| Message ID | 1470596269-20572-2-git-send-email-hdegoede@redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Sunday, August 7, 2016 8:57:49 PM CEST Hans de Goede wrote: > +Espressif ESP8089 wireless SDIO devices > + > +This node provides properties for controlling the ESP8089 wireless device. > +The node is expected to be specified as a child node to the SDIO controller > +that connects the device to the system. > + > +Required properties: > + > + - compatible : Should be "esp,esp8089". I think it would be good to standardize a compatible string for sdio like we have it for other discoverable buses, based on the vendor/device ID (or whatever this uses), using something like compatible = "sdio1234,5678", "esp,esp8089"; > +Optional properties: > + - esp,crystal_26M_en: Integer value for the crystal_26M_en firmware parameter better use '-' instead of '_' for property names. Arnd
08.08.2016, 02:58, "Hans de Goede" <hdegoede@redhat.com>: > The ESP8089 chips can mostly be enumerated via their sdio interface, > but they are clocked by an external crystal which may differ from one > board to the other. > > This commit adds a binding for the sdio child node for these chips, > allowing to specify the external crystal type (for now, this binding > could be be extended with e.g. OOB irq support later). > > The Android driver for this chip uses a text file with key,value pairs > which gets loaded as firmware to pass this info to the firmware. > The "esp,crystal_26M_en" name is chosen to match the crystal_26M_en > key-name in that text file. > > Note that at this point there only is an out of tree driver for this > hardware, there is no clear timeline / path for merging this. Still > I believe it would be good to specify the binding for this in tree > now, so that any future migration to an in tree driver will not cause > compatiblity issues. > > Cc: Icenowy Zheng <icenowy@aosc.xyz> > Signed-off-by: Hans de Goede <hdegoede@redhat.com> > --- > .../bindings/net/wireless/esp,esp8089.txt | 31 ++++++++++++++++++++++ > 1 file changed, 31 insertions(+) > create mode 100644 Documentation/devicetree/bindings/net/wireless/esp,esp8089.txt > > diff --git a/Documentation/devicetree/bindings/net/wireless/esp,esp8089.txt b/Documentation/devicetree/bindings/net/wireless/esp,esp8089.txt > new file mode 100644 > index 0000000..898a149 > --- /dev/null > +++ b/Documentation/devicetree/bindings/net/wireless/esp,esp8089.txt > @@ -0,0 +1,31 @@ > +Espressif ESP8089 wireless SDIO devices > + > +This node provides properties for controlling the ESP8089 wireless device. > +The node is expected to be specified as a child node to the SDIO controller > +that connects the device to the system. > + > +Required properties: > + > + - compatible : Should be "esp,esp8089". There's also esp8266. (And the difference of 8089 and 8266 is mainly the difference of crystal frequency... 26MHz for 8266, 40MHz for 8089) So maybe it should be the difference of dt compatible? > + > +Optional properties: > + - esp,crystal_26M_en: Integer value for the crystal_26M_en firmware parameter > + > +Example: > + > +&mmc1 { > + #address-cells = <1>; > + #size-cells = <0>; > + > + vmmc-supply = <®_dldo1>; > + mmc-pwrseq = <&wifi_pwrseq>; > + bus-width = <4>; > + non-removable; > + status = "okay"; > + > + esp8089: sdio_wifi@1 { > + compatible = "esp,esp8089"; > + reg = <1>; > + esp,crystal_26M_en = <2>; > + }; > +}; > -- > 2.7.4
Hi Arnd, On Mon, Aug 8, 2016 at 3:59 AM, Arnd Bergmann <arnd@arndb.de> wrote: > On Sunday, August 7, 2016 8:57:49 PM CEST Hans de Goede wrote: >> +Espressif ESP8089 wireless SDIO devices >> + >> +This node provides properties for controlling the ESP8089 wireless device. >> +The node is expected to be specified as a child node to the SDIO controller >> +that connects the device to the system. >> + >> +Required properties: >> + >> + - compatible : Should be "esp,esp8089". > > I think it would be good to standardize a compatible string for sdio > like we have it for other discoverable buses, based on the > vendor/device ID (or whatever this uses), using something like > > compatible = "sdio1234,5678", "esp,esp8089"; Is there a central repository for SDIO vendor/device IDs similar to PCI or USB IDs? The vendor ID Espressif uses is 0x6666, which seems kind of bogus to me. ChenYu > >> +Optional properties: >> + - esp,crystal_26M_en: Integer value for the crystal_26M_en firmware parameter > > better use '-' instead of '_' for property names. > > Arnd > >
