Re: [PATCH 2/2] btrfs: optimize free space tree bitmap conversion

[Howard McLauchlan <hmclauchlan@fb.com>]

On 04/18/2018 03:26 PM, David Sterba wrote:
> On Wed, Apr 18, 2018 at 02:30:59PM -0700, Howard McLauchlan wrote:
>> Presently, convert_free_space_to_extents() does a linear scan of the
>> bitmap. We can speed this up with find_next_{bit,zero_bit}_le().
>>
>> This patch replaces the linear scan with find_next_{bit,zero_bit}_le().
>> Testing shows a 20-33% decrease in execution time for
>> convert_free_space_to_extents().
>>
>> Suggested-by: Omar Sandoval <osandov@osandov.com>
>> Signed-off-by: Howard McLauchlan <hmclauchlan@fb.com>
>> ---
>>
>> Since we change bitmap to be unsigned long, we have to do some casting for the
>> bitmap cursor. In le_bitmap_set() it makes sense to use u8, as we are doing
>> bit operations. Everywhere else, we're just using it for pointer arithmetic and
>> not directly accessing it, so char seems more appropriate.
> Ok, makes sense for just passing the pointers around. I'll add the text
> to changelog and apply the patch to next.
>
>> -		bit = !!le_test_bit(bitnr, bitmap);
> This is the last use of le_test_bit, so it can be removed (in another
> patch).
I just found an issue in this patch that should be fixed. I'll address
moving le_bitmap_set(), le_test_bit and send a V2 for both these patches.

Howard