diff mbox

[1/1] Add Wondermedia Technologies (wmt) to docs/devicetree/vendor-bindings.txt

Message ID 76F764B079F92A4E843589C893D0A022C380F1D4@SERVER.prisktech.co.nz (mailing list archive)
State New, archived
Headers show

Commit Message

Tony Prisk July 17, 2012, 7:11 a.m. UTC
From 756ae5c2f475d679649adff99058679b651af8d9 Mon Sep 17 00:00:00 2001
From: Tony Prisk <linux@prisktech.co.nz>
Date: Tue, 17 Jul 2012 19:09:31 +1200
Subject: [PATCH] Add Wondermedia Technologies (wmt) to vendor-prefixes.txt
 Signed-off-by: Tony Prisk <linux@prisktech.co.nz>

---
 .../devicetree/bindings/vendor-prefixes.txt        |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

Comments

Arnd Bergmann July 17, 2012, 12:50 p.m. UTC | #1
On Tuesday 17 July 2012, Tony Prisk wrote:
> From 756ae5c2f475d679649adff99058679b651af8d9 Mon Sep 17 00:00:00 2001
> From: Tony Prisk <linux@prisktech.co.nz>
> Date: Tue, 17 Jul 2012 19:09:31 +1200
> Subject: [PATCH] Add Wondermedia Technologies (wmt) to vendor-prefixes.txt
>  Signed-off-by: Tony Prisk <linux@prisktech.co.nz>
 
Something went wrong with exporting the patch. You probably used git-show
rather than git-format-patch, so you got the wrong headers.

It would also be helpful to describe why the specific string is used here.
Normally we use the stock ticker symbol and "wmt" refers to Wal-Mart,
which would be unlikely but not impossible to produce their own hardware
in the future. The same thing applies to "via", which is VIACOM. The symbol
for VIA technologies is "2388.TW", but that is not particularly descriptive
or well-known.

Maybe someone on the devicetree-discuss mailing list has a better
suggestion. I think using "via" is probably the best compromise, as
it's a well-known name and most of the components of the wondermedia
chips were first made by VIA anyway.

	Arnd
Alexey Charkov July 17, 2012, 1:30 p.m. UTC | #2
2012/7/17 Arnd Bergmann <arnd@arndb.de>:
> It would also be helpful to describe why the specific string is used here.
> Normally we use the stock ticker symbol and "wmt" refers to Wal-Mart,
> which would be unlikely but not impossible to produce their own hardware
> in the future. The same thing applies to "via", which is VIACOM. The symbol
> for VIA technologies is "2388.TW", but that is not particularly descriptive
> or well-known.
>
> Maybe someone on the devicetree-discuss mailing list has a better
> suggestion. I think using "via" is probably the best compromise, as
> it's a well-known name and most of the components of the wondermedia
> chips were first made by VIA anyway.

In fact, "wmt" is the name that the vendor itself uses in their GPL
code (they also use mach-wmt as the arch-specific folder name). I
believe that we could just as well use the full "wondermedia" word to
avoid possible confusion, in case "wmt" is unacceptable for any
reason.

Best,
Alexey
Tony Prisk July 17, 2012, 1:47 p.m. UTC | #3
Arnd,

I don't know what has happened with the patch(es).
I used git format-patch to export them. I had to copy them to another PC to send the emails, so perhaps that's the problem.

Re: 'wmt' - as Alexey mentioned, this is the name the vendor uses in their own sources so it seemed like the obvious choice. I will post on devicetree-discuss and see what they think.

Regard,
Tony P
Tony Prisk July 19, 2012, 4:56 a.m. UTC | #4
Given that there have been no replies to this thread, and the restart patch has
now been accepted using 'wmt' as the vendor namespace - would it be safe to
assume we should formalize this in the vendor-bindings.txt?

I'd like to try and have basic device tree support added for the next merge
window (or this one if it all goes well :) ), but it relies on the vendor binding being
formalized.

