSID event with failed status, hence the connect is
> unsuccessful. Do you have any control over the AP? Could you try without
> WPA security to quickly confirm connection works? May help in root
> causing the issue.

I tried disabled WPA security, it's still not working.

-- 
Felipe Contreras

Re: brcmfmac not working on ECS LIVA

[Arend van Spriel]

On 11/14/14 18:03, Felipe Contreras wrote:
> On Thu, Nov 13, 2014 at 4:47 AM, Arend van Spriel<arend@broadcom.com>  wrote:
>> On 12-11-14 16:23, Felipe Contreras wrote:
>>> On Wed, Nov 12, 2014 at 4:40 AM, Arend van Spriel<arend@broadcom.com>  wrote:
>
>>> I tried this patch, and although it seems to make a difference, it
>>> still doesn't work.
>>>
>>> http://people.freedesktop.org/~felipec/dmesg-brcm-2.txt
>>
>> Hi Felipe,
>>
>> All the messages to the firmware seem ok. If connection would succeed I
>> normally see a LINK event followed by SET_SSID event. In your log I only
>> see the SET_SSID event with failed status, hence the connect is
>> unsuccessful. Do you have any control over the AP? Could you try without
>> WPA security to quickly confirm connection works? May help in root
>> causing the issue.
>
> I tried disabled WPA security, it's still not working.

Thanks, Felipe

Could you check the runtime-pm for mmc1 controller. You need to find 
sdhci-acpi driver in sysfs. I think it is under 
/sys/bus/platform/drivers. One of the bound devices is mmc1 (although 
probably will have different nameing there) and there is subfolder power 
with file named control. Can you cat that file and do 'echo on > 
control' if it says auto.

Regards,
Arend