Skip to contentSkip to conversation

Revert "ath: add support for special 0x0 regulatory domain"

PATCH In review Félix Sipma 5 participants 7 messages Oct 22, 2020 – Dec 21, 2020 1 message not in the archive Qualcomm Atheros ATH10K wireless driver

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Revert "ath: add support for special 0x0 regulatory domain"

Hello,

Sorry if I break the thread, I'm not subscribed and do not have access to the message-id.

I noticed the last message about this issue was on August:
https://lists.infradead.org/pipermail/ath10k/2020-August/011785.html

I got caught by this bug, too. I use a Compex wle900vx (qca988x).

Is there something preventing https://patchwork.kernel.org/project/ath10k/patch/[email protected]/
from being merged?

Regards,

Signature · 1 line
Félix

_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k

A message that is not in the archive (sent off-list or not yet received): CANe27jKpYm29QOjYOZ_jwHiRxuWx66J+th8-zgbXK4geiCU0_Q@mail.gmail.com

#2 lore raw

Hi,

Thanks for your detailed answer.

I made a detailed report, with dmesg outputs of different kernel/firmware-atheros and wireless-regdb combinations at https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=970679

Massimo Maggi pointed me to this ath10k issue on the last message.

Regards,

Quoted 16 lines · Jouni Malinen

What is the exact issue you are seeing?

Quoted 4 lines

Is there something preventing

https://patchwork.kernel.org/project/ath10k/patch/[email protected]/
from being merged?

That seems to have a risk of mapping the default country code (CTRY_DEFAULT is defined to be 0 in regd.h) being mapped to something else than a safe world roaming case and this could have an impact throughout all drivers using the generic ath/regd.c implementation. It looks like the correct thing to do is to map an unexpected value to the safest option instead of allowing this to be mapped to something that could end up enabling incorrect regulatory rules.

- Jouni

Signature · 1 line
Félix

_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k

#3 lore raw
Quoted 3 lines · Félix Sipma

I made a detailed report, with dmesg outputs of different kernel/firmware-atheros and wireless-regdb combinations at
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=970679

So the issue is in not being able to operate an AP on the 5 GHz band? That sounds like the expected behavior for any device that has not been calibrated and provisioned for a specific country where regulatory rules allow operation on the 5 GHz band. I understand that this may look like a regression since the commit removed functionality, but it feels like a bug fix to me since that functionality should not have been enabled by default in the first place. The goal here is to avoid inappropriate operation on the band without explicit configuration to enable such operation. In AP devices, the device should have been provisioned for a specific country to be able to enforce the correct frequency range restrictions. The safe default for a device that does not have such explicit configuration within the WLAN component itself is to use the world roaming mode which prevents initialization of radiation (i.e., does not allow AP mode to be started but allows station mode operation to connect to an already started AP) on the 5 GHz band.

- Jouni

_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k

#4 lore raw
Quoted 6 lines · Jouni Malinen
Quoted 3 lines · Félix Sipma

I made a detailed report, with dmesg outputs of different kernel/firmware-atheros and wireless-regdb combinations at
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=970679

So the issue is in not being able to operate an AP on the 5 GHz band?

Yes, sorry for not being clear about this.

Quoted 7 lines

That sounds like the expected behavior for any device that has not been calibrated and provisioned for a specific country where regulatory rules allow operation on the 5 GHz band. I understand that this may look like a regression since the commit removed functionality, but it feels like a bug fix to me since that functionality should not have been enabled by default in the first place. The goal here is to avoid inappropriate operation on the band

Quoted 8 lines

without explicit configuration to enable such operation. In AP devices, the device should have been provisioned for a specific country to be able to enforce the correct frequency range restrictions. The safe default for a device that does not have such explicit configuration within the WLAN component itself is to use the world roaming mode which prevents initialization of radiation (i.e., does not allow AP mode to be started but allows station mode operation to connect to an already started AP) on the 5 GHz band.

So, if I understand well, the new behaviour (non working AP with ac) is expected, given that the device was not provisioned with the settings specifics to my zone? And to prevent people from operating an AP with the wrong settings, we force them to use hardware that is specific to their zone? If that's the intention, I'm not sure it is very effective: I can still buy a German wifi card and use it in France. It will probably suits most of the hardware vendors, though: people moving to another country are forced to buy new stuff and they can't buy cheaper stuff abroad, but, seen from another angle, it sounds like planned obsolescence.