Regards,

Tony P

> -----Original Message-----
> From: linux-arm-kernel-bounces@lists.infradead.org [mailto:linux-arm-kernel-bounces@lists.infradead.org] On Behalf Of Alexey
> Charkov
> Sent: Wednesday, 18 July 2012 1:31 a.m.
> To: vt8500-wm8505-linux-kernel@googlegroups.com
> Cc: Tony Prisk; devicetree-discuss@lists.ozlabs.org; Russell King; linux-kernel@vger.kernel.org; linux-arm-kernel@lists.infradead.org
> Subject: Re: [PATCH 1/1] Add Wondermedia Technologies (wmt) to docs/devicetree/vendor-bindings.txt
> 
> 2012/7/17 Arnd Bergmann <arnd@arndb.de>:
> > It would also be helpful to describe why the specific string is used here.
> > Normally we use the stock ticker symbol and "wmt" refers to Wal-Mart,
> > which would be unlikely but not impossible to produce their own hardware
> > in the future. The same thing applies to "via", which is VIACOM. The symbol
> > for VIA technologies is "2388.TW", but that is not particularly descriptive
> > or well-known.
> >
> > Maybe someone on the devicetree-discuss mailing list has a better
> > suggestion. I think using "via" is probably the best compromise, as
> > it's a well-known name and most of the components of the wondermedia
> > chips were first made by VIA anyway.
> 
> In fact, "wmt" is the name that the vendor itself uses in their GPL
> code (they also use mach-wmt as the arch-specific folder name). I
> believe that we could just as well use the full "wondermedia" word to
> avoid possible confusion, in case "wmt" is unacceptable for any
> reason.
> 
> Best,
> Alexey
> 
> _______________________________________________
> linux-arm-kernel mailing list
> linux-arm-kernel@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Arnd Bergmann July 19, 2012, 11:16 a.m. UTC | #5
On Thursday 19 July 2012, Tony Prisk wrote:
> Given that there have been no replies to this thread, and the restart patch has
> now been accepted using 'wmt' as the vendor namespace - would it be safe to
> assume we should formalize this in the vendor-bindings.txt?
> 
> I'd like to try and have basic device tree support added for the next merge
> window (or this one if it all goes well :) ), but it relies on the vendor binding being
> formalized.

What about devices that come from via though? Anything that first showed up
in vt8500 or earlier should probably get a prefix identifying VIA rather
than WonderMedia.

	Arnd
Tony Prisk July 19, 2012, 11:45 a.m. UTC | #6
From: Arnd Bergmann [arnd@arndb.de]
Sent: 19 July 2012 23:16
To: vt8500-wm8505-linux-kernel@googlegroups.com
Cc: Tony Prisk; Alexey Charkov; devicetree-discuss@lists.ozlabs.org; Russell King; linux-kernel@vger.kernel.org; linux-arm-kernel@lists.infradead.org
Subject: Re: [PATCH 1/1] Add Wondermedia Technologies (wmt) to docs/devicetree/vendor-bindings.txt

On Thursday 19 July 2012, Tony Prisk wrote:
> Given that there have been no replies to this thread, and the restart patch has
> now been accepted using 'wmt' as the vendor namespace - would it be safe to
> assume we should formalize this in the vendor-bindings.txt?
>
> I'd like to try and have basic device tree support added for the next merge
> window (or this one if it all goes well :) ), but it relies on the vendor binding being
> formalized.

What about devices that come from via though? Anything that first showed up
in vt8500 or earlier should probably get a prefix identifying VIA rather
than WonderMedia.

        Arnd


I have no issue with that.

I am mainly concerned with the Wondermedia products at the moment.

Adding device tree support for WM8505 (and WM8650) is my main goal as the WM8650
currently isn't supported and trying to get another board file added seems unlikely with
everything moving to device tree. The VT8500 requires additional/different patches and
as I don't have a vt8500 I can't test them.

It does raise the question of applying 'wmt, prizm-pmc' to the vt8500 though. While it's
fine for the restart patch I submitted, the vt8500 registers are the same but the bit-fields
are different and therefore not entirely compatible with 'prizm-pmc'.
When it comes to other PMC-related features, in particular the clocks, the VT8500 is quite
different in its bit-fields.

The vt8500 isn't technically in the Prizm family, as that is Wondermedia's name for the SoC's.


