On Thu, Oct 20, 2022 at 04:18:08PM +0300, Andy Shevchenko wrote:
On Wed, Oct 19, 2022 at 05:21:01PM -0700, Yury Norov wrote:
quoted
On Tue, Oct 18, 2022 at 04:00:23PM +0200, Alexander Lobakin wrote:
...
quoted
quoted
+static inline size_t bitmap_arr32_size(size_t nbits)
+{
+ return array_size(BITS_TO_U32(nbits), sizeof(u32));
To me this looks simpler: round_up(nbits, 32) / BITS_PER_BYTE.
Can you check which generates better code?
It's not only about better code, but also about overflow protection, which your
proposal is missing.
Can you explain how this may overflow?
#define __round_mask(x, y) ((__typeof__(x))((y)-1))
#define round_up(x, y) ((((x)-1) | __round_mask(x, y))+1)
Thanks,
Yury