The vendor of my (new) hardware is from Switzerland, and is selling to (at least) everywhere in Europe, and they sell the same hardware (that used to work) for everybody. So, now, they are supposed to sell specific hardware for each country? I think the card was not provisioned with a specific country (I'm in France, and I tried FR, CH, US, ...), so I guess I have to throw it to the bin anyway (and they should do the same with all their stocks).

I hope I have misunderstood your message, else I have to admit I'm a little disappointed... Thanks for your explanation, though!

If that's really the intention, is there a workaround for irresponsible people like me, who just wan't to avoid buying new stuff?

Regards,

Signature · 1 line
Félix

_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k

#5 lore raw

​Hello,

Has a decision been reached as to whether or not this commit will be reverted? It appears to still exist in kernel 5.10.

I too have a Compex WLE900VX QCA986x/988x 802.11ac card and am affected. I purchased the card from Switzerland and am running it in Canada.

I've been using the v5.4 LTS kernel as a temporary workaround, but given 5.10 will be the next LTS it would be helpful to have a definitive answer one way or another on this issue so those of us who are affected can plan accordingly.

The last update I'm aware of is Alvin's argument to revert the patch: https://lists.infradead.org/pipermail/ath10k/2020-October/012049.html .

Kind regards,
Julian
​

From: ath10k <[email protected]> on behalf of Félix Sipma <[email protected]>
Sent: October 30, 2020 8:51 AM
To: Jouni Malinen <[email protected]>
Cc: [email protected] <[email protected]>
Subject: Re: [PATCH] Revert "ath: add support for special 0x0 regulatory domain"

Quoted 6 lines · Jouni Malinen
Quoted 3 lines · Félix Sipma

I made a detailed report, with dmesg outputs of different kernel/firmware-atheros and wireless-regdb combinations at
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=970679

So the issue is in not being able to operate an AP on the 5 GHz band?

Yes, sorry for not being clear about this.

Quoted 7 lines

That sounds like the expected behavior for any device that has not been calibrated and provisioned for a specific country where regulatory rules allow operation on the 5 GHz band. I understand that this may look like a regression since the commit removed functionality, but it feels like a bug fix to me since that functionality should not have been enabled by default in the first place. The goal here is to avoid inappropriate operation on the band

Quoted 8 lines

without explicit configuration to enable such operation. In AP devices, the device should have been provisioned for a specific country to be able to enforce the correct frequency range restrictions. The safe default for a device that does not have such explicit configuration within the WLAN component itself is to use the world roaming mode which prevents initialization of radiation (i.e., does not allow AP mode to be started but allows station mode operation to connect to an already started AP) on the 5 GHz band.

So, if I understand well, the new behaviour (non working AP with ac) is expected, given that the device was not provisioned with the settings specifics to my zone? And to prevent people from operating an AP with the wrong settings, we force them to use hardware that is specific to their zone? If that's the intention, I'm not sure it is very effective:
I can still buy a German wifi card and use it in France. It will probably suits most of the hardware vendors, though: people moving to another country are forced to buy new stuff and they can't buy cheaper stuff abroad, but, seen from another angle, it sounds like planned obsolescence.

The vendor of my (new) hardware is from Switzerland, and is selling to (at least) everywhere in Europe, and they sell the same hardware (that used to work) for everybody. So, now, they are supposed to sell specific hardware for each country? I think the card was not provisioned with a specific country (I'm in France, and I tried FR, CH, US,  ...), so I guess I have to throw it to the bin anyway (and they should do the same with all their stocks).

I hope I have misunderstood your message, else I have to admit I'm a little disappointed... Thanks for your explanation, though!

If that's really the intention, is there a workaround for irresponsible people like me, who just wan't to avoid buying new stuff?

