Le 02/07/2022 à 21:32, Andy Shevchenko a écrit :
On Sat, Jul 02, 2022 at 09:24:24PM +0200, Christophe JAILLET wrote:
Le 02/07/2022 à 20:54, Andy Shevchenko a écrit :
On Sat, Jul 02, 2022 at 08:29:09PM +0200, Christophe JAILLET wrote:
...
- memset(set->bitmap, 0, bitmap_size(num_ssid, num_id));
+ memset(set->bitmap, 0, idset_bitmap_size(num_ssid, num_id));
Why not to use bitmap_zero()?
...
- memset(set->bitmap, 0xff, bitmap_size(set->num_ssid, set->num_id));
+ memset(set->bitmap, 0xff, idset_bitmap_size(set->num_ssid, set->num_id));
Why not to use bitmap_fill() ?
For this initial step, I wanted to keep changes as minimal as possible (i.e
just function renaming)
In fact, I plan to send a follow-up patch on this file.
This would remove the newly renamed idset_bitmap_size() function, use the
bitmap API directly (as you pointed-out) with
"set->num_ssid * set->num_id" as size.
It is already done this way in idset_is_empty(), so it would be more
consistent.
If the serie needs a v2 (or if required), I can add an additional 5th patch
for it. Otherwise it will send separatly later.
If you use bitmap APIs as I suggested above as the first patch, the rest will
have less unneeded churn, no?
Makes sense.
I'll wait for some other potential comments 1 day or 2 and send a v2
with the simplification you propose as an initial step.
Thanks for your feed-back.
CJ
--
dm-devel mailing list
dm-devel@xxxxxxxxxx
https://listman.redhat.com/mailman/listinfo/dm-devel
