| Message ID | 20160525173153.GA4219@makrotopia.org |
|---|---|
| State | Accepted |
| Headers | show |
On Wed, 25 May 2016 19:32:01 +0200 Daniel Golle <daniel@makrotopia.org> wrote: > It turns out most device vendors don't set the correct country code > in their devices' on-flash-EEPROM sections as they apparently rather > provide a complete per-target-market firmware with patched drivers > instead of just setting the country code. > This results in the driver to incorrectly assume the value stored in > the on-flash-EERPOM (usually US or China) being the regulatory domain > inside which the device is being used. > To work around this issue, OpenWrt introduced the ATH_USER_REGD config > variable to decide during build whether or not to allow the user to > override the regulatory domain setting. This option, however, is not > enabled by default and thus ends up being disabled for snapshots > builds and released binaries. > As we know for a long time that most devices got borked regulatory > domain values set in their EEPROMs we should allow our users to > respect their local law (instead of just assume US or China laws). > Note that also the current default has great potential of users not > ever setting their regulatory domain and thus using inapproriate and > potentially illegal frequencies and/or tx-power settings > > Signed-off-by: Daniel Golle <daniel@makrotopia.org> > --- > Imho we should set World regulatory domain by default but also allow > users to set their actual regulatory domain (e.g. ETSI). > Obviously this will require the user to enter her country before being > able to use AP, IBSS and 802.11s mode in the 5 GHz bands. > Most wireless mesh communities' firmware projects are currently > implemented as source-wrappers around OpenWrt instead of just using > the SDK and ImageBuilder because of the official releases being > useless in most places in the world due to ATH_USER_REGD being > disabled. I really like this commit. Can you set the country *explicit* to world in /etc/config/wireless before appling this patch? best, lynxis
On Thu, 2016-05-26 at 02:57 +0200, Alexander Couzens wrote: > On Wed, 25 May 2016 19:32:01 +0200 > Daniel Golle <daniel@makrotopia.org> wrote: > > I really like this commit. > Can you set the country *explicit* to world in /etc/config/wireless > before appling this patch? > If the FCC hadn't already completely knobbish and basically eliminated (or least required one to be much more determined and skilled) the ability to use OpenWrt/LEDE on new routers in the US anyway, I might have some objection (I'm not in the US but Canada will almost certainly adopt similar legislation since our government tends to do what the US wants, putzes that they are) on the grounds of leading to trouble. As it is, for new wireless routers, the US, and probably soon Canada, are shut out from participating in OpenWrt/LEDE anyway, it's pretty much moot point for those of us here, and in sane parts of the world it makes sense. Regards, Daniel
Hi Daniel! On Wed, May 25, 2016 at 09:29:23PM -0400, Daniel Dickinson wrote: > ... > If the FCC hadn't already completely knobbish and basically eliminated > (or least required one to be much more determined and skilled) the > ability to use OpenWrt/LEDE on new routers in the US anyway, I might > have some objection (I'm not in the US but Canada will almost certainly > adopt similar legislation since our government tends to do what the US > wants, putzes that they are) on the grounds of leading to trouble. > > As it is, for new wireless routers, the US, and probably soon Canada, > are shut out from participating in OpenWrt/LEDE anyway, it's pretty much > moot point for those of us here, and in sane parts of the world it makes > sense. As we have learned at BattleMesh, this whole lockdown madness was actually started in some back-office of the European Comission back in 2014 without anyone of us noticing. As usually, if things like that happend, we Europeans are quick to blame others eventhough it was out own governments making that decission and then encouraging the FCC to do the same on the other side of the Atlantic. Independently of that whole issue, I believe we should always try our bests to make it as easy as possible for users to obey to radio regulations. One improvement could be to make sure the user enters her country code before starting to use the firmware. Currently, users outside of the US may just enable the 5 GHz radio and without ever being asked, the device will just assume FCC regulations and most likely use frequencies and/or transmit power inappropriate to the region it is actually being used. Cheers Daniel
On Thu, 26 May 2016, Daniel Golle wrote: > Hi Daniel! > > On Wed, May 25, 2016 at 09:29:23PM -0400, Daniel Dickinson wrote: >> ... >> If the FCC hadn't already completely knobbish and basically eliminated >> (or least required one to be much more determined and skilled) the >> ability to use OpenWrt/LEDE on new routers in the US anyway, I might >> have some objection (I'm not in the US but Canada will almost certainly >> adopt similar legislation since our government tends to do what the US >> wants, putzes that they are) on the grounds of leading to trouble. >> >> As it is, for new wireless routers, the US, and probably soon Canada, >> are shut out from participating in OpenWrt/LEDE anyway, it's pretty much >> moot point for those of us here, and in sane parts of the world it makes >> sense. > > As we have learned at BattleMesh, this whole lockdown madness was > actually started in some back-office of the European Comission back > in 2014 without anyone of us noticing. As usually, if things like > that happend, we Europeans are quick to blame others eventhough it > was out own governments making that decission and then encouraging > the FCC to do the same on the other side of the Atlantic. It's also important to note that the FCC has not passed any rules requiring a lockdown. The posted some proposed rules that would require a lockdown, but after the backlash have made public statements that they do not intend to block OpenWRT or similar (and they started the 'free the settop box' project to break existing cable lockdowns) Linksys would not be able to say that they are not going to lock down their devices if the FCC actually required it. Back on topic, I think it's a good thing to default to world-wide and allow the user to set the proper country (they should not need to know what the codes are) be sure to have the appropriate dire warnings about deliberatly setting it incorrectly and the trouble you could get into :-) David Lang AG6AH