Hi, On 08-08-16 04:36, Icenowy Zheng wrote: > > > 08.08.2016, 02:58, "Hans de Goede" <hdegoede@redhat.com>: >> The ESP8089 chips can mostly be enumerated via their sdio interface, >> but they are clocked by an external crystal which may differ from one >> board to the other. >> >> This commit adds a binding for the sdio child node for these chips, >> allowing to specify the external crystal type (for now, this binding >> could be be extended with e.g. OOB irq support later). >> >> The Android driver for this chip uses a text file with key,value pairs >> which gets loaded as firmware to pass this info to the firmware. >> The "esp,crystal_26M_en" name is chosen to match the crystal_26M_en >> key-name in that text file. >> >> Note that at this point there only is an out of tree driver for this >> hardware, there is no clear timeline / path for merging this. Still >> I believe it would be good to specify the binding for this in tree >> now, so that any future migration to an in tree driver will not cause >> compatiblity issues. >> >> Cc: Icenowy Zheng <icenowy@aosc.xyz> >> Signed-off-by: Hans de Goede <hdegoede@redhat.com> >> --- >> .../bindings/net/wireless/esp,esp8089.txt | 31 ++++++++++++++++++++++ >> 1 file changed, 31 insertions(+) >> create mode 100644 Documentation/devicetree/bindings/net/wireless/esp,esp8089.txt >> >> diff --git a/Documentation/devicetree/bindings/net/wireless/esp,esp8089.txt b/Documentation/devicetree/bindings/net/wireless/esp,esp8089.txt >> new file mode 100644 >> index 0000000..898a149 >> --- /dev/null >> +++ b/Documentation/devicetree/bindings/net/wireless/esp,esp8089.txt >> @@ -0,0 +1,31 @@ >> +Espressif ESP8089 wireless SDIO devices >> + >> +This node provides properties for controlling the ESP8089 wireless device. >> +The node is expected to be specified as a child node to the SDIO controller >> +that connects the device to the system. >> + >> +Required properties: >> + >> + - compatible : Should be "esp,esp8089". > There's also esp8266. (And the difference of 8089 and 8266 is mainly the difference of crystal frequency... 26MHz for 8266, 40MHz for 8089) > So maybe it should be the difference of dt compatible? 2 of my tablets (both Polaroid ones) use a value of 2 (rather then 0 or 1) for crystal_26M_en and using a different value does not work. Another generic q8 form-factor tablet I've with an a23 needs crystal_26M_en=1. In all these tablets the chip is clearly marked esp8089, so the crystal_26M_en setting seems to not be related to the chip model and deriving this from the compatible string seems like a bad idea to me. Regards, Hans >> + >> +Optional properties: >> + - esp,crystal_26M_en: Integer value for the crystal_26M_en firmware parameter >> + >> +Example: >> + >> +&mmc1 { >> + #address-cells = <1>; >> + #size-cells = <0>; >> + >> + vmmc-supply = <®_dldo1>; >> + mmc-pwrseq = <&wifi_pwrseq>; >> + bus-width = <4>; >> + non-removable; >> + status = "okay"; >> + >> + esp8089: sdio_wifi@1 { >> + compatible = "esp,esp8089"; >> + reg = <1>; >> + esp,crystal_26M_en = <2>; >> + }; >> +}; >> -- >> 2.7.4
