2430 FB driver issues

2430 FB driver issues

[nishanth menon]

Hi All,

On compiling the kernel for 2430sdp with defconfig, I see the
following error messages on shell:
Power Management for OMAP2 initializing
PRCM revision 1.0
omapfb omapfb: irq error status 40e2
omapfb omapfb: irq error status 0040
omapfb omapfb: irq error status 4022
omapfb omapfb: irq error status 4000
<snip>
omapfb omapfb: irq error status 4000
<snip>
I believe this is from drivers/video/omap/dispc.c:876. I have tried
two versions of uboot: one with PRCM 5a and the other with PRCM2.
neither worked. The kernel version is the latest pull from omap-git
from source.mvista.com.

On disabling framebuffer, it boots up fine.

Regards,
Nishanth Menon

Re: 2430 FB driver issues

[Kevin Hilman]

nishanth menon wrote:
> Hi All,
> 
> On compiling the kernel for 2430sdp with defconfig, I see the
> following error messages on shell:
> Power Management for OMAP2 initializing
> PRCM revision 1.0
> omapfb omapfb: irq error status 40e2
> omapfb omapfb: irq error status 0040
> omapfb omapfb: irq error status 4022
> omapfb omapfb: irq error status 4000
> <snip>
> omapfb omapfb: irq error status 4000
> <snip>
> I believe this is from drivers/video/omap/dispc.c:876. I have tried
> two versions of uboot: one with PRCM 5a and the other with PRCM2.
> neither worked. The kernel version is the latest pull from omap-git
> from source.mvista.com.
> 
> On disabling framebuffer, it boots up fine.

FWIW, I have the same errors, but have not debugged it further.

Kevin

RE: 2430 FB driver issues

[Syed Mohammed, Khasim]

>nishanth menon wrote:
>> Hi All,
>>
>> On compiling the kernel for 2430sdp with defconfig, I see the
>> following error messages on shell:
>> Power Management for OMAP2 initializing
>> PRCM revision 1.0
>> omapfb omapfb: irq error status 40e2
>> omapfb omapfb: irq error status 0040
>> omapfb omapfb: irq error status 4022
>> omapfb omapfb: irq error status 4000
>> <snip>
>> omapfb omapfb: irq error status 4000
>> <snip>
>> I believe this is from drivers/video/omap/dispc.c:876. I have tried
>> two versions of uboot: one with PRCM 5a and the other with PRCM2.
>> neither worked. The kernel version is the latest pull from omap-git
>> from source.mvista.com.
>>
>> On disabling framebuffer, it boots up fine.
>
>FWIW, I have the same errors, but have not debugged it further.
>

I am not seeing any such messages and it boots happily. 
I am using u-boot 1.1.4 Apr 20 2007, PRCM 2, 2430 SDP V4.0.

According to Nishanth's dump, The status shows LCD sync lost, which happens when there is no data to sync up or when DSS is operating at low frequency (not able to feed continuous data to LCD). I verified register set with TI's DSS drivers, every thing looks reasonable. The PCD is also proper (8) the DSS FCLK is 41.25 Mhz. I don't see any issues. If I get some time I will try with other board.

Regards,
Khasim

RE: 2430 FB driver issues

[Syed Mohammed, Khasim]

>>
>>FWIW, I have the same errors, but have not debugged it further.
>>
>
>I am not seeing any such messages and it boots happily.
>I am using u-boot 1.1.4 Apr 20 2007, PRCM 2, 2430 SDP V4.0.
>
>According to Nishanth's dump, The status shows LCD sync lost, which happens
>when there is no data to sync up or when DSS is operating at low frequency
>(not able to feed continuous data to LCD). I verified register set with
>TI's DSS drivers, every thing looks reasonable. The PCD is also proper (8)
>the DSS FCLK is 41.25 Mhz. I don't see any issues. If I get some time I
>will try with other board.
>

I do see some configurations that will result in Sync-Lost.

The optimal way of operating GFX pipeline is to
 - Configure GFX DMA burst size to 16x32
 - The difference between GFX HIGH-FIFO to GFX LOW-FIFO should be equal to DMA burst size (16x4 = 64).

