[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: fsck: i_blocks N, should be N/2



On 18-Oct-01 at 22:57, Andrew Morton (akpm zip com au) wrote:

> >    Inode XXX, i_blocks is 16, should be 8.  Fix<y>?

> In 2.4.12, Linus' kernel gained the new quota code, and the
> i_blocks accounting is all tied up into that. It's possible that
> something went wrong and we're double-accounting.  I'll check.
> 
> But I'm rather surprised that it happened on 2.4.10 as well.  Are
> you sure the merge/patch went OK?

Ahem.  I guess I know what happend, and it was my own fault.
I actually applied 2.4.10 patches to 2.4.11, and in a few places
where it did not apply I looked at the code and guessed what should
have been there.  This included manually taking the quota code out of
#ifdef NEW...QUOTA...  Possibly the errors appeared while I ran this
crippled 2.4.11 and then when I ran fsck under 2.4.12, they just
came up to the surface.

Anyway, I cannot reproduce it right now with proper 2.4.12.

Sorry for false alarm.  *If* I am wrong and the problem still
reappears I'll post a report.

> Could you please run:
> 
>      cd /usr/src/linux/fs/ext3
>      grep i_blocks *.c
> 
> The resulting output should NOT have any code which performs
> additions to or subtractions from inode->i_blocks.

it does not...

Sorry again

Eugene





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]