[U-Boot-Users] [PATCH] Motorola i.MX support (1/2)
Hello List,
These patches add Motorola MM1 support for U-boot. The patches are against U-boot-1.1.1 and add support for two custom boards and the Davicom dm9000 ethernet chip.
All comments and feedback welcome,
Sascha Hauer
patch-u-boot-1.1.1-imx.bz2 - Motorola MX1 support patch-u-boot-1.1.1-mx1fs2.bz2 - mx1fs2 board support patch-u-boot-1.1.1-scb9328.bz2 - scb9328 board support patch-u-boot-1.1.1-dm9000.bz2 - dm9000 network driver support (used in scb9328)
Sascha Hauer wrote:
Hello List,
These patches add Motorola MM1 support for U-boot. The patches are against U-boot-1.1.1 and add support for two custom boards and the Davicom dm9000 ethernet chip.
Looks like there are at least three (!) ports for Motorola's DragonBall i.MX (MC9328MX1/L) now:
1.) Ming-Len Wu minglen_wu@techware.com.tw already in CVS without (announcing on the list) 2.) Sascha Hauer sascha@saschahauer.de posted on 2004-06-17 3.) Bryan Larsen bryan.larsen@lumenera.com mentioned on list on 2003-09-19
Is that correct, Wolfgang?
Steven
Dear Steven,
in message 40D2D012.80807@imc-berlin.de you wrote:
These patches add Motorola MM1 support for U-boot. The patches are against U-boot-1.1.1 and add support for two custom boards and the Davicom dm9000 ethernet chip.
[I didn't check these patyches yet; I just put them on my stack ;-) ]
Looks like there are at least three (!) ports for Motorola's DragonBall i.MX (MC9328MX1/L) now:
1.) Ming-Len Wu minglen_wu@techware.com.tw already in CVS without (announcing on the list)
Correct. Checked in some time ago.
2.) Sascha Hauer sascha@saschahauer.de posted on 2004-06-17
Sasha, I think it makes sense that you check what's in the CVS first. Then let's discuss how to proceed. I think for now I will just put your patches on hold, as they seem obsoleted by the current code in CVS (sorry, but this is why ESR says "release early, release often").
3.) Bryan Larsen bryan.larsen@lumenera.com mentioned on list on 2003-09-19
But I haven't seen any actual code from him, AFAICT.
Is that correct, Wolfgang?
Yes, it is.
Best regards,
Wolfgang Denk
On Fri, Jun 18, 2004 at 03:19:37PM +0200, Wolfgang Denk wrote:
[I didn't check these patyches yet; I just put them on my stack ;-) ]
You should ;)
Sasha, I think it makes sense that you check what's in the CVS first. Then let's discuss how to proceed. I think for now I will just put your patches on hold, as they seem obsoleted by the current code in CVS (sorry, but this is why ESR says "release early, release often").
The design in CVS is crap. i.MX is not a separate CPU architecture, it is just ARM920T. There is no reason for a new directory in cpu/.
Robert
In message 20040618142346.GG24767@pengutronix.de you wrote:
[I didn't check these patyches yet; I just put them on my stack ;-) ]
You should ;)
I will. As time permits.
The design in CVS is crap. i.MX is not a separate CPU architecture, it is just ARM920T. There is no reason for a new directory in cpu/.
Maybe you want to submit a patch to clean this up (after negotiating with the current maintainer, of course) ?
Best regards,
Wolfgang Denk
On Fri, Jun 18, 2004 at 05:02:57PM +0200, Wolfgang Denk wrote:
Maybe you want to submit a patch to clean this up (after negotiating with the current maintainer, of course) ?
Well, Sascha's patch was our attempt to clean it up ;)
Robert
In message 20040618152118.GA27424@pengutronix.de you wrote:
Well, Sascha's patch was our attempt to clean it up ;)
But AFAICT they are against an OLD version of U-Boot, which did not yet include Ming-Len Wu's patches.
So what is your suggestion?
We can:
(1) add your patch (2) add your patch and revert Ming-Len Wu's patch (3) use Ming-Len Wu's code as base for which you submit fixes/improvements
(1) means that we have the same architecture / board supported twice differently. Not good. rejected.
