Hi Ruben, I have now pushed the yaffs2 changes so you should see them and no longer get segfaults.
What I also have tried is to improve the PIC control code (not yet committed). It appears that I have two different hardware models. One did (because it is now bricked) respond to the code in GIT, while the other just prints pic-shutdown but the kernel continues to run. So I don't know if power is controlled on this model through GPIO 69 or I2C or even both. Both models must have different PIC firmware since the older model just switches the screen on (which shows random pixels), then needs one second until the SYNC comes. The newer one keeps the screen dark for 1 second... Finally, I have tried to work on the PIC interrupt (to teach the power button to send ctl-alt-del). But there I have a strange situation. As soon as I register my IRQ for GPIO 65 the kernel no longer boots but hangs with an uninitialized IRQ for GPIO 64. This is the MMC card detect intererupt. What I don't understand: why is this INT occurring, but uninitialized? And, why only if I request_irq(IRQ_GPIO_0 + 65, ...)? Well, it may be a race condition combined with the initialization of the whole interrupt system. If the kernel does not respond to any intererupt before the first request_irq() is done, and the card detect line is already active (because I boot from SD), this could explain the situation. Best regards, NIkolaus Am 20.06.2010 um 13:22 schrieb Ruben Viets: > Hi Nikolaus, > > Downloaded the latest git sources. git clone > git://git.goldelico.com/letux-400.git > > Two issues, may need additional fix or upload of your local patches: > - issue with nand mounting. Showing errors on yaffs2 tragedy.. cause kernel > segfault and stuck > - The key's to startup from different boot don't work, alway's rooted from > /dev/mtdblock4. > > Grt Ruben. > > From: r_vi...@hotmail.com > To: h...@computer.org > Date: Sat, 19 Jun 2010 10:10:11 +0000 > CC: mipsbook-devel@linuxtogo.org > Subject: Re: [Mipsbook-devel] yaffs2 > > Hi Nikolaus, > > I have simular result. Used df to see the size and available space on all > partitions. > > - It looks like the /dev/mtdblock3 is in use by jffs2 blocks. > dmesg shows some prioritised blocks when mounting as yaffs2. So better to > mount as jffs2 I think. > > So far this kernel 2.6.24.3 and the updated sources are very usable now. ;-) > > Grt Ruben. > > > > Subject: Re: [Mipsbook-devel] yaffs2 > > From: h...@computer.org > > Date: Wed, 16 Jun 2010 08:44:33 +0200 > > CC: r_vi...@hotmail.com; mipsbook-devel@linuxtogo.org > > To: h...@computer.org > > > > > > Am 16.06.2010 um 08:31 schrieb Dr. H. Nikolaus Schaller: > > > > > > > > Am 15.06.2010 um 20:34 schrieb Ruben Viets: > > > > > >> Hi Nikolaus, > > >> > > >> You already have all sources if you tar xvjf the linux-2.6.24.3 and > > >> afterwards patch -p1 < patch20080530 + patchv33_withoutyaffs2 > > > > > > Ok, this was quite easy. > > > > > > I did not find many differnces: > > > > > > include/mtd/mtd-abi.h -- different size for oob fields (which may be the > > > issue) > > > fs-fs-writeback.c -- different strategy for locking > > > fs/mpage.c -- "It seems that there is a IO coherent bug in > > > generic_file_mmap()" > > > fs/yaffs/* -- about 50% of the files differ > > > > > > > > >> > > >> Don't need to tar and send my files then. Question is if the below dirs > > >> you mention are enough. You may need additional files. > > >> So better start from scratch in a different dir and do a diff with your > > >> letux-400 development directory. > > >> > > >> My latest config (CONFIG_RV9) is located at > > >> http://projects.kwaak.net/twiki/pub/Epc700/KernelPort26Compiling/CONFIG_RV9 > > > > > > Here I did not have CONFIG_MTD_SW_HM_ECC set and none of the VERIFY > > > options > > > > > > Now it *appears* to work. There are still some strange messages like > > > "Freechunks verification failure" but I did not do a flash_eraseall - I > > > just tried to continue to use the file system created with the broken > > > yaffs2. > > > > Ok, I have erased the minifs partition and mounted as yaffs2. The message > > has disappeared. > > > > But there is a mismatch between used/available blocks and what I see as > > files (an empty lost+found directory). And 'du' reports 40% used. > > > > When initializing as jffs2 I get 8% used and a completely empty device. > > > > Or is yaffs inappropriate for small NAND partitions? > > > > Nikolaus > > New Windows 7: Simplify what you do everyday. Find the right PC for you. > New Windows 7: Simplify what you do everyday. Find the right PC for you. _______________________________________________ Mipsbook-devel mailing list Mipsbook-devel@linuxtogo.org http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/mipsbook-devel