So the Ideal configuration would have been, 
480504A0  = 0000008D (GFX Attribute)
480504A4  = 00FF00C0 (FIFO threshold)

What's configured is:
480504A0  = 0000004D (GFX Attribute)
480504A4  = 00C00040 (FIFO threshold)

This might cause LCD sync lost. Some one can try the correct settings and see if this is still resulting in Sync lost interrupts. The Clock seems to be configured correctly.

Regards,
Khasim

Re: 2430 FB driver issues

[Paul Walmsley]

Hi Kevin,

On Fri, 6 Jul 2007, Kevin Hilman wrote:

> FWIW, I have the same errors, but have not debugged it further.

You mentioned earlier that this code formerly worked for you before the 
PRCM changes.  Do you have a commit ID or revision from a working version? 
if so then perhaps Nishanth or someone else with a 2430 SDP might be able 
to git-bisect to find which commit broke it.


- Paul

Re: 2430 FB driver issues

[Hunyue Yau]

Paul,

It should have been working when it went it. The commit entry 
adding it should be:

commit 4ed427693dd7e0d4fd00a493988cefaa17accebc
Author: Hunyue Yau <hyau@mvista.com>
Date:   Fri May 25 15:33:08 2007 -0700

    OMAP: 2430SDP: Support LCD display as a FB device
    
    Add glue to control the 2430SDP LCD as a frame buffer device
    using the existing dispc.c driver under omapfb.
    
    Signed-off-by: Hunyue Yau <hyau@mvista.com>
    Signed-off-by: Kevin Hilman <khilman@mvista.com>
    Signed-off-by: Tony Lindgren <tony@atomide.com>

-- Hunyue

On Sunday 08 July 2007 16:01, Paul Walmsley wrote:
> Hi Kevin,
> 
> On Fri, 6 Jul 2007, Kevin Hilman wrote:
> 
> > FWIW, I have the same errors, but have not debugged it further.
> 
> You mentioned earlier that this code formerly worked for you before the 
> PRCM changes.  Do you have a commit ID or revision from a working version? 
> if so then perhaps Nishanth or someone else with a 2430 SDP might be able 
> to git-bisect to find which commit broke it.
> 
> 
> - Paul
> _______________________________________________
> Linux-omap-open-source mailing list
> Linux-omap-open-source@linux.omap.com
> http://linux.omap.com/mailman/listinfo/linux-omap-open-source
> 

Re: 2430 FB driver issues

[Nishanth Menon]

All,
Apologies on a late response..
Hunyue Yau stated on 7/9/2007 2:04 PM:
> Paul,
>
> It should have been working when it went it. The commit entry 
> adding it should be:
>
> commit 4ed427693dd7e0d4fd00a493988cefaa17accebc
> Author: Hunyue Yau <hyau@mvista.com>
>
>   
to keep things simple, here is the set of facts that i gathered:
a) With Hunyue patch(4ed427693dd7e0d4fd00a493988cefaa17accebc),
Framebuffer is enabled, BUT LCD backlight does not come on, nor does
anything come to the display. Intrestingly, as Hunyue mentioned, NO errors.
b) Patch from Paul (5c6cd5d71bdb1a8536238c7812f1dbd6a68bc542), enables
LCD on 2430SDP with LCD backlight and all.. This is the first patch that
makes 2430SDP LCD work and the error that i started this chain with exists!
c) all intermediate patches between (a) and (b) including the following:
Richard W (bbe8d2b1654809c39be2e9ecc91dd5632006a2cb)
Tony (517928578afaea422975fd8216242fdd067b56a2)
Tony(dd1d9a62f7cd60b544013eb429d6a3137617f0b2)
all retain the same state as I mentioned in (a).

Not much useful information except that my guess is FB LCD is broken on
2430SDP as of date and it is possibly broken from the time it is
enabled.. Since i do see Tony punching in a bunch every weekend, did
Hunyue test with the git latest once Tony finished pushing all patches?
- esp since May 25th is the time when the major clock/prcm  framework
patches from Paul W went into git.
> On Sunday 08 July 2007 16:01, Paul Walmsley wrote:
>   
>> You mentioned earlier that this code formerly worked for you before the 
>> PRCM changes.  Do you have a commit ID or revision from a working version? 
>> if so then perhaps Nishanth or someone else with a 2430 SDP might be able 
>> to git-bisect to find which commit broke it.
>>
>>     
Hmm... I did try this, and i failed badly.. this is a topic of a
seperate mail..

