summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorRasmus Villemoes <linux@rasmusvillemoes.dk>2016-05-19 17:10:52 -0700
committerLinus Torvalds <torvalds@linux-foundation.org>2016-05-19 19:12:14 -0700
commitd64e85d3e1c59c3664b9ec1183052ec4641ea1e2 (patch)
tree88906c1e88f34ffb1c726405118e84051c103dd2
parent0139aa7b7fa12ceef095d99dc36606a5b10ab83a (diff)
downloadlinux-d64e85d3e1c59c3664b9ec1183052ec4641ea1e2.tar.gz
linux-d64e85d3e1c59c3664b9ec1183052ec4641ea1e2.tar.bz2
linux-d64e85d3e1c59c3664b9ec1183052ec4641ea1e2.zip
compiler.h: add support for malloc attribute
gcc as far back as at least 3.04 documents the function attribute __malloc__. Add a shorthand for attaching that to a function declaration. This was also suggested by Andi Kleen way back in 2002 [1], but didn't get applied, perhaps because gcc at that time generated the exact same code with and without this attribute. This attribute tells the compiler that the return value (if non-NULL) can be assumed not to alias any other valid pointers at the time of the call. Please note that the documentation for a range of gcc versions (starting from around 4.7) contained a somewhat confusing and self-contradicting text: The malloc attribute is used to tell the compiler that a function may be treated as if any non-NULL pointer it returns cannot alias any other pointer valid when the function returns and *that the memory has undefined content*. [...] Standard functions with this property include malloc and *calloc*. (emphasis mine). The intended meaning has later been clarified [2]: This tells the compiler that a function is malloc-like, i.e., that the pointer P returned by the function cannot alias any other pointer valid when the function returns, and moreover no pointers to valid objects occur in any storage addressed by P. What this means is that we can apply the attribute to kmalloc and friends, and it is ok for the returned memory to have well-defined contents (__GFP_ZERO). But it is not ok to apply it to kmemdup(), nor to other functions which both allocate and possibly initialize the memory with existing pointers. So unless someone is doing something pretty perverted kstrdup() should also be a fine candidate. [1] http://thread.gmane.org/gmane.linux.kernel/57172 [2] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56955 Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk> Cc: Christoph Lameter <cl@linux.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: David Rientjes <rientjes@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Andi Kleen <ak@linux.intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
-rw-r--r--include/linux/compiler-gcc.h1
-rw-r--r--include/linux/compiler.h4
2 files changed, 5 insertions, 0 deletions
diff --git a/include/linux/compiler-gcc.h b/include/linux/compiler-gcc.h
index 3d5202eda22f..e2949397c19b 100644
--- a/include/linux/compiler-gcc.h
+++ b/include/linux/compiler-gcc.h
@@ -142,6 +142,7 @@
#if GCC_VERSION >= 30400
#define __must_check __attribute__((warn_unused_result))
+#define __malloc __attribute__((__malloc__))
#endif
#if GCC_VERSION >= 40000
diff --git a/include/linux/compiler.h b/include/linux/compiler.h
index b5ff9881bef8..793c0829e3a3 100644
--- a/include/linux/compiler.h
+++ b/include/linux/compiler.h
@@ -357,6 +357,10 @@ static __always_inline void __write_once_size(volatile void *p, void *res, int s
#define __deprecated_for_modules
#endif
+#ifndef __malloc
+#define __malloc
+#endif
+
/*
* Allow us to avoid 'defined but not used' warnings on functions and data,
* as well as force them to be emitted to the assembly file.