Numonyx NOR and chip->mutex bug?

[Michael Cashwell]

On Feb 5, 2011, at 2:19 PM, Michael Cashwell wrote:

>> I won't worry about this until you test my patch and still have a problem :) Please test it.
> 
> My plan is to retest your patch against a pristine copy of the file and report back. Unfortunately this has to wait until Monday as I don't have access to the hardware over the weekend.

OK, I retested Joakim's patch from 4-feb on a clean source file and it doesn't help my problem:

[cashwell at mec-fedora14 linux-2.6.35.7-arm-gum]$ cat ../joakim.patch | patch -p1
patching file drivers/mtd/chips/cfi_cmdset_0001.c
Hunk #1 succeeded at 1227 (offset -18 lines).
Hunk #2 succeeded at 1266 (offset -18 lines).

(The offsets are just that we're on slightly different kernel versions, I assume. It applies OK.)

#### Exporting kernel linux-2.6.35.7-arm-gum to netboot.
Image Name:   2.6.35.7-gum1
Created:      Mon Feb  7 08:22:31 2011
Image Type:   ARM Linux Kernel Image (uncompressed)
Data Size:    1919908 Bytes = 1874.91 kB = 1.83 MB
Load Address: 0xA1000000
Entry Point:  0xA1000000

Booted and then did ubiformat, mount, rsync of about 20MB, and then during the umount of that file system:

UBI error: ubi_io_write: error -30 while writing 1968 bytes to PEB 17:37528, written 0 bytes
UBI warning: ubi_eba_write_leb: failed to write data to PEB 17
UBI warning: ubi_ro_mode: switch to read-only mode

which I'd previously traced to the -EROFS return in do_write_buffer because it thinks it got a block-locked error during write. I've also seen that sometimes see the bad Vpp error too. (Why one of those has a printk and not the other I don't know.)

-Mike