This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.
Messages - tangrs
Pages: 1 ... 5 6 [7] 8 9 ... 16
91
« on: January 19, 2013, 05:12:56 am »
Hmmm, I'm starting to implement a CPU frequency scaling driver. At the moment, it simply reports the CPU frequency.
Now, I need some help with frequency scaling. If we assume that the base speed won't change, what would be a good way to determine a table of "safe" speeds to scale to? Should we be touching the base speed? What happens if the user overclocks the calculator? How do we ensure that AHB/CPU speeds are always within safe limits?
Any ideas?
Edit: I'll try a list of "safe" divisors for now.
92
« on: January 17, 2013, 04:49:22 am »
Maybe instead of compiling it for Ndless it could be programmed for tangrs Linux port for nspire? Then you could use C++.
One of the motivators of porting Linux to the calculator was to have an open source alternative to the Nspire CAS software  In fact, if the software was originally built for and already has a GUI for Linux, it shouldn't be harder than just simply cross comping to work on Nspire Linux.
93
« on: January 14, 2013, 05:08:29 am »
Edit2: I compiled psplash and now we have a nice progress bar on boot:
Very nice! Edit: Added a backlight driver for Nspire Linux so the backlight also turns off as the LCD powers off. That should save some power and not show the ugly white fading screen when the LCD turns off. I am also planning to get a contrast driver working for the classics so we can remove the ugly /proc/contrast and /proc/backlight files.
94
« on: January 12, 2013, 07:00:30 pm »
Maybe I'm testing too much?
Now I'm worrying about my reset-button and flash..
Oh, now I remember what we have to support, too:
Charging the battery while running linux!
It doesn't work yet.
That's part of what I'm working on. I'm still trying to figure out the USB PHY.
95
« on: January 12, 2013, 06:52:28 pm »
Did it happen to you, too?
If not, my device really hates me (even the reset-button doesn't work always)
It happens but definitely no where as often as yours o_O (only happened 3 times so far in all of booting Linux)
96
« on: January 12, 2013, 06:49:00 pm »
Yeah, you should fix that 
I don't know how, maybe don't map it at all?
Now, I'm stumped. The NAND isn't actually mapped at all.
97
« on: January 12, 2013, 06:46:07 pm »
Oh really? Damn...
98
« on: January 12, 2013, 06:42:59 pm »
There's actually another write that I put into the start of nspire_classic_init. Now I think about it, it should work because the bits should represent devices rather than addresses (which isn't what I initially thought).Never mind 
99
« on: January 12, 2013, 06:38:34 pm »
But it works! 
At least there weren't any problems after 1<<7.
Hmmm. I see. In that case, move the disable from classic.c to common.c. I'm in the middle of a huge, unstaged mess at the moment
100
« on: January 12, 2013, 06:33:11 pm »
But one problem does still exist:
Edit: After reboot in linux the touchpad doesn't work anymore with nucleus, you have to reboot.
But it doesn't boot anymore and you have to run NAND-Tests everytime you reboot in linux.
I'll try disabling access with 1<<7 to 0x900B0018
With 1<<7 it doesn't damage the flash anymore, but the touchpad does still work only after a reboot.
I don't think the touchpad problem is too easy to find, but the flash problem has to be solved asap!
I didn't lose my data yet, but why is linux doing something with the NAND at all?
The 1<<7 to protect the NAND only applies to the classics AFAIK. The CX uses some different ways to access the NAND so 1<<7 is probably undefined on the CX.
101
« on: January 12, 2013, 06:24:34 pm »
Though, some one did ask about it in a mailing list in 2008 and probably succeeded in writing one. I read:
It sounds like it isn't 'NAND' at all, as far as you're concerned. It's
a disk drive of some form. Usually these things present themselves as
IDE, so you'd use the IDE driver. If it has its own 'special' interface
then you probably want a dedicated block driver for it; nothing to do
with the MTD subsystem at all.WTF? Flash controller with IDE interface for ARM SoCs?
I think the person who replied didn't know what he was talking about.
102
« on: January 12, 2013, 06:19:47 pm »
Vogtinator, you have royally screwed your desktop. You extracted an ARM linux install on top of your x86 linux root. You're lucky that it still boots, likely thanks to various configuration differences between the two. Nope, it didn't boot. Had to restore the files from another laptop:
"export list=`tar -tf xconfig.tar.bz2`; for i in $list; do echo $i; rm /mnt/$i; rsync -lptgoDvzP /media/$i /mnt/$i; done;"
The NAND chip itself is supported and always has been. But the NAND controller isn't Is it the same on all nspires?
I haven't looked at the classic nspires in depth yet but I'm pretty sure there isn't a existing driver for the CX. Though, some one did ask about it in a mailing list in 2008 and probably succeeded in writing one. Think it's worth asking if he has any source code available? Nice, I look forward to seeing it Here you go:
That is really awesome!
103
« on: January 12, 2013, 06:08:27 pm »
Look what I found out: http://www.linux-mtd.infradead.org/nand-data/nanddata.html
The NAND is already actually supported by linux!
I was a bit frustrated because of this silly synaptics_i2c polling too slow, so I did a bit of research.
Edit: After reboot in linux the touchpad doesn't work anymore with nucleus, you have to reboot. But it doesn't boot anymore and you have to run NAND-Tests everytime you reboot in linux. I'll try disabling access with 1<<7 to 0x900B0018
Edit2: The TOUCHPAD WORKS! I'll upload a video later/tomorrow
The NAND chip itself is supported and always has been. But the NAND controller isn't Nice, I look forward to seeing it
104
« on: January 12, 2013, 07:34:26 am »
The filesystem, we'll leave til later to solve. Perhaps the recently released nLaunch could help us here? Not nLaunch itself, but...
Edit: I've seen some interesting code hidden in the OS that formats the flash.
Maybe there's some sort of data structure that we could modify to trick the OS into believing we have less NAND than we actually have and have it format the filesystem to only utilize that much. Under Linux, we could partition the NAND up and only use the portion that the Nspire OS doesn't use. ... in the community, we've known for nearly three years the boot2 1.4.1571 for Clickpad/Touchpad contains tons of debugging and testing code for Nucleus and its elements, especially filesystem operations 
The boot2 3.1.131 is smaller, and presumably doesn't contain as much.
Reckon it's possible to utilize a few to do some dirty, partitioning work for us?
105
« on: January 12, 2013, 05:45:55 am »
I'll get cracking on the hardest device then - the NAND. Yay! But we don't have any filesystem driver for it..
But the driver shouldn't be TOO hard, the emulator source should be enough, maybe there's already a driver for our chip available?
I'm pretty sure there isn't one for the CX at least. The CX one is hard because it uses some weird scheme of embedding NAND commands into the address then performing a write  Not sure about the classic models though. The filesystem, we'll leave til later to solve. Perhaps the recently released nLaunch could help us here? Edit: I've seen some interesting code hidden in the OS that formats the flash. Maybe there's some sort of data structure that we could modify to trick the OS into believing we have less NAND than we actually have and have it format the filesystem to only utilize that much. Under Linux, we could partition the NAND up and only use the portion that the Nspire OS doesn't use.
Pages: 1 ... 5 6 [7] 8 9 ... 16
|
Show Posts