Re: [PATCH] x86: Align jump targets to 1 byte boundaries

[Ingo Molnar <mingo@kernel.org>]

* Markus Trippelsdorf <markus@trippelsdorf.de> wrote:

> On 2015.04.14 at 07:38 +0200, Ingo Molnar wrote:
> > 
> > Just to make sure, could you please also apply the 3 alignment patches 
> > attached below? There's a lot of noise from extra alignment.
> 
> Here's an updated table:
> 
>    text    data     bss     dec     filename
>    8746230  970072  802816 10519118 ./vmlinux gcc-5 (lto) 
>    9202488  978512  811008 10992008 ./vmlinux gcc-5
>    8036915  970296  802816 9810027  ./vmlinux gcc-5 (lto -fno-guess-branch-probability)
>    8593615  978512  811008 10383135 ./vmlinux gcc-5 (-fno-guess-branch-probability)
>    8202614  970072  802816 9975502  ./vmlinux gcc-5 (lto + Ingo's patch)
>    8801016  978512  811008 10590536 ./vmlinux gcc-5 (Ingo's patch) 
>    8733943  952088  798720 10484751 ./vmlinux gcc-5 (lto + -malign-data=abi)
>    9186105  958320  806912 10951337 ./vmlinux gcc-5 (-malign-data=abi)
>    8190327  952088  798720 9941135  ./vmlinux gcc-5 (lto + Ingo's patch + -malign-data=abi)
>    8784633  958320  806912 10549865 ./vmlinux gcc-5 (Ingo's patch + -malign-data=abi) 
> 
> For the "lto + Ingo's patch + -malign-data=abi" combination there is a
> 10% text size reduction.

Lets rename "Ingo's patch" to "code_align=1". The interesting one 
would be to compare:

           code_align=1 + -fno-guess-branch-probability
vs.
     lto + code_align=1 + -fno-guess-branch-probability

Or rather, you could try "Ingo's combo patch" further below, with and 
without LTO.

I'd expect LTO to still be in the 5% reduction range.

> -malign-data is a new option for gcc-5 that controls how the 
> compiler aligns variables. "abi" aligns variables according to psABI 
> and give the tightest packing.

I'm not so sure about that one, our data access patterns are usually a 
lot more well thought out than our code alignment (which is really 
mostly compiler controlled). It also gives limited savings:

  9202488 vmlinux gcc-5
  9186105 vmlinux gcc-5 (-malign-data=abi)

Which is 0.1%. I've got a handful of options in that size range:

+        # Reduces vmlinux size by 0.25%:
+        KBUILD_CFLAGS += -fno-caller-saves
+
+        # Reduces vmlinux size by 1.10%:
+        KBUILD_CFLAGS += -fno-inline-small-functions
+
+        # Reduces vmlinux size by about 0.95%:
+        KBUILD_CFLAGS += -fno-tree-ch

but obviously they are more obscure and thus riskier. Find below an 
updated "Ingo's combo patch". It gives more than 10% savings here on 
x86 defconfig using gcc 4.9, without LTO.

Thanks,

	Ingo

---------------->
 arch/x86/Makefile | 24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)

diff --git a/arch/x86/Makefile b/arch/x86/Makefile
index 5ba2d9ce82dc..999e94685d12 100644
--- a/arch/x86/Makefile
+++ b/arch/x86/Makefile
@@ -77,10 +77,34 @@ else
         KBUILD_AFLAGS += -m64
         KBUILD_CFLAGS += -m64
 
+        # Pack jump targets tightly, don't align them to the default 16 bytes:
+        KBUILD_CFLAGS += -falign-jumps=1
+
+        # Pack functions tightly as well:
+        KBUILD_CFLAGS += -falign-functions=1
+
+        # Pack loops tightly as well:
+        KBUILD_CFLAGS += -falign-loops=1
+
         # Don't autogenerate traditional x87 instructions
         KBUILD_CFLAGS += $(call cc-option,-mno-80387)
         KBUILD_CFLAGS += $(call cc-option,-mno-fp-ret-in-387)
 
+        #
+        # Don't guess branch probabilities, follow the code and unlikely()/likely() hints,
+        # which reduces vmlinux size by about 5.4%:
+        #
+        KBUILD_CFLAGS += -fno-guess-branch-probability
+
+        # Reduces vmlinux size by 0.25%:
+        KBUILD_CFLAGS += -fno-caller-saves
+
+        # Reduces vmlinux size by 1.10%:
+        KBUILD_CFLAGS += -fno-inline-small-functions
+
+        # Reduces vmlinux size by about 0.95%:
+        KBUILD_CFLAGS += -fno-tree-ch
+
 	# Use -mpreferred-stack-boundary=3 if supported.
 	KBUILD_CFLAGS += $(call cc-option,-mpreferred-stack-boundary=3)