Re: More JFFS problems...

[Alexander Larsson <alex@xxxxxxx.se>]

On Wed, 2 Aug 2000, David Woodhouse wrote:

> 
> Test program:
> 
> 	fd = open("testfile", O_CREAT| O_RDWR, S_IRUSR);
> 	write(fd, "Thrunge\n", 8);
> 	write(fd, "Wibble\n",7);
> 	lseek(fd, 2, SEEK_SET);
> 	write(fd, "xx", 2);
> 	fsync(fd);
> 	close(fd);
> 
> 
> Result:
> [root@alpha1 /mnt]# /home/dwmw2/testfile-alpha 
> [root@alpha1 /mnt]# cat testfile 
> Thrunge
> Wibble
> [root@xxxxxxx.
> [root@alpha1 /]# umount /mnt
> [root@alpha1 /]# mount /dev/mtdblock0 /mnt -t jffs
> [root@alpha1 /]# cat mnt/testfile 
> Thrunge
> Wibble
> xx[root@alpha1 /]#              (note the 'xx' is at the end)
> 
>  The structure on the media seems to be correct - the third node has a 
> start location of '2'. However, the result of reading back the file, either 
> immediately, or after a remount, is incorrect.

I've suspected problems like this since Sébastien managed to have his
changes to mkfs.jffs work even though he didn't change the offset of the
nodes... I think the handling of partial node rewriting isn't working.

/ Alex