From your posting it seems clear that your position is (2) [is it?];
and to me it seems you have at least a valid argument. But before doing this I would like to have an agreement between you (resp. Sascha Hauer) and Ming-Len Wu.
Best regards,
Wolfgang Denk
I've just had a look at the i.MX (MC9328 MX1, or however this thing is called) Port in current CVS. The main differences are:
- Like Robert said, my patch introduces less duplication of code - the serial driver in CVS supports only a fixed (hardcoded) baudrate - the architecture include file from cvs only has register definitions while my port also has bitmasks for most registers as well as a convenience function imx_gpio_mode which lets you set the gpio primary/alternate functions without having to play with hex values. - different boards supported. Two custom boards <-> MX1ADS. Adding support for MX1ADS should be no problem, though, since we only have to rewrite some register definitions. Unfortunately I don't have the hardware.
On Fri, Jun 18, 2004 at 05:59:13PM +0200, Wolfgang Denk wrote:
In message 20040618152118.GA27424@pengutronix.de you wrote:
Well, Sascha's patch was our attempt to clean it up ;)
But AFAICT they are against an OLD version of U-Boot, which did not yet include Ming-Len Wu's patches.
No problem, I can send a patch against current CVS.
So what is your suggestion?
We can:
(1) add your patch (2) add your patch and revert Ming-Len Wu's patch (3) use Ming-Len Wu's code as base for which you submit fixes/improvements
(1) means that we have the same architecture / board supported twice differently. Not good. rejected.
From your posting it seems clear that your position is (2) [is it?];
So I would prefer position (2), too.
Greetings
Sascha Hauer
Hi,
On Fri, Jun 18, 2004 at 09:30:20PM +0200, Sascha Hauer wrote:
I've just had a look at the i.MX (MC9328 MX1, or however this thing is called) Port in current CVS. The main differences are:
I would prefer imx for the architecture - it's i.MX, not only MC9328 or even MX1.
- different boards supported. Two custom boards <-> MX1ADS. Adding support for MX1ADS should be no problem, though, since we only have to rewrite some register definitions. Unfortunately I don't have the hardware.
Having support for the ADS would definitely a Good Thing (TM).
(1) add your patch (2) add your patch and revert Ming-Len Wu's patch (3) use Ming-Len Wu's code as base for which you submit fixes/improvements
(1) means that we have the same architecture / board supported twice differently. Not good. rejected.
From your posting it seems clear that your position is (2) [is it?];
So I would prefer position (2), too.
I would prefer to do (2) with a little bit of (3) ;) It shouldn't be too much work to port Mng-Len Wu's port to the more generic variant.
Robert
Robert Schwebel wrote:
Hi,
On Fri, Jun 18, 2004 at 09:30:20PM +0200, Sascha Hauer wrote:
I've just had a look at the i.MX (MC9328 MX1, or however this thing is called) Port in current CVS. The main differences are:
I would prefer imx for the architecture - it's i.MX, not only MC9328 or even MX1.
How do you seperate the MX1 from the MXL? Maybe it's not neccessary for U-Boot. We will hardly support the BlueTooth stuff in U-Boot... On the other hand the MX1 got one more UART...
BTW: Does anyone got a working BDI2000 config file for the MC9328MX1/L? I want to be able to burn the flash when we got the ADS board...
Thanks a million!
In message 40D67C99.5040102@imc-berlin.de you wrote:
Maybe it's not neccessary for U-Boot. We will hardly support the BlueTooth stuff in U-Boot... On the other hand the MX1 got one more UART...
Don't count on that. We have funny things in U-Boot - like VLAN support or a console over USB on a mobile phone...
Best regards,
Wolfgang Denk
Robert Schwebel wrote:
Hi, On Fri, Jun 18, 2004 at 09:30:20PM +0200, Sascha Hauer wrote:
I've just had a look at the i.MX (MC9328 MX1, or however this thing is called) Port in current CVS. The main differences are:
I would prefer imx for the architecture - it's i.MX, not only MC9328 or even MX1.
How do you seperate the MX1 from the MXL? Maybe it's not neccessary for U-Boot. We will hardly support the BlueTooth stuff in U-Boot... On the other hand the MX1 got one more UART...
