Rockbox Development > New Ports
Ingenic Jz4740/Jz4732 players
mcuelenaere:
Current status:
the 'crashing' problem have been identified as TLB Load/Store/Ifetch exceptions.
Also, TLB refill's have been requested by the MMU.
Current workaround: disable interrupts :/
Next blocker is the NAND: for some reason it just won't respond to any READ commands (I've tried several timing options, none worked).
The only NAND-related stuff that works is it's identification..
The Onda VX767 has been added as a target and will probably differ not much from the VX747.
I found the official datasheets, but currently the parts what I'm interested in are either missing or not helping (NAND & MMU/TLB)..
So my current options are rather limited; with the main one being RE.
mcuelenaere:
Good news :)
I think almost all previous bugs have been fixed and only 1 new one has been introduced ;)
What was the cause of the weird TLB/MMU/Exception/Interrupt bug?
In the interrupt/exception handler all CPU registers get stored on the stack and when leaving from an interrupt, they get restored from stack.
_But_ instead of loading the data of register $1 into $1, it got loaded into $v1 (stupid typo, I know).
This messed up quite some other unrelated code, because of course this meant $v1 could get changed at _any_ time (interrupts occur 'randomly').
Today I also got threading code to work - needs thorough testing though, but at least the Onda ran >2min non-stop with interrupts on (while before it would crash at about 15-20sec).
I also dumped some NAND data through USB to my PC to try locating the FAT partition, but didn't came far. As expected, the data at page 0 in block 0 contains executable code (this gets executed as second-stage bootloader), but RE it doesn't provide much info.
And last but not least: the new bug makes the backlight blink when doing button_read_device(); I haven't investigated why this occurs and it started doing so when I implemented thread handling and added yield() to the bootloader.
here2code:
Thanks Saratoga for the help on irc channel the other day.
Hi Maurus,
I am able to load the bootloader in SDRAM and run using your code and usbtools. Next step for me is to try to load the bootloader in NAND and run it from there. Have you tried doing that?
Have you tried running rockbox from SDRAM using usbtools?
thanks.
mcuelenaere:
--- Quote from: here2code on September 26, 2008, 02:41:38 PM ---Thanks Saratoga for the help on irc channel the other day.
Hi Maurus,
I am able to load the bootloader in SDRAM and run using your code and usbtools. Next step for me is to try to load the bootloader in NAND and run it from there. Have you tried doing that?
Have you tried running rockbox from SDRAM using usbtools?
thanks.
--- End quote ---
No, I haven't tried running the bootloader from NAND but AFAI checked, it should work (I made the SDRAM environment as close as possible to the NAND one).
And there's no point running Rockbox from SDRAM as no FAT partition can be mounted (yet), and as Rockbox will need to load some files it is rather pointless.
Currently what we need to find out is what NAND wearout scheme/NAND layout structure has been used by ChinaChip (I'm betting on either the one implemented in µCos II, as a lot of that source code has been used in the OF or the MTD ones in Linux)
==off topic==
BTW have you turned on nick name highlighting in mIRC? Because I've tried to speak to you thrice now and you don't seem to have responded (all text in #rockbox is logged).
LeJ:
Hi mcuelenaere,
I have just read the talk between you and wpyh.
http://www.rockbox.org/irc/rockbox-20080910.txt
I got the same problem with VX767 lcd, the backlight turns on, and I get nothing displayed.
On my experience in linux lcd driver. I think that maybe:
1. The LCD controller was not been configured properly.
2. The LCM need power on sequence
I don't know if the LCM is I2C, if so we can saw the sequence in oscilloscope.
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version