On Monday, August 8, 2016 11:59:25 AM CEST Chen-Yu Tsai wrote: > Hi Arnd, > > On Mon, Aug 8, 2016 at 3:59 AM, Arnd Bergmann <arnd@arndb.de> wrote: > > On Sunday, August 7, 2016 8:57:49 PM CEST Hans de Goede wrote: > >> +Espressif ESP8089 wireless SDIO devices > >> + > >> +This node provides properties for controlling the ESP8089 wireless device. > >> +The node is expected to be specified as a child node to the SDIO controller > >> +that connects the device to the system. > >> + > >> +Required properties: > >> + > >> + - compatible : Should be "esp,esp8089". > > > > I think it would be good to standardize a compatible string for sdio > > like we have it for other discoverable buses, based on the > > vendor/device ID (or whatever this uses), using something like > > > > compatible = "sdio1234,5678", "esp,esp8089"; > > Is there a central repository for SDIO vendor/device IDs similar to > PCI or USB IDs? No idea. I believe the existing repositories for PCI and USB are not public either, there is just the information at http://pcidatabase.com/, http://pci-ids.ucw.cz/ or http://www.linux-usb.org/usb.ids that are maintained by volunteers and that are not authoritarive. I found one list at https://github.com/systemd/systemd/blob/master/hwdb/sdio.ids but it's probably very incomplete. > The vendor ID Espressif uses is 0x6666, which seems kind of bogus to me. It probably is, but that's not important as long as nothing else uses that ID. Arnd
Hi, 08.08.2016, 15:34, "Hans de Goede" <hdegoede@redhat.com>: > Hi, > > On 08-08-16 04:36, Icenowy Zheng wrote: >> 08.08.2016, 02:58, "Hans de Goede" <hdegoede@redhat.com>: >>> The ESP8089 chips can mostly be enumerated via their sdio interface, >>> but they are clocked by an external crystal which may differ from one >>> board to the other. >>> >>> This commit adds a binding for the sdio child node for these chips, >>> allowing to specify the external crystal type (for now, this binding >>> could be be extended with e.g. OOB irq support later). >>> >>> The Android driver for this chip uses a text file with key,value pairs >>> which gets loaded as firmware to pass this info to the firmware. >>> The "esp,crystal_26M_en" name is chosen to match the crystal_26M_en >>> key-name in that text file. >>> >>> Note that at this point there only is an out of tree driver for this >>> hardware, there is no clear timeline / path for merging this. Still >>> I believe it would be good to specify the binding for this in tree >>> now, so that any future migration to an in tree driver will not cause >>> compatiblity issues. >>> >>> Cc: Icenowy Zheng <icenowy@aosc.xyz> >>> Signed-off-by: Hans de Goede <hdegoede@redhat.com> >>> --- >>> .../bindings/net/wireless/esp,esp8089.txt | 31 ++++++++++++++++++++++ >>> 1 file changed, 31 insertions(+) >>> create mode 100644 Documentation/devicetree/bindings/net/wireless/esp,esp8089.txt >>> >>> diff --git a/Documentation/devicetree/bindings/net/wireless/esp,esp8089.txt b/Documentation/devicetree/bindings/net/wireless/esp,esp8089.txt >>> new file mode 100644 >>> index 0000000..898a149 >>> --- /dev/null >>> +++ b/Documentation/devicetree/bindings/net/wireless/esp,esp8089.txt >>> @@ -0,0 +1,31 @@ >>> +Espressif ESP8089 wireless SDIO devices >>> + >>> +This node provides properties for controlling the ESP8089 wireless device. >>> +The node is expected to be specified as a child node to the SDIO controller >>> +that connects the device to the system. >>> + >>> +Required properties: >>> + >>> + - compatible : Should be "esp,esp8089". >> There's also esp8266. (And the difference of 8089 and 8266 is mainly the difference of crystal frequency... 26MHz for 8266, 40MHz for 8089) >> So maybe it should be the difference of dt compatible? > > 2 of my tablets (both Polaroid ones) use a value of 2 (rather > then 0 or 1) for crystal_26M_en and using a different value > does not work. Another generic q8 form-factor tablet I've with > an a23 needs crystal_26M_en=1. In all these tablets the chip is > clearly marked esp8089, so the crystal_26M_en setting seems to > not be related to the chip model and deriving this from the > compatible string seems like a bad idea to me. Oh such strange. Thanks, Icenowy > > Regards, > > Hans > >>> + >>> +Optional properties: >>> + - esp,crystal_26M_en: Integer value for the crystal_26M_en firmware parameter >>> + >>> +Example: >>> + >>> +&mmc1 { >>> + #address-cells = <1>; >>> + #size-cells = <0>; >>> + >>> + vmmc-supply = <®_dldo1>; >>> + mmc-pwrseq = <&wifi_pwrseq>; >>> + bus-width = <4>; >>> + non-removable; >>> + status = "okay"; >>> + >>> + esp8089: sdio_wifi@1 { >>> + compatible = "esp,esp8089"; >>> + reg = <1>; >>> + esp,crystal_26M_en = <2>; >>> + }; >>> +}; >>> -- >>> 2.7.4
