blk_dev[93].request_fn == NULL bug is still there

blk_dev[93].request_fn == NULL bug is still there

[Levi Khatskevitch]

> > I've just tested the new code, the doc2001 use count still gets 
>increasingly
> > negative on each access to /dev/mtd0. Like:
>
>Are you sure you rebuilt mtdchar.o?
>
>--
>dwmw2
>
>

Ok, I was wrong I used the old vmlinux file :-). Module use count is correct 
now.

Now I've just reproduced the blk_dev[93].request_fn == NULL bug in the 
latest code. Here is how:

1) build the kernel with nftl*.o build in and doc*.o as modules.

2) insmod docecc.o; insmod doc2001.o; insmod docprobe.o
you'll see:

M-Systems DiskOnChip driver. (C) 1999 Machine Vision Holdings, Inc.
DiskOnChip Millennium found at address 0xFFFC8000
Flash chip found: Manufacture ID: 98, Chip ID: E6 (Toshiba TC58V64AFT/DC)
1 flash chips found. Total DiskOnChip size: 8 Mbytes
Cannot calculate an NFTL geometry to match size of 0x37c0.
Using C:1019 H:14 S:1 (== 0x37ba sects)
nftla:<3>ll_rw_block: Trying to read nonexistent block-device 5d:00 (0)
unable to read partition table
...

where "ll_rw_block: Trying to read nonexistent block-device 5d:00 (0)" is 
caused by blk_dev[0x5d].request_fn == NULL
any subsequent access to /dev/nftla results in the same error.

at the same time nftl *is* registered to MAJOR# 93 according to:
[root@c1000s-dtd /]# cat /proc/devices
Character devices:
  1 mem
  2 pty
  3 ttyp
  4 ttyS
  5 cua
10 misc
90 mtd
128 ptm
136 pts

Block devices:
  1 ramdisk
  2 fd
  7 loop
93 nftl
^^^^^^^^^

Regards,
- Levi

_____________________________________________________________________________________
Get more from the Web.  FREE MSN Explorer download : http://explorer.msn.com



To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org

Re: blk_dev[93].request_fn == NULL bug is still there

[David Woodhouse]

On Thu, 30 Nov 2000, Levi Khatskevitch wrote:

> Ok, I was wrong I used the old vmlinux file :-). Module use count is correct
> now.

Hehe. I did _exactly_ the same thing :)

> Now I've just reproduced the blk_dev[93].request_fn == NULL bug in the
> latest code. Here is how:

I thought we'd found that this was because we were initialising the MTD
code before device_setup(). So I moved init_mtd() down a couple of lines
in init/main.c.

I'll try to reproduce this myself as soon as I get into the office.


-- 
dwmw2




To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org

Re: blk_dev[93].request_fn == NULL bug is still there

[David Woodhouse]

lkhatskevitch@hotmail.com said:
>  Now I've just reproduced the blk_dev[93].request_fn == NULL bug in
> the  latest code. Here is how:

Works for me now. Check you've rebuilt it all with the latest patch.

Note init/main.c should now look like this:


        /* Set up devices .. */
        device_setup();

#ifdef CONFIG_MTD
        init_mtd();
#endif


--
dwmw2




To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org

Re: blk_dev[93].request_fn == NULL bug is still there

[Levi Khatskevitch]

On Fri, 1 Dec 2000, David Woodhouse wrote:
>I thought we'd found that this was because we were initialising the MTD
>code before device_setup(). So I moved init_mtd() down a couple of lines
>in init/main.c.
>
>I'll try to reproduce this myself as soon as I get into the office.
>
>--
>dwmw2
>
>

BTW Initializing MTD before device_setup was different (actially worse), it 
caused blk_dev[xxx].plug_tq.routine == NULL, that contition is not checked 
by run_task_queue() and was causing kernel to crash.

- Levi
_____________________________________________________________________________________
Get more from the Web.  FREE MSN Explorer download : http://explorer.msn.com



To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org

Re: blk_dev[93].request_fn == NULL bug is still there

[David Woodhouse]

lkhatskevitch@hotmail.com said:
>  BTW Initializing MTD before device_setup was different (actially
> worse), it  caused blk_dev[xxx].plug_tq.routine == NULL, that
> contition is not checked  by run_task_queue() and was causing kernel
> to crash. 

Ah. Sorry, I'd misread it and hadn't noticed the difference. 
Now I'm confused. Did it not complain about failing to register the blkdev?


--
dwmw2




To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org