Re: 2.6.0test9 Reiserfs boot time "buffer layer error at fs/buffer.c:431"

[lkml-031028@amos.mailshell.com]

On Tue, Oct 28, 2003 at 02:13:29PM -0800, Andrew Morton wrote:
> I've been waiting a year for someone who can reproduce this.

Looks like it's reproducible right now - I got exactly the same
message when I boot again later.

I'll try not to move things too much so I can test this more.
(You don't think it's critical, do you?)

> Are you using initrd?

Nope. Just plain old /boot/vmlinuz on a simple IDE disk with
a single ReiserFS partition and ReiserFS code compiled into
the kernel.

> Could you please add this patch, and send the new dmesg output?

Will do it gladly, when I'm back home tonight.

Thanks for everything,

--Amos

PS I'm not on lkml, so please keep me cc'ed about this.

> 
> 
>  25-akpm/fs/block_dev.c |   15 ++++++++++++---
>  1 files changed, 12 insertions(+), 3 deletions(-)
> 
> diff -puN fs/block_dev.c~a fs/block_dev.c
> --- 25/fs/block_dev.c~a	Tue Oct 28 14:11:20 2003
> +++ 25-akpm/fs/block_dev.c	Tue Oct 28 14:11:24 2003
> @@ -50,17 +50,26 @@ int set_blocksize(struct block_device *b
>  {
>  	int oldsize;
>  
> +	printk("%s: size=%d\n", __FUNCTION__, size);
> +
>  	/* Size must be a power of two, and between 512 and PAGE_SIZE */
> -	if (size > PAGE_SIZE || size < 512 || (size & (size-1)))
> +	if (size > PAGE_SIZE || size < 512 || (size & (size-1))) {
> +		printk("%s: EINVAL 1\n", __FUNCTION__);
>  		return -EINVAL;
> +	}
>  
>  	/* Size cannot be smaller than the size supported by the device */
> -	if (size < bdev_hardsect_size(bdev))
> +	if (size < bdev_hardsect_size(bdev)) {
> +		printk("%s: %d < %d\n", __FUNCTION__, size,
> +					bdev_hardsect_size(bdev));
>  		return -EINVAL;
> +	}
>  
>  	oldsize = bdev->bd_block_size;
> -	if (oldsize == size)
> +	if (oldsize == size) {
> +		printk("%s: %d OK\n", __FUNCTION__, size);
>  		return 0;
> +	}
>  
>  	/* Ok, we're actually changing the blocksize.. */
>  	sync_blockdev(bdev);
> 
> _
> 
> 
>