Messages in this thread | | | Date | Fri, 11 Dec 2009 08:34:46 +0200 | From | Pekka Enberg <> | Subject | Re: [PATCH 1/2] tracing: Define kmem_trace_alloc_notrace unconditionally |
| |
Li Zefan kirjoitti: > Pekka Enberg wrote: >> Li Zefan wrote: >>> Always define kmem_trace_alloc_{,node}_notrace(), otherwise >>> perf-kmem will show wrong stats ifndef CONFIG_KMEMTRACE, >>> because a kmalloc() memory allocation may be traced by >>> both trace_kmalloc() and trace_kmem_cache_alloc(). >>> >>> Signed-off-by: Li Zefan <lizf@cn.fujitsu.com> >> Did you check how much this will make kernel text bigger because of the >> inlining happening in kmem_cache_alloc_notrace()? >> > > I'm not sure I understood what you meant, but I'm not inlining > kmem_cache_alloc_notrace(), and instead I'm removing the inline > version in !CONFIG_KMEMTRACE case.
In SLUB, slab_alloc() will be inlined to kmem_cache_alloc_notrace() increasing mm/slub.o size so we don't want to define kmem_cache_alloc_notrace() unconditionally.
Pekka
| |