On Thursday 17 July 2008 22:24:56 Amit Sharma wrote:
> Hi
>
> We are using yaffs2 for onenand Samsung and we are finding all fsstress
> process goes to un-interruptable state.we checked grosslock count
> state by adding following line in dump_proc function of yaffs2
> buf += sprintf(buf, "lock status......... %d\n", dev->grossLock.count);
ALthough it could be something within yaffs, I suspect that your nand driver
is hanging.
>
> lock count value is -1 when it hangs.
>
> hang board memory usage
>
> 198 root 1000 S /bin/bash /fsstress_script.sh
> 201 root 456 S /fsstress -p 3 -n 10000 -d /tmp -l 0
> 390 root 148 S sleep 3600
> * 411 root 460 D /fsstress -p 3 -n 10000 -d /tmp -l 0
> 412 root 412 D /fsstress -p 3 -n 10000 -d /tmp -l 0
> 413 root 372 D /fsstress -p 3 -n 10000 -d /tmp -l 0
> * 414 root 280 R ps
> # free
> total used free shared
> buffers
> Mem: 62264 60412 *1852 * 0 8192
> Swap: 0 0 0
> Total: 62264 60412 1852
>
> commonly seen in all boards by using wchan
>
> 1. Yaffs_clear_inode
> 2. Yaffs_commit_write
> 3. Yaffs_lookup
>
> Any member can support me , i will thankful to you.
> thanks in advance
>
> we have tested it on many board 90%l are hanging.
(text/plain)