Regards,
Nishanth Menon

RE: 2430 FB driver issues

[Woodruff, Richard]

 > Not much useful information except that my guess is FB LCD is broken on
> 2430SDP as of date and it is possibly broken from the time it is
> enabled.. Since i do see Tony punching in a bunch every weekend, did
> Hunyue test with the git latest once Tony finished pushing all patches?
> - esp since May 25th is the time when the major clock/prcm  framework
> patches from Paul W went into git.

It is working now right?  I updated and booted my local version and all seems happy.

I have a random patch or two differences from the stock kernel but I'm not sure they matter.

Regards,
Richard W.

----------------
root@128.247.75.30:/tst# uname -a
Linux 128.247.75.30 2.6.22-omap1-g2a1c4dfb-dirty #1 PREEMPT Thu Jul 19 10:24:19
CDT 2007 armv6l GNU/Linux

root@128.247.75.30:/tst#./powertop
    PowerTOP version 1.0       (C) 2007 Intel Corporation

< Detailed C-state information is only available on Mobile CPUs (laptops) >
tail: `-190' option is obsolete; use `-n 190'
Try `tail --help' for more information.


Top causes for wakeups:
  86.7%        <interrupt> : gp timer
  11.1%        <interrupt> : eth0
   2.2%        <interrupt> : serial

Suggestion: Enable the CONFIG_USB_SUSPEND kernel configuration option.
This option will automatically disable UHCI USB when not in use, and may
save approximately 1 Watt of power.

RE: 2430 FB driver issues

[Syed Mohammed, Khasim]

> > Not much useful information except that my guess is FB LCD is broken on
>> 2430SDP as of date and it is possibly broken from the time it is
>> enabled.. Since i do see Tony punching in a bunch every weekend, did
>> Hunyue test with the git latest once Tony finished pushing all patches?
>> - esp since May 25th is the time when the major clock/prcm  framework
>> patches from Paul W went into git.
>
>It is working now right?  I updated and booted my local version and all
>seems happy.
>
>I have a random patch or two differences from the stock kernel but I'm not
>sure they matter.
>

Some times it works with out any issues and few times it starts generating Sync Lost error. I was not getting this error for a week, but now even I notice these errors on same board :)

Regards,
Khasim

RE: 2430 FB driver issues

[Woodruff, Richard]

Yes, I saw it now.

If I dump the registers in Lauterbach and compare against our internal settings and the settings in the git tree I see a number of differences.

The FIFO thresholds are quite different, the dma burst size is different, and the pixel clock is running fast on the git tree.

I'd venture a guess if you match up FIFO, PCD, and DMA burst the problem will go away.  When you under your FIFO bad things happen and that is just what the 0x40 in the 0x4040 says is happening.

If anyone is interested I can send some .png's of registers.  They are like 300k so I won't post it here.

Regards,
Richard W.

> -----Original Message-----
> From: Syed Mohammed, Khasim
> Sent: Thursday, July 19, 2007 11:54 AM
> To: Woodruff, Richard; Nishanth Menon; Hunyue Yau
> Cc: Paul Walmsley; linux-omap-open-source@linux.omap.com
> Subject: RE: 2430 FB driver issues
> 
> > > Not much useful information except that my guess is FB LCD is broken
> on
> >> 2430SDP as of date and it is possibly broken from the time it is
> >> enabled.. Since i do see Tony punching in a bunch every weekend, did
> >> Hunyue test with the git latest once Tony finished pushing all patches?
> >> - esp since May 25th is the time when the major clock/prcm  framework
> >> patches from Paul W went into git.
> >
> >It is working now right?  I updated and booted my local version and all
> >seems happy.
> >
> >I have a random patch or two differences from the stock kernel but I'm
> not
> >sure they matter.
> >
> 
> Some times it works with out any issues and few times it starts generating
> Sync Lost error. I was not getting this error for a week, but now even I
> notice these errors on same board :)
> 
> Regards,
> Khasim

2430 FB driver issues

[Timothy Tang]

Is there any update to the fix of this issue?  I also have to disable the 
FB currently.

Regards,
Tim