Regard

Tony P
Alexey Charkov July 19, 2012, 12:12 p.m. UTC | #7
2012/7/19 Tony Prisk <linux@prisktech.co.nz>:
> From: Arnd Bergmann [arnd@arndb.de]
> Sent: 19 July 2012 23:16
> To: vt8500-wm8505-linux-kernel@googlegroups.com
> Cc: Tony Prisk; Alexey Charkov; devicetree-discuss@lists.ozlabs.org; Russell King; linux-kernel@vger.kernel.org; linux-arm-kernel@lists.infradead.org
> Subject: Re: [PATCH 1/1] Add Wondermedia Technologies (wmt) to docs/devicetree/vendor-bindings.txt
>
> On Thursday 19 July 2012, Tony Prisk wrote:
>> Given that there have been no replies to this thread, and the restart patch has
>> now been accepted using 'wmt' as the vendor namespace - would it be safe to
>> assume we should formalize this in the vendor-bindings.txt?
>>
>> I'd like to try and have basic device tree support added for the next merge
>> window (or this one if it all goes well :) ), but it relies on the vendor binding being
>> formalized.
>
> What about devices that come from via though? Anything that first showed up
> in vt8500 or earlier should probably get a prefix identifying VIA rather
> than WonderMedia.
>
>         Arnd
>
>
> I have no issue with that.
>
> I am mainly concerned with the Wondermedia products at the moment.
>
> Adding device tree support for WM8505 (and WM8650) is my main goal as the WM8650
> currently isn't supported and trying to get another board file added seems unlikely with
> everything moving to device tree. The VT8500 requires additional/different patches and
> as I don't have a vt8500 I can't test them.

I believe that Arnd's point was about the hardware that WonderMedia
chips have inherited from their VIA ancestors, like NAND, SD/MMC,
graphics engine, USB EHCI glue, PWM, PS/2 controller, parts of PMC
etc. Those would be more appropriately named with something
VIA-specific rather than WMT as the vendor prefix (this would apply
for the PWM and the raster engine, which are unchanged between VT8500
and WM8505, to my knowledge).

On the other hand, all of these parts might well be some kind of
licensed third-party design anyway, which makes this whole VIA vs. WMT
a bikeshed discussion mostly. I would personally not feel too guilty
about calling some device that first appeared in a VT8xxx chip 'wmt'
until there's technical reason for different 'compatible' strings
matching different chip versions (VIA age vs. WMT age). So far, we
don't have that need.

Best,
Alexey
Tony Prisk July 19, 2012, 12:19 p.m. UTC | #8
From: Alexey Charkov [alchark@gmail.com]
Sent: 20 July 2012 00:12
To: Tony Prisk
Cc: Arnd Bergmann; vt8500-wm8505-linux-kernel@googlegroups.com; devicetree-discuss@lists.ozlabs.org; Russell King; linux-kernel@vger.kernel.org; linux-arm-kernel@lists.infradead.org
Subject: Re: [PATCH 1/1] Add Wondermedia Technologies (wmt) to docs/devicetree/vendor-bindings.txt

2012/7/19 Tony Prisk <linux@prisktech.co.nz>:
> From: Arnd Bergmann [arnd@arndb.de]
> Sent: 19 July 2012 23:16
> To: vt8500-wm8505-linux-kernel@googlegroups.com
> Cc: Tony Prisk; Alexey Charkov; devicetree-discuss@lists.ozlabs.org; Russell King; linux-kernel@vger.kernel.org; linux-arm-kernel@lists.infradead.org
> Subject: Re: [PATCH 1/1] Add Wondermedia Technologies (wmt) to docs/devicetree/vendor-bindings.txt
>
> On Thursday 19 July 2012, Tony Prisk wrote:
>> Given that there have been no replies to this thread, and the restart patch has
>> now been accepted using 'wmt' as the vendor namespace - would it be safe to
>> assume we should formalize this in the vendor-bindings.txt?
>>
>> I'd like to try and have basic device tree support added for the next merge
>> window (or this one if it all goes well :) ), but it relies on the vendor binding being
>> formalized.
>
> What about devices that come from via though? Anything that first showed up
> in vt8500 or earlier should probably get a prefix identifying VIA rather
> than WonderMedia.
>
>         Arnd
>
>
> I have no issue with that.
>
> I am mainly concerned with the Wondermedia products at the moment.
>
> Adding device tree support for WM8505 (and WM8650) is my main goal as the WM8650
> currently isn't supported and trying to get another board file added seems unlikely with
> everything moving to device tree. The VT8500 requires additional/different patches and
> as I don't have a vt8500 I can't test them.

