Message ID | 20250326155200.39895-1-yury.norov@gmail.com |
---|---|
State | New |
Headers | show |
Series | wifi: carl9170: micro-optimize carl9170_tx_shift_bm() | expand |
On Wed, Mar 26, 2025 at 09:00:33PM +0100, Christian Lamparter wrote: > Hi, > > On Wed, Mar 26, 2025 at 4:52 PM Yury Norov <yury.norov@gmail.com> wrote: > > > > The function calls bitmap_empty() just before find_first_bit(). Both > > functions are O(N). Because find_first_bit() returns >= nbits in case of > > empty bitmap, the bitmap_empty() test may be avoided. > > > > I looked up bitmap_empty(): > <https://web.git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/include/linux/bitmap.h#n423> > > apart from the small_const_nbits stuff (which carl9170 likely does not qualify > for since from what I remember it's a 128bits bitmap) the function just does: > > | return find_first_bit(src, nbits) == nbits; > > so yes, find_first_bit runs twice with same parameters... Unless the > compiler is smart > enough to detect this and (re-)use the intermediate result later. But > I haven't check > if this is the case with any current, old or future compilers. Has anyone? > > Anyway, Sure. > > > Signed-off-by: Yury Norov <yury.norov@gmail.com> > > Acked-by: Christian Lamparter <chunkeey@gmail.com> Thanks, Chrustian. So, how is that supposed to be merged? I can move it with bitmap-for-next, unless there's no better branch. Thanks, Yury
On 4/27/2025 8:25 AM, Yury Norov wrote: > On Wed, Mar 26, 2025 at 09:00:33PM +0100, Christian Lamparter wrote: >> Hi, >> >> On Wed, Mar 26, 2025 at 4:52 PM Yury Norov <yury.norov@gmail.com> wrote: >>> >>> The function calls bitmap_empty() just before find_first_bit(). Both >>> functions are O(N). Because find_first_bit() returns >= nbits in case of >>> empty bitmap, the bitmap_empty() test may be avoided. >>> >> >> I looked up bitmap_empty(): >> <https://web.git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/include/linux/bitmap.h#n423> >> >> apart from the small_const_nbits stuff (which carl9170 likely does not qualify >> for since from what I remember it's a 128bits bitmap) the function just does: >> >> | return find_first_bit(src, nbits) == nbits; >> >> so yes, find_first_bit runs twice with same parameters... Unless the >> compiler is smart >> enough to detect this and (re-)use the intermediate result later. But >> I haven't check >> if this is the case with any current, old or future compilers. Has anyone? >> >> Anyway, Sure. >> >>> Signed-off-by: Yury Norov <yury.norov@gmail.com> >> >> Acked-by: Christian Lamparter <chunkeey@gmail.com> > > Thanks, Chrustian. So, how is that supposed to be merged? > I can move it with bitmap-for-next, unless there's no better > branch. > > Thanks, > Yury > Yury, did you take this? If not, I'll take it through the ath tree.
On Tue, May 20, 2025 at 09:24:14AM -0700, Jeff Johnson wrote: > On 4/27/2025 8:25 AM, Yury Norov wrote: > > On Wed, Mar 26, 2025 at 09:00:33PM +0100, Christian Lamparter wrote: > >> Hi, > >> > >> On Wed, Mar 26, 2025 at 4:52 PM Yury Norov <yury.norov@gmail.com> wrote: > >>> > >>> The function calls bitmap_empty() just before find_first_bit(). Both > >>> functions are O(N). Because find_first_bit() returns >= nbits in case of > >>> empty bitmap, the bitmap_empty() test may be avoided. > >>> > >> > >> I looked up bitmap_empty(): > >> <https://web.git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/include/linux/bitmap.h#n423> > >> > >> apart from the small_const_nbits stuff (which carl9170 likely does not qualify > >> for since from what I remember it's a 128bits bitmap) the function just does: > >> > >> | return find_first_bit(src, nbits) == nbits; > >> > >> so yes, find_first_bit runs twice with same parameters... Unless the > >> compiler is smart > >> enough to detect this and (re-)use the intermediate result later. But > >> I haven't check > >> if this is the case with any current, old or future compilers. Has anyone? > >> > >> Anyway, Sure. > >> > >>> Signed-off-by: Yury Norov <yury.norov@gmail.com> > >> > >> Acked-by: Christian Lamparter <chunkeey@gmail.com> > > > > Thanks, Chrustian. So, how is that supposed to be merged? > > I can move it with bitmap-for-next, unless there's no better > > branch. > > > > Thanks, > > Yury > > > > Yury, did you take this? > If not, I'll take it through the ath tree. No. Please take with ath.
diff --git a/drivers/net/wireless/ath/carl9170/tx.c b/drivers/net/wireless/ath/carl9170/tx.c index 0226c31a6cae..b7717f9e1e9b 100644 --- a/drivers/net/wireless/ath/carl9170/tx.c +++ b/drivers/net/wireless/ath/carl9170/tx.c @@ -366,8 +366,7 @@ static void carl9170_tx_shift_bm(struct ar9170 *ar, if (WARN_ON_ONCE(off >= CARL9170_BAW_BITS)) return; - if (!bitmap_empty(tid_info->bitmap, off)) - off = find_first_bit(tid_info->bitmap, off); + off = min(off, find_first_bit(tid_info->bitmap, off)); tid_info->bsn += off; tid_info->bsn &= 0x0fff;
The function calls bitmap_empty() just before find_first_bit(). Both functions are O(N). Because find_first_bit() returns >= nbits in case of empty bitmap, the bitmap_empty() test may be avoided. Signed-off-by: Yury Norov <yury.norov@gmail.com> --- drivers/net/wireless/ath/carl9170/tx.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)