|
Sometimes gcc mysteriously doesn't inline
very small functions we expect to be inlined. See
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66122
With this .config:
http://busybox.net/~vda/kernel_config_OPTIMIZE_INLINING_and_Os,
the following functions get deinlined many times.
Examples of disassembly:
<get_unaligned_be16> (12 copies, 51 calls):
66 8b 07 mov (%rdi),%ax
55 push %rbp
48 89 e5 mov %rsp,%rbp
86 e0 xchg %ah,%al
5d pop %rbp
c3 retq
<get_unaligned_be32> (12 copies, 135 calls):
8b 07 mov (%rdi),%eax
55 push %rbp
48 89 e5 mov %rsp,%rbp
0f c8 bswap %eax
5d pop %rbp
c3 retq
<get_unaligned_be64> (2 copies, 20 calls):
48 8b 07 mov (%rdi),%rax
55 push %rbp
48 89 e5 mov %rsp,%rbp
48 0f c8 bswap %rax
5d pop %rbp
c3 retq
<__swab16p> (16 copies, 146 calls):
55 push %rbp
89 f8 mov %edi,%eax
86 e0 xchg %ah,%al
48 89 e5 mov %rsp,%rbp
5d pop %rbp
c3 retq
<__swab32p> (43 copies, ~560 calls):
55 push %rbp
89 f8 mov %edi,%eax
0f c8 bswap %eax
48 89 e5 mov %rsp,%rbp
5d pop %rbp
c3 retq
<__swab64p> (21 copies, 119 calls):
55 push %rbp
48 89 f8 mov %rdi,%rax
48 0f c8 bswap %rax
48 89 e5 mov %rsp,%rbp
5d pop %rbp
c3 retq
<__swab32s> (6 copies, 47 calls):
8b 07 mov (%rdi),%eax
55 push %rbp
48 89 e5 mov %rsp,%rbp
0f c8 bswap %eax
89 07 mov %eax,(%rdi)
5d pop %rbp
c3 retq
This patch fixes this via s/inline/__always_inline/.
Code size decrease after the patch is ~4.5k:
text data bss dec hex filename
92202377 20826112 36417536 149446025 8e85d89 vmlinux
92197848 20826112 36417536 149441496 8e84bd8 vmlinux5_swap_after
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
Acked-by: Ingo Molnar <mingo@kernel.org>
Cc: Thomas Graf <tgraf@suug.ch>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
Since GCC 4.4, there have been __builtin_bswap32() and __builtin_bswap16()
intrinsics. A __builtin_bswap16() came a little later (4.6 for PowerPC,
48 for other platforms).
By using these instead of the inline assembler that most architectures
have in their __arch_swabXX() macros, we let the compiler see what's
actually happening. The resulting code should be at least as good, and
much *better* in the cases where it can be combined with a nearby load
or store, using a load-and-byteswap or store-and-byteswap instruction
(e.g. lwbrx/stwbrx on PowerPC, movbe on Atom).
When GCC is sufficiently recent *and* the architecture opts in to using
the intrinsics by setting CONFIG_ARCH_USE_BUILTIN_BSWAP, they will be
used in preference to the __arch_swabXX() macros. An architecture which
does not set ARCH_USE_BUILTIN_BSWAP will continue to use its own
hand-crafted macros.
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Acked-by: H. Peter Anvin <hpa@linux.intel.com>
|