Hi, 08.08.2016, 17:30, "Arnd Bergmann" <arnd@arndb.de>: > On Monday, August 8, 2016 11:59:25 AM CEST Chen-Yu Tsai wrote: >> Hi Arnd, >> >> On Mon, Aug 8, 2016 at 3:59 AM, Arnd Bergmann <arnd@arndb.de> wrote: >> > On Sunday, August 7, 2016 8:57:49 PM CEST Hans de Goede wrote: >> >> +Espressif ESP8089 wireless SDIO devices >> >> + >> >> +This node provides properties for controlling the ESP8089 wireless device. >> >> +The node is expected to be specified as a child node to the SDIO controller >> >> +that connects the device to the system. >> >> + >> >> +Required properties: >> >> + >> >> + - compatible : Should be "esp,esp8089". >> > >> > I think it would be good to standardize a compatible string for sdio >> > like we have it for other discoverable buses, based on the >> > vendor/device ID (or whatever this uses), using something like >> > >> > compatible = "sdio1234,5678", "esp,esp8089"; >> >> Is there a central repository for SDIO vendor/device IDs similar to >> PCI or USB IDs? > > No idea. I believe the existing repositories for PCI and USB are not > public either, there is just the information at http://pcidatabase.com/, > http://pci-ids.ucw.cz/ or http://www.linux-usb.org/usb.ids that > are maintained by volunteers and that are not authoritarive. > > I found one list at https://github.com/systemd/systemd/blob/master/hwdb/sdio.ids > but it's probably very incomplete. Maybe it should be maintained as a independent project... > >> The vendor ID Espressif uses is 0x6666, which seems kind of bogus to me. > > It probably is, but that's not important as long as nothing else uses > that ID. Yes... I think the USB vendor ID of Allwinner, 0x1f3a, is also nothing authoritative. > > Arnd Thanks, Icenowy
diff --git a/Documentation/devicetree/bindings/net/wireless/esp,esp8089.txt b/Documentation/devicetree/bindings/net/wireless/esp,esp8089.txt new file mode 100644 index 0000000..898a149 --- /dev/null +++ b/Documentation/devicetree/bindings/net/wireless/esp,esp8089.txt @@ -0,0 +1,31 @@ +Espressif ESP8089 wireless SDIO devices + +This node provides properties for controlling the ESP8089 wireless device. +The node is expected to be specified as a child node to the SDIO controller +that connects the device to the system. + +Required properties: + + - compatible : Should be "esp,esp8089". + +Optional properties: + - esp,crystal_26M_en: Integer value for the crystal_26M_en firmware parameter + +Example: + +&mmc1 { + #address-cells = <1>; + #size-cells = <0>; + + vmmc-supply = <®_dldo1>; + mmc-pwrseq = <&wifi_pwrseq>; + bus-width = <4>; + non-removable; + status = "okay"; + + esp8089: sdio_wifi@1 { + compatible = "esp,esp8089"; + reg = <1>; + esp,crystal_26M_en = <2>; + }; +};
The ESP8089 chips can mostly be enumerated via their sdio interface, but they are clocked by an external crystal which may differ from one board to the other. This commit adds a binding for the sdio child node for these chips, allowing to specify the external crystal type (for now, this binding could be be extended with e.g. OOB irq support later). The Android driver for this chip uses a text file with key,value pairs which gets loaded as firmware to pass this info to the firmware. The "esp,crystal_26M_en" name is chosen to match the crystal_26M_en key-name in that text file. Note that at this point there only is an out of tree driver for this hardware, there is no clear timeline / path for merging this. Still I believe it would be good to specify the binding for this in tree now, so that any future migration to an in tree driver will not cause compatiblity issues. Cc: Icenowy Zheng <icenowy@aosc.xyz> Signed-off-by: Hans de Goede <hdegoede@redhat.com> --- .../bindings/net/wireless/esp,esp8089.txt | 31 ++++++++++++++++++++++ 1 file changed, 31 insertions(+) create mode 100644 Documentation/devicetree/bindings/net/wireless/esp,esp8089.txt