Merge tag 'bitmap-6.1-rc1' of https://github.com/norov/linux

Pull bitmap updates from Yury Norov:

 - Fix unsigned comparison to -1 in CPUMAP_FILE_MAX_BYTES (Phil Auld)

 - cleanup nr_cpu_ids vs nr_cpumask_bits mess (me)

   This series cleans that mess and adds new config FORCE_NR_CPUS that
   allows to optimize cpumask subsystem if the number of CPUs is known
   at compile-time.

 - optimize find_bit() functions (me)

   Reworks find_bit() functions based on new FIND_{FIRST,NEXT}_BIT()
   macros.

 - add find_nth_bit() (me)

   Adds find_nth_bit(), which is ~70 times faster than bitcounting with
   for_each() loop:

	for_each_set_bit(bit, mask, size)
		if (n-- == 0)
			return bit;

   Also adds bitmap_weight_and() to let people replace this pattern:

	tmp = bitmap_alloc(nbits);
	bitmap_and(tmp, map1, map2, nbits);
	weight = bitmap_weight(tmp, nbits);
	bitmap_free(tmp);

   with a single bitmap_weight_and() call.

 - repair cpumask_check() (me)

   After switching cpumask to use nr_cpu_ids, cpumask_check() started
   generating many false-positive warnings. This series fixes it.

 - Add for_each_cpu_andnot() and for_each_cpu_andnot() (Valentin
   Schneider)

   Extends the API with one more function and applies it in sched/core.

* tag 'bitmap-6.1-rc1' of https://github.com/norov/linux: (28 commits)
  sched/core: Merge cpumask_andnot()+for_each_cpu() into for_each_cpu_andnot()
  lib/test_cpumask: Add for_each_cpu_and(not) tests
  cpumask: Introduce for_each_cpu_andnot()
  lib/find_bit: Introduce find_next_andnot_bit()
  cpumask: fix checking valid cpu range
  lib/bitmap: add tests for for_each() loops
  lib/find: optimize for_each() macros
  lib/bitmap: introduce for_each_set_bit_wrap() macro
  lib/find_bit: add find_next{,_and}_bit_wrap
  cpumask: switch for_each_cpu{,_not} to use for_each_bit()
  net: fix cpu_max_bits_warn() usage in netif_attrmask_next{,_and}
  cpumask: add cpumask_nth_{,and,andnot}
  lib/bitmap: remove bitmap_ord_to_pos
  lib/bitmap: add tests for find_nth_bit()
  lib: add find_nth{,_and,_andnot}_bit()
  lib/bitmap: add bitmap_weight_and()
  lib/bitmap: don't call __bitmap_weight() in kernel code
  tools: sync find_bit() implementation
  lib/find_bit: optimize find_next_bit() functions
  lib/find_bit: create find_first_zero_bit_le()
  ...

1 files changed, 16 insertions, 24 deletions

diff --git a/lib/cpumask.c b/lib/cpumask.c
index f0ae119be8c4..c7c392514fd3 100644
--- a/lib/cpumask.c
+++ b/lib/cpumask.c
@@ -128,23 +128,21 @@ unsigned int cpumask_local_spread(unsigned int i, int node)
 	i %= num_online_cpus();
 
 	if (node == NUMA_NO_NODE) {
-		for_each_cpu(cpu, cpu_online_mask)
-			if (i-- == 0)
-				return cpu;
+		cpu = cpumask_nth(i, cpu_online_mask);
+		if (cpu < nr_cpu_ids)
+			return cpu;
 	} else {
 		/* NUMA first. */
-		for_each_cpu_and(cpu, cpumask_of_node(node), cpu_online_mask)
-			if (i-- == 0)
-				return cpu;
-
-		for_each_cpu(cpu, cpu_online_mask) {
-			/* Skip NUMA nodes, done above. */
-			if (cpumask_test_cpu(cpu, cpumask_of_node(node)))
-				continue;
-
-			if (i-- == 0)
-				return cpu;
-		}
+		cpu = cpumask_nth_and(i, cpu_online_mask, cpumask_of_node(node));
+		if (cpu < nr_cpu_ids)
+			return cpu;
+
+		i -= cpumask_weight_and(cpu_online_mask, cpumask_of_node(node));
+
+		/* Skip NUMA nodes, done above. */
+		cpu = cpumask_nth_andnot(i, cpu_online_mask, cpumask_of_node(node));
+		if (cpu < nr_cpu_ids)
+			return cpu;
 	}
 	BUG();
 }
@@ -168,10 +166,8 @@ unsigned int cpumask_any_and_distribute(const struct cpumask *src1p,
 	/* NOTE: our first selection will skip 0. */
 	prev = __this_cpu_read(distribute_cpu_mask_prev);
 
-	next = cpumask_next_and(prev, src1p, src2p);
-	if (next >= nr_cpu_ids)
-		next = cpumask_first_and(src1p, src2p);
-
+	next = find_next_and_bit_wrap(cpumask_bits(src1p), cpumask_bits(src2p),
+					nr_cpumask_bits, prev + 1);
 	if (next < nr_cpu_ids)
 		__this_cpu_write(distribute_cpu_mask_prev, next);
 
@@ -185,11 +181,7 @@ unsigned int cpumask_any_distribute(const struct cpumask *srcp)
 
 	/* NOTE: our first selection will skip 0. */
 	prev = __this_cpu_read(distribute_cpu_mask_prev);
-
-	next = cpumask_next(prev, srcp);
-	if (next >= nr_cpu_ids)
-		next = cpumask_first(srcp);
-
+	next = find_next_bit_wrap(cpumask_bits(srcp), nr_cpumask_bits, prev + 1);
 	if (next < nr_cpu_ids)
 		__this_cpu_write(distribute_cpu_mask_prev, next);