I believe that Arnd's point was about the hardware that WonderMedia
chips have inherited from their VIA ancestors, like NAND, SD/MMC,
graphics engine, USB EHCI glue, PWM, PS/2 controller, parts of PMC
etc. Those would be more appropriately named with something
VIA-specific rather than WMT as the vendor prefix (this would apply
for the PWM and the raster engine, which are unchanged between VT8500
and WM8505, to my knowledge).

On the other hand, all of these parts might well be some kind of
licensed third-party design anyway, which makes this whole VIA vs. WMT
a bikeshed discussion mostly. I would personally not feel too guilty
about calling some device that first appeared in a VT8xxx chip 'wmt'
until there's technical reason for different 'compatible' strings
matching different chip versions (VIA age vs. WMT age). So far, we
don't have that need.

Best,
Alexey

-----

The problem is that we do have that need now - as I pointed out the PMC
on the VT8500 should not use the 'wmt,prizm-pmc' compatibility as it isn't
actually compatible. Any time you wanted to modify a bit-field you would
need to check whether you had a VT8500 or WM8xxx.

I appreciate that a lot of the devices are identical, but we do still need to
sort out those devices that aren't and what we're going to do with them.

The framebuffer is a good example - we have two separate drivers, one
for VT8500 and one for WM8xxx.

Regards,
Tony
Arnd Bergmann July 19, 2012, 12:28 p.m. UTC | #9
On Thursday 19 July 2012, Tony Prisk wrote:
> I have no issue with that.
> 
> I am mainly concerned with the Wondermedia products at the moment.
> 
> Adding device tree support for WM8505 (and WM8650) is my main goal as the WM8650
> currently isn't supported and trying to get another board file added seems unlikely with
> everything moving to device tree. The VT8500 requires additional/different patches and
> as I don't have a vt8500 I can't test them.
> 
> It does raise the question of applying 'wmt, prizm-pmc' to the vt8500 though. While it's
> fine for the restart patch I submitted, the vt8500 registers are the same but the bit-fields
> are different and therefore not entirely compatible with 'prizm-pmc'.
> When it comes to other PMC-related features, in particular the clocks, the VT8500 is quite
> different in its bit-fields.
> 
> The vt8500 isn't technically in the Prizm family, as that is Wondermedia's name for the SoC's.
> 

My point is that for each peripheral, we should decide separately what the
most descriptive name for it is. If the pmc is common for all wmt prizm
chips, then using the "wmt,prizm-pmc" string makes sense, and vt8500
can use "via,vt8500-pmc" or something like that.

However for stuff like the rtc that appears to be unchanged between them,
I would use the compatible string of the oldest known chip using it,
so even a wmt8650 should just claim compatibility with vt8500 there and
list both "wmt,wm8650-rtc" and "via,vt8500-rtc" there.

	Arnd
Tony Prisk July 19, 2012, 12:39 p.m. UTC | #10

Arnd Bergmann July 19, 2012, 12:55 p.m. UTC | #11
On Thursday 19 July 2012, Tony Prisk wrote:
> Which brings us back to the original question (and raises another one) - can we assume yet that
> 'wmt' will be the vendor-binding for Wondermedia? Also, what are we going to use for VIA-based
> SoC's given that VIA is the ticker for VIACOM. 
> 
> I did do a bit of a search on some stock tickers - 'WM' is Waste Management Inc. on the NYSE.
> Presumably this would be as safe, or safer, than WMT (Walmart).
> VIACOM, being a media content company, seems unlikely to produce any hardware but its possible
> they will rebrand others hardware.

"wm" has the danger of being confused with Wolfson Microelectronics, but
they are already listed in the documentation as using "wlf".

I'd really like to hear other opions from people on the devicetree-discuss
list, but if nobody replies, I'd suggest we go with "via" and "wm".

	Arnd
diff mbox

Patch

diff --git a/Documentation/devicetree/bindings/vendor-prefixes.txt b/Documentation/devicetree/bindings/vendor-prefixes.txt
index db4d3af..b87dc6b 100644
--- a/Documentation/devicetree/bindings/vendor-prefixes.txt
+++ b/Documentation/devicetree/bindings/vendor-prefixes.txt
@@ -48,4 +48,5 @@  st	STMicroelectronics
 stericsson	ST-Ericsson
 ti	Texas Instruments
 wlf	Wolfson Microelectronics
+wmt	Wondermedia Technologies
 xlnx	Xilinx