| 06 May 2010 09:38 AM |
|
|
I have tried to load CE 6.0 through the serial port and it fails with a data abort. Has anyone been able to load the DEV image off their website?
Here's what it does.
ls
D : tftp 0
D : etc 0
D : dev 0
losh> erase B1 B4095 /dev/nand0
erasing nand: 100%
erased '/dev/nand0' start=0x1: len=0xfff bytes/blocks skipped 0
losh> mount fatfs /cf
losh> cd cf
losh> load bin NK.BIN
loading from NK.BIN:
R
.............................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................
image launch addr: 0x80201000
Verifying checksums: >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
Checksums good.
file loaded
losh> video-open REI 16
video-open display: width: 800 height: 600 bpp: 16 disp: REI
losh> draw-test
losh>
losh> exec rtc:rtc_imx31_pmic:dbg_leds:1:dbg_serial:IMX31_UART:skiplcdcinit:1
INFO:OALLogSetZones: dpCurSettings.ulZoneMask: 0xb
Windows CE Kernel for ARM (Thumb Enabled) Built on Jan 21 2009 at 18:02:34
INFO:OALLogSetZones: dpCurSettings.ulZoneMask: 0xb
LoLo Parameter: rtc:rtc_imx31_pmic:dbg_leds:1:dbg_serial:IMX31_UART:skiplcdcinit:1
-----------------------------------------
iMX31 Clock Information
.......................................
MCU PLL = 399000000 Hz
USB PLL = 240000000 Hz
SER PLL = 220147200 Hz
ARM CLOCK = 399000000 Hz
IPU CLOCK = 133000000 Hz
AHB CLOCK = 133000000 Hz
More Information Available in Debug Build
-----------------------------------------
-----------------------------------------
iMX31 PMIC Information
.......................................
Information Available in Debug Build
-----------------------------------------
OAL: SWITCHING TO HIGH SPEED ARM CLOCK = 399000000 Hz
LAN9118: Chip Id 117 Revision 2
WARN: OALKitlInit: KITL Disabled
-----------------------------------------
LogicPD i.MX31 SOM Revision Information
.......................................
BSP Revision : 1.0.0
Built for Hardware : i.MX31 SOM-LV
-----------------------------------------
LogicPD i.MX31 SOM CoProc Information
.......................................
L1 Cache : Enabled
L2 Cache : Enabled
Program Flow Prediction : Enabled
-----------------------------------------
LogicPD i.MX31 SOM Processor Information
.......................................
Silicon Rev : 0x00000028
CPU Main ID Register : 0x4107b364
Architecture ID : 0x07
ARM Ver : ARMv6
-----------------------------------------
OEM Initialization Done.
OAL: Debug Serial in Use.
OAL: Disabling "Drivers\BuiltIn\COM1" driver
VBridge:: VB_INITIALIZED returns [0]
IPU: Boot string 'lcdcinit' set - display driver will skip LCD initialization.
Exception 'Data Abort' (4): Thread-Id=04140002(pth=8c79f000), Proc-Id=00400002(pprc=8925a308) 'NK.EXE', VM-active=01800002(pprc=8f657c54) 'udevice.exe'
PC=c080e05c(ddraw_ipu.dll+0x0000e05c) RA=c0805b2c(ddraw_ipu.dll+0x00005b2c) SP=d0f1f960, BVA=000000b4
|
|
|
|
richard.laborde@logicpd.com Basic Member Posts:247  
| 10 May 2010 03:50 PM |
|
|
Alan - try it with the "5" display. That has worked for the show image for me. "7" does too.
Which LoLo version are you running? I tried both 2.4.8 (unreleased) and 2.4.9.
|
|
|
|
| 11 May 2010 08:35 AM |
|
|
I was using 2.4.0p1, but I updated to 2.4.9 with no effect.
I tried using displays 5 and 7. First I executed the "draw-test" command to see how it would appear on my display. It showed the draw test well enough, but repeated itself towards the right / bottom of the screen. Of course, it doesn't look right, but it should be ok to boot. (I believe I've done the same on 5.00 and it looked wrong but worked).
The good news is, when booting with display 5 or 7 it did not crash with the ddraw data abort! The bad news is, the screen remained all white... and I could not see anything. however, activesync connected and it seems to be running.
Also, I would like to tell you that I had pre-release version 0.2.0 and it boots and the display works. I wonder, what changed between 0.2.0 and 1.0.0?
|
|
|
|
| 11 May 2010 08:56 AM |
|
|
Well.. I have a trial version of this program called "beyond compare" it allows me to compare the source code for the drivers across the entire folder. I chose to start at the SRC\DRIVERS\IPU\DISPLAY\COMMON folder. The files I see that are different are:
ddipu.cpp
ddipu.h
ddipu_line.cpp
ddipu_misc.cpp
ddipu_overlay.cpp
ddipu_surf.cpp
halcaps.cpp
Some of these files only have very minor changes of course. I dunno if any of this would help, but I could supply the 0.2.0 code I have if it helps, a diff tool might shed some light.
|
|
|
|
richard.laborde@logicpd.comBasic Member Posts:247
| 11 May 2010 09:35 AM |
|
|
Alan,
Do you have one of our display kits? I connected the 12.1 and used the "5" screen setting and it boots to the standard desktop. (it is repeated and the touch is way off) but it booted at least. I have no idea what they changed between the prerelease and 1.0.0. I'll forward your info to the sw guys.
|
|
|
|
| 11 May 2010 09:52 AM |
|
|
No I don't believe that I have one of your displays.
|
|
|
|
| 11 May 2010 09:52 AM |
|
|
No I don't believe that I have one of your displays. I will check though.
|
|
|
|
| 11 May 2010 02:27 PM |
|
|
No we, don't have one. It would be nice to try, but not sure it would help.
I've also tried 3 more SOMs. I found two REV B boards, and another REV A, same thing with all of them..
What's the next step?
|
|
|
|
| 21 May 2010 10:32 AM |
|
|
I found the problem after receiving a copy of platform builder 6.0 which allowed me to debug the driver source.
From: PLATFORM\LPD_iMX31\SRC\INC\image_cfg.h
// PP internal working buffer, code buffer
#define IMAGE_WINCE_PP_RAM_OFFSET (0x9000)
#define IMAGE_WINCE_PP_RAM_PA_START (CSP_BASE_MEM_PA_CSD0+IMAGE_WINCE_PP_RAM_OFFSET)
#define IMAGE_WINCE_PP_RAM_SIZE (1.6*1024*1024)
For my display this is not enough memory for an 800x600 display.
800x600 = 480,000 x (2 bytes per pixel) = 960,000 * (2 surface buffers) = 1,920,000
After changing this to 1.9 MB my device boots and the display seems to be working just fine.
This was causing SetupVideoMemory() to fail, which would cause a CleanUp routine to be called that was supposed to free memory was crashing with an access violation. I didn't look into the cleanup function because I fixed the actual underlying issue with the memory allocation.
|
|
|
|