Regards,

Signature · 5 lines
Félix
_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k
#6 lore raw
In reply to Jouni Malinen

Hi,

Quoted 8 lines · Jouni Malinen

So the issue is in not being able to operate an AP on the 5 GHz band? That sounds like the expected behavior for any device that has not been calibrated and provisioned for a specific country where regulatory rules allow operation on the 5 GHz band. I understand that this may look like a regression since the commit removed functionality, but it feels like a bug fix to me since that functionality should not have been enabled by default in the first place. The goal here is to avoid inappropriate operation on the band

Quoted 8 lines

without explicit configuration to enable such operation. In AP devices, the device should have been provisioned for a specific country to be able to enforce the correct frequency range restrictions. The safe default for a device that does not have such explicit configuration within the WLAN component itself is to use the world roaming mode which prevents initialization of radiation (i.e., does not allow AP mode to be started but allows station mode operation to connect to an already started AP) on the 5 GHz band.

This explanation makes perfect sense, but are you sure that that is what the 0x0 regulatory domain means? For example the pages [1][2] state unequivocally that 0x0 means US.

If we assume 0x0 = US, do you agree that an EEPROM programmed with 0x0 should imply that the device is suitably calibrated for operation in the United States, and as such should permit initiating radiation on the 5 GHz band (etc.)? Or have I missed something?

I also tried to get the patch in question reverted based on this rationale, see [3] for more info. I hope somebody can help clear up this misunderstanding, since it seems to me that the main issue is: what is the true meaning of 0x0?

Btw, the hardware that Félix is referring to does have an FCC ID[4] and seems to be suitably certified for the US.

Kind regards,
Alvin

[1]
https://wireless.wiki.kernel.org/en/users/drivers/ath#the_0x0_regulatory_domain
[2]
https://web.archive.org/web/20150117040022/http://article.gmane.org/gmane.linux.kernel.wireless.general/38410
[3]
https://patchwork.kernel.org/project/ath10k/patch/[email protected]/
[4] https://fccid.io/TK4WLE900VX
_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k

#7 lore raw

Am 30.10.20 um 14:23 schrieb Alvin Sipraga:

Quoted 39 lines

Hi,

Quoted 8 lines · Jouni Malinen

So the issue is in not being able to operate an AP on the 5 GHz band? That sounds like the expected behavior for any device that has not been calibrated and provisioned for a specific country where regulatory rules allow operation on the 5 GHz band. I understand that this may look like a regression since the commit removed functionality, but it feels like a bug fix to me since that functionality should not have been enabled by default in the first place. The goal here is to avoid inappropriate operation on the band

Quoted 8 lines

without explicit configuration to enable such operation. In AP devices, the device should have been provisioned for a specific country to be able to enforce the correct frequency range restrictions. The safe default for a device that does not have such explicit configuration within the WLAN component itself is to use the world roaming mode which prevents initialization of radiation (i.e., does not allow AP mode to be started but allows station mode operation to connect to an already started AP) on the 5 GHz band.

This explanation makes perfect sense, but are you sure that that is what the 0x0 regulatory domain means? For example the pages [1][2] state unequivocally that 0x0 means US.

If we assume 0x0 = US, do you agree that an EEPROM programmed with 0x0 should imply that the device is suitably calibrated for operation in the United States, and as such should permit initiating radiation on the 5 GHz band (etc.)? Or have I missed something?

I also tried to get the patch in question reverted based on this rationale, see [3] for more info. I hope somebody can help clear up this misunderstanding, since it seems to me that the main issue is: what is the true meaning of 0x0?

Btw, the hardware that Félix is referring to does have an FCC ID[4] and seems to be suitably certified for the US.

Kind regards,
Alvin

Quoted 11 lines

[1]
https://wireless.wiki.kernel.org/en/users/drivers/ath#the_0x0_regulatory_domain
[2]
https://web.archive.org/web/20150117040022/http://article.gmane.org/gmane.linux.kernel.wireless.general/38410
[3]
https://patchwork.kernel.org/project/ath10k/patch/[email protected]/
[4] https://fccid.io/TK4WLE900VX
_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k

Hello, this is not correct...

I have such a card with 0x0 in the eeprom
due to the patch the complete midband in station mode is missing So the card can't see AP's in midband or connect to AP's either....

why are the cards mapped to 0x64 and not 0x6c?

https://bugs.launchpad.net/ubuntu/+source/linux-firmware/+bug/1895333
https://www.mail-archive.com/[email protected]/msg13040.html

the patch has made the card unusable for me and I  had to deal extensively with the thematics ...

and yes my card also has a FCC ID and it would be new to me that there is no midband in the USA...

 best regards

(sorry for breaking the list)

_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k