Skip to contentSkip to conversation

sched: add the inline keyword to the prototype.

PATCH In review Il Han 3 participants 4 messages Nov 15, 2011 – Nov 16, 2011

lore
Share & tools
lore

Il Han

author
#0 lore raw

PATCH 1/1 sched: add the inline keyword to the prototype.

The prototype has to agree with the definition.
Add the inline keyword to the prototype also

---
 kernel/sched_fair.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)
kernel/sched_fair.c +2 -2
index 5c9e679..c7c93d2 100644
@@ -306,7 +306,7 @@ find_matching_se(struct sched_entity **se, struct sched_entity **pse)
 
 #endif	/* CONFIG_FAIR_GROUP_SCHED */
 
-static void account_cfs_rq_runtime(struct cfs_rq *cfs_rq,
+static __always_inline void account_cfs_rq_runtime(struct cfs_rq *cfs_rq,
 				   unsigned long delta_exec);
 
 /**************************************************************
@@ -1052,7 +1052,7 @@ static void clear_buddies(struct cfs_rq *cfs_rq, struct sched_entity *se)
 		__clear_buddies_skip(se);
 }
 
-static void return_cfs_rq_runtime(struct cfs_rq *cfs_rq);
+static __always_inline void return_cfs_rq_runtime(struct cfs_rq *cfs_rq);
 
 static void
 dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags)
Signature · 1 line
1.7.4.1
#1 lore raw
Quoted 3 lines · Il Han

The prototype has to agree with the definition.
Add the inline keyword to the prototype also

Hmm... this would then cause the prototypes to disagree in the !CONFIG_CFS_BANDWIDTH case.

Can you update?

Quoted 27 lines
---
  kernel/sched_fair.c |    4 ++--
  1 files changed, 2 insertions(+), 2 deletions(-)
kernel/sched_fair.c +2 -2
index 5c9e679..c7c93d2 100644
@@ -306,7 +306,7 @@ find_matching_se(struct sched_entity **se, struct sched_entity **pse)
 
  #endif	/* CONFIG_FAIR_GROUP_SCHED */
 
-static void account_cfs_rq_runtime(struct cfs_rq *cfs_rq,
+static __always_inline void account_cfs_rq_runtime(struct cfs_rq *cfs_rq,
  				   unsigned long delta_exec);
 
  /**************************************************************
@@ -1052,7 +1052,7 @@ static void clear_buddies(struct cfs_rq *cfs_rq, struct sched_entity *se)
  		__clear_buddies_skip(se);
  }
 
-static void return_cfs_rq_runtime(struct cfs_rq *cfs_rq);
+static __always_inline void return_cfs_rq_runtime(struct cfs_rq *cfs_rq);
 
  static void
  dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags)
#2 lore raw
Quoted 7 lines · Paul Turner
Quoted 3 lines · Il Han

The prototype has to agree with the definition.
Add the inline keyword to the prototype also

Hmm... this would then cause the prototypes to disagree in the !CONFIG_CFS_BANDWIDTH case.

I'm very sure I've compiled a lot of kernels yesterday with various permutations including FAIR_GROUP && !CFS_BANDWIDTH and I'm also very sure I haven't actually seen any complaints like this.

It this some new fangled GCC 'feature'? I'm currently still on 4.5 which admittedly is somewhat stale.

#3 lore raw
Quoted 9 lines · Peter Zijlstra
Quoted 7 lines · Paul Turner
Quoted 3 lines · Il Han

The prototype has to agree with the definition.
Add the inline keyword to the prototype also

Hmm... this would then cause the prototypes to disagree in the !CONFIG_CFS_BANDWIDTH case.

Quoted 6 lines

I'm very sure I've compiled a lot of kernels yesterday with various permutations including FAIR_GROUP&& !CFS_BANDWIDTH and I'm also very sure I haven't actually seen any complaints like this.

It this some new fangled GCC 'feature'? I'm currently still on 4.5 which admittedly is somewhat stale.

gcc (even 4.6) won't give warnings either way about the inline attribute on a forward declaration not matching in either direction. I would imagine this was found by code-inspection.

I was only pointing out with this we have:

#if CONFIG_CFS_BANDWIDTH
static __always_inline xxx() { stuff; } [1]
#else
static xxx() {} [2]
#endif

And that by only changing the pre-declaration to match [1] it now similarly (silently) would not match [2] in the !CONFIG case.

- Paul