Hi David, On Thu, May 26, 2016 at 02:51:17AM -0700, David Lang wrote: > On Thu, 26 May 2016, Daniel Golle wrote: > > > Hi Daniel! > > > > On Wed, May 25, 2016 at 09:29:23PM -0400, Daniel Dickinson wrote: > > > ... > > > If the FCC hadn't already completely knobbish and basically eliminated > > > (or least required one to be much more determined and skilled) the > > > ability to use OpenWrt/LEDE on new routers in the US anyway, I might > > > have some objection (I'm not in the US but Canada will almost certainly > > > adopt similar legislation since our government tends to do what the US > > > wants, putzes that they are) on the grounds of leading to trouble. > > > > > > As it is, for new wireless routers, the US, and probably soon Canada, > > > are shut out from participating in OpenWrt/LEDE anyway, it's pretty much > > > moot point for those of us here, and in sane parts of the world it makes > > > sense. > > > > As we have learned at BattleMesh, this whole lockdown madness was > > actually started in some back-office of the European Comission back > > in 2014 without anyone of us noticing. As usually, if things like > > that happend, we Europeans are quick to blame others eventhough it > > was out own governments making that decission and then encouraging > > the FCC to do the same on the other side of the Atlantic. > > It's also important to note that the FCC has not passed any rules requiring > a lockdown. The posted some proposed rules that would require a lockdown, > but after the backlash have made public statements that they do not intend > to block OpenWRT or similar (and they started the 'free the settop box' > project to break existing cable lockdowns) > > Linksys would not be able to say that they are not going to lock down their > devices if the FCC actually required it. Well, Linksys (and others using mv88w8864 or ath10k) can opt for really only locking the radio firmware and thereby create a niche-market for 'open-source-friendly' routers. However, this is obviously not possible for ath9k and might be hard or even impossible to achieve using other pure softmac (ie. firmware-less) chips as well. > > > Back on topic, I think it's a good thing to default to world-wide and allow > the user to set the proper country (they should not need to know what the > codes are) Ack. Imho the ISO country code is the right thing to store in UCI, (translated) country names should be exposed to users on the web-interface, but that's really only on the view-level. It also doesn't make much sense to store the country-code for each radio, it's a system-wide property and should thus go into /etc/config/system (opinions?). > > be sure to have the appropriate dire warnings about deliberatly setting it > incorrectly and the trouble you could get into :-) Obviously there should be warnings in the documentation and web- interface as well as comments in the configuration file. Cheers Daniel > > David Lang > AG6AH
diff --git a/package/kernel/mac80211/Makefile b/package/kernel/mac80211/Makefile index 8b053d4..28da0ec 100644 --- a/package/kernel/mac80211/Makefile +++ b/package/kernel/mac80211/Makefile @@ -161,6 +161,7 @@ define KernelPackage/ath/config if PACKAGE_kmod-ath config ATH_USER_REGD bool "Force Atheros drivers to respect the user's regdomain settings" + default y help Atheros' idea of regulatory handling is that the EEPROM of the card defines the regulatory limits and the user is only allowed to restrict the settings
It turns out most device vendors don't set the correct country code in their devices' on-flash-EEPROM sections as they apparently rather provide a complete per-target-market firmware with patched drivers instead of just setting the country code. This results in the driver to incorrectly assume the value stored in the on-flash-EERPOM (usually US or China) being the regulatory domain inside which the device is being used. To work around this issue, OpenWrt introduced the ATH_USER_REGD config variable to decide during build whether or not to allow the user to override the regulatory domain setting. This option, however, is not enabled by default and thus ends up being disabled for snapshots builds and released binaries. As we know for a long time that most devices got borked regulatory domain values set in their EEPROMs we should allow our users to respect their local law (instead of just assume US or China laws). Note that also the current default has great potential of users not ever setting their regulatory domain and thus using inapproriate and potentially illegal frequencies and/or tx-power settings Signed-off-by: Daniel Golle <daniel@makrotopia.org> --- Imho we should set World regulatory domain by default but also allow users to set their actual regulatory domain (e.g. ETSI). Obviously this will require the user to enter her country before being able to use AP, IBSS and 802.11s mode in the 5 GHz bands. Most wireless mesh communities' firmware projects are currently implemented as source-wrappers around OpenWrt instead of just using the SDK and ImageBuilder because of the official releases being useless in most places in the world due to ATH_USER_REGD being disabled. package/kernel/mac80211/Makefile | 1 + 1 file changed, 1 insertion(+)