At 11:38 AM 6/20/2002 -0700, Trevin Beattie wrote:
>At 01:46 PM 6/19/2002 -0700, Konrad Schroder wrote:
>>>   needed = fsb + btofsb(...) + fsbtodb(...);
>>>   Whatever the units of fsb are (expected to be), btofsb given a number of
>>> frags, and fsbtodb gives a number of sectors.
>>>
>>> Is the last item a bug?  Since the next statement compares 'needed' to
>>
>>Yes, I think it is a bug.  I'll look at it tonight, though I think that
>>just putting the first btofsb inside the fsbtodb might take care of it.
>
>I just came along this function again from a different path, and it looks
>like my previous assessment of the terms may have been incorrect.
>lfs_truncate() calls lfs_reserve() with units of fsb (which, as you said,
>varies depending on the LFS version); this gets passed in turn to lfs_fits
>through the fsb parameter.  If that is the correct unit, then the third
>term in the sum must be incorrect, and the fix is to remove the fsbtodb()
>conversion.

I looked closer at the terms inside fsbtodb(), and found another possible
problem.  "howmany(fs->lfs_uinodes + 1, INOPB(fs))" and fs->lfs_segtabsz
have units of blocks, not fsb's.

>It would also mean that all of the functions lfs_reserve(), lfs_fits(), and
>lfs_availwait(); and the struct dlfs members lfs_bfree, lfs_avail, and
>lfs_ravail; are all in units of fsb instead of blocks or sectors.  Is that
>correct?

Add lfs_effnblocks to that list (struct lfs_inode_ext).

-----------------------
Trevin Beattie          "Do not meddle in the affairs of wizards,
trevin@xmission.com     for you are crunchy and good with ketchup."
      {:->                                     --unknown