aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorLinus Torvalds <torvalds@linux-foundation.org>2015-08-31 21:37:33 -0400
committerLinus Torvalds <torvalds@linux-foundation.org>2015-08-31 21:37:33 -0400
commit5757bd6157f523ff0d448a4ec93938523ca441a9 (patch)
tree713885bff513324e1935874653e546162cdbe540
parent7073bc66126e3ab742cce9416ad6b4be8b03c4f7 (diff)
parentaccd0b9ec015d611eb7783dd86f1bb31bf8d62ab (diff)
Merge branch 'core-types-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull inlining tuning from Ingo Molnar: "A handful of inlining optimizations inspired by x86 work but applicable in general" * 'core-types-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: jiffies: Force inlining of {m,u}msecs_to_jiffies() x86/hweight: Force inlining of __arch_hweight{32,64}() linux/bitmap: Force inlining of bitmap weight functions
-rw-r--r--arch/x86/include/asm/arch_hweight.h13
-rw-r--r--include/linux/bitmap.h2
-rw-r--r--include/linux/bitops.h6
-rw-r--r--include/linux/jiffies.h4
4 files changed, 14 insertions, 11 deletions
diff --git a/arch/x86/include/asm/arch_hweight.h b/arch/x86/include/asm/arch_hweight.h
index 9686c3d9ff73..259a7c1ef709 100644
--- a/arch/x86/include/asm/arch_hweight.h
+++ b/arch/x86/include/asm/arch_hweight.h
@@ -21,7 +21,7 @@
21 * ARCH_HWEIGHT_CFLAGS in <arch/x86/Kconfig> for the respective 21 * ARCH_HWEIGHT_CFLAGS in <arch/x86/Kconfig> for the respective
22 * compiler switches. 22 * compiler switches.
23 */ 23 */
24static inline unsigned int __arch_hweight32(unsigned int w) 24static __always_inline unsigned int __arch_hweight32(unsigned int w)
25{ 25{
26 unsigned int res = 0; 26 unsigned int res = 0;
27 27
@@ -42,20 +42,23 @@ static inline unsigned int __arch_hweight8(unsigned int w)
42 return __arch_hweight32(w & 0xff); 42 return __arch_hweight32(w & 0xff);
43} 43}
44 44
45#ifdef CONFIG_X86_32
45static inline unsigned long __arch_hweight64(__u64 w) 46static inline unsigned long __arch_hweight64(__u64 w)
46{ 47{
47 unsigned long res = 0;
48
49#ifdef CONFIG_X86_32
50 return __arch_hweight32((u32)w) + 48 return __arch_hweight32((u32)w) +
51 __arch_hweight32((u32)(w >> 32)); 49 __arch_hweight32((u32)(w >> 32));
50}
52#else 51#else
52static __always_inline unsigned long __arch_hweight64(__u64 w)
53{
54 unsigned long res = 0;
55
53 asm (ALTERNATIVE("call __sw_hweight64", POPCNT64, X86_FEATURE_POPCNT) 56 asm (ALTERNATIVE("call __sw_hweight64", POPCNT64, X86_FEATURE_POPCNT)
54 : "="REG_OUT (res) 57 : "="REG_OUT (res)
55 : REG_IN (w)); 58 : REG_IN (w));
56#endif /* CONFIG_X86_32 */
57 59
58 return res; 60 return res;
59} 61}
62#endif /* CONFIG_X86_32 */
60 63
61#endif 64#endif
diff --git a/include/linux/bitmap.h b/include/linux/bitmap.h
index ea17cca9e685..9653fdb76a42 100644
--- a/include/linux/bitmap.h
+++ b/include/linux/bitmap.h
@@ -295,7 +295,7 @@ static inline int bitmap_full(const unsigned long *src, unsigned int nbits)
295 return find_first_zero_bit(src, nbits) == nbits; 295 return find_first_zero_bit(src, nbits) == nbits;
296} 296}
297 297
298static inline int bitmap_weight(const unsigned long *src, unsigned int nbits) 298static __always_inline int bitmap_weight(const unsigned long *src, unsigned int nbits)
299{ 299{
300 if (small_const_nbits(nbits)) 300 if (small_const_nbits(nbits))
301 return hweight_long(*src & BITMAP_LAST_WORD_MASK(nbits)); 301 return hweight_long(*src & BITMAP_LAST_WORD_MASK(nbits));
diff --git a/include/linux/bitops.h b/include/linux/bitops.h
index 297f5bda4fdf..e63553386ae7 100644
--- a/include/linux/bitops.h
+++ b/include/linux/bitops.h
@@ -57,7 +57,7 @@ extern unsigned long __sw_hweight64(__u64 w);
57 (bit) < (size); \ 57 (bit) < (size); \
58 (bit) = find_next_zero_bit((addr), (size), (bit) + 1)) 58 (bit) = find_next_zero_bit((addr), (size), (bit) + 1))
59 59
60static __inline__ int get_bitmask_order(unsigned int count) 60static inline int get_bitmask_order(unsigned int count)
61{ 61{
62 int order; 62 int order;
63 63
@@ -65,7 +65,7 @@ static __inline__ int get_bitmask_order(unsigned int count)
65 return order; /* We could be slightly more clever with -1 here... */ 65 return order; /* We could be slightly more clever with -1 here... */
66} 66}
67 67
68static __inline__ int get_count_order(unsigned int count) 68static inline int get_count_order(unsigned int count)
69{ 69{
70 int order; 70 int order;
71 71
@@ -75,7 +75,7 @@ static __inline__ int get_count_order(unsigned int count)
75 return order; 75 return order;
76} 76}
77 77
78static inline unsigned long hweight_long(unsigned long w) 78static __always_inline unsigned long hweight_long(unsigned long w)
79{ 79{
80 return sizeof(w) == 4 ? hweight32(w) : hweight64(w); 80 return sizeof(w) == 4 ? hweight32(w) : hweight64(w);
81} 81}
diff --git a/include/linux/jiffies.h b/include/linux/jiffies.h
index 535fd3bb1ba8..1ba48a18c1d7 100644
--- a/include/linux/jiffies.h
+++ b/include/linux/jiffies.h
@@ -351,7 +351,7 @@ static inline unsigned long _msecs_to_jiffies(const unsigned int m)
351 * directly here and from __msecs_to_jiffies() in the case where 351 * directly here and from __msecs_to_jiffies() in the case where
352 * constant folding is not possible. 352 * constant folding is not possible.
353 */ 353 */
354static inline unsigned long msecs_to_jiffies(const unsigned int m) 354static __always_inline unsigned long msecs_to_jiffies(const unsigned int m)
355{ 355{
356 if (__builtin_constant_p(m)) { 356 if (__builtin_constant_p(m)) {
357 if ((int)m < 0) 357 if ((int)m < 0)
@@ -405,7 +405,7 @@ static inline unsigned long _usecs_to_jiffies(const unsigned int u)
405 * directly here and from __msecs_to_jiffies() in the case where 405 * directly here and from __msecs_to_jiffies() in the case where
406 * constant folding is not possible. 406 * constant folding is not possible.
407 */ 407 */
408static inline unsigned long usecs_to_jiffies(const unsigned int u) 408static __always_inline unsigned long usecs_to_jiffies(const unsigned int u)
409{ 409{
410 if (__builtin_constant_p(u)) { 410 if (__builtin_constant_p(u)) {
411 if (u > jiffies_to_usecs(MAX_JIFFY_OFFSET)) 411 if (u > jiffies_to_usecs(MAX_JIFFY_OFFSET))