Re: [PATCH -V2] ext4: unmap the underlying metadata when allocating blocks via fallocate

[Eric Sandeen <sandeen@redhat.com>]

tytso@mit.edu wrote:
> On Mon, Jan 25, 2010 at 10:42:05AM +0530, Aneesh Kumar K. V wrote:
>> My goal was to see if we can drop the BH_New flag when returning
>> preallocated blocks. But now i look at it again i guess you don't need
>> to take this patch. We cannot unmap underlying meta data during
>> fallocate because even if we remove the old mapping for the blocks,
>> we could reboot and later somebody(e2fsprogs) can directly read the blocks
>> and create a bh with old data. So i guess we should be doing
>> unmap_underlying_blocks when we are returning blocks to userspace or
>> when we actually zero out blocks.
> 
> OK, so I'll drop this patch from the patch queue, and will do a full
> test run (XFSQA -g auto using both 1k and 4k block sizes) of the
> following patches:
> 
> Aneesh Kumar K.V (3):
>       ext4: Handle -EDQUOT error on write
>       ext4: Fix quota accounting error with fallocate
>       ext4: Drop EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE flag

You may wish to add the patch from Jiaying (via me) as well, to handle
the blocks-past-eof case, if it's had sufficient review.

-Eric

> Of all of the patches in the ext4 patch queue, these seem to be the
> ones which need to be urgently pushed to Linus before 2.6.33.
> 
> Any objections, or nominations for other patches that should go in
> before 2.6.33, please speak up now...
> 
> 						- Ted