BTW: Does anyone got a working BDI2000 config file for the MC9328MX1/L? I want to be able to burn the flash when we got the ADS board...
Thanks a million!
Hi Steven,
I have a M9328MXLADS 2.0 board right on my desktop. If the time permits, I'll be able to give some help about testing the code in the next weeks.
Regards,
llandre
DAVE Electronics System House - R&D Department www.dave-tech.it
On Mon, Jun 21, 2004 at 08:13:45AM +0200, Steven Scholz wrote:
How do you seperate the MX1 from the MXL? Maybe it's not neccessary for U-Boot. We will hardly support the BlueTooth stuff in U-Boot... On the other hand the MX1 got one more UART...
... which has the same register layout (well, at least that ones you need in u-boot) as the other ones. The basic serial support in cpu/arm920t/imx_serial.c (well, not exactly a good place for a serial driver, but that's how it was done for the other ports) can easily be extended to support more than two serial ports.
BTW: Does anyone got a working BDI2000 config file for the MC9328MX1/L? I want to be able to burn the flash when we got the ADS board...
Attached. I'm not sure about the memory situation on the ADS board, our board has 32 Bit flash.
Robert
Robert Schwebel wrote:
On Mon, Jun 21, 2004 at 08:13:45AM +0200, Steven Scholz wrote:
How do you seperate the MX1 from the MXL? Maybe it's not neccessary for U-Boot. We will hardly support the BlueTooth stuff in U-Boot... On the other hand the MX1 got one more UART...
... which has the same register layout (well, at least that ones you need in u-boot) as the other ones. The basic serial support in cpu/arm920t/imx_serial.c (well, not exactly a good place for a serial driver, but that's how it was done for the other ports)...
Bad excuse for doing the same bad thing... ;-)
But where would be a goog place to store such code? The same goes for the serial driver of the AT91RM9200 (which is ARM9 too). Wolfgang?
BTW: Does anyone got a working BDI2000 config file for the MC9328MX1/L? I want to be able to burn the flash when we got the ADS board...
Thanks a million for the config. I'll test it with the ADS as soon as we get it.
Steven
On Mon, Jun 21, 2004 at 10:17:27AM +0200, Steven Scholz wrote:
Bad excuse for doing the same bad thing... ;-)
But where would be a goog place to store such code? The same goes for the serial driver of the AT91RM9200 (which is ARM9 too).
It would probably belong into something like drivers/serial. The question is what else had to be changed when we do something like this.
Robert
Sascha Hauer wrote:
I've just had a look at the i.MX (MC9328 MX1, or however this thing is called) Port in current CVS. The main differences are:
- Like Robert said, my patch introduces less duplication of code
- the serial driver in CVS supports only a fixed (hardcoded) baudrate
- the architecture include file from cvs only has register definitions while my port also has bitmasks for most registers as well as a convenience function imx_gpio_mode which lets you set the gpio primary/alternate functions without having to play with hex values.
- different boards supported. Two custom boards <-> MX1ADS. Adding support for MX1ADS should be no problem, though, since we only have to rewrite some register definitions. Unfortunately I don't have the hardware.
We will have one in the next few days. I'd love to help testing it.
Steven
Steven Scholz wrote:
Sascha Hauer wrote:
I've just had a look at the i.MX (MC9328 MX1, or however this thing is called) Port in current CVS. The main differences are:
- Like Robert said, my patch introduces less duplication of code
- the serial driver in CVS supports only a fixed (hardcoded) baudrate
- the architecture include file from cvs only has register definitions while my port also has bitmasks for most registers as well as a convenience function imx_gpio_mode which lets you set the gpio primary/alternate functions without having to play with hex values.
- different boards supported. Two custom boards <-> MX1ADS. Adding support for MX1ADS should be no problem, though, since we only have to rewrite some register definitions. Unfortunately I don't have the hardware.
We will have one in the next few days. I'd love to help testing it.
I just got my brand new toy! :o)
So any news on the U-Boot i.MX support?
Steven
participants (6)
-
llandre -
Robert Schwebel -
Robert Schwebel -
Sascha Hauer -
Steven Scholz -
Wolfgang Denk