summaryrefslogtreecommitdiff
path: root/arch/x86/crypto/aesni-intel_avx-x86_64.S
diff options
context:
space:
mode:
authorDenys Vlasenko <dvlasenk@redhat.com>2017-01-20 00:33:04 +0300
committerHerbert Xu <herbert@gondor.apana.org.au>2017-01-23 17:50:29 +0300
commite183914af00e15eb41ae666d44e323bfa154be13 (patch)
tree177ded54a5bd45ec1ae6c544d28b0a45a31118f1 /arch/x86/crypto/aesni-intel_avx-x86_64.S
parent587d531b8f67ebe62f8326849a7a685a03cbc904 (diff)
downloadlinux-e183914af00e15eb41ae666d44e323bfa154be13.tar.xz
crypto: x86 - make constants readonly, allow linker to merge them
A lot of asm-optimized routines in arch/x86/crypto/ keep its constants in .data. This is wrong, they should be on .rodata. Mnay of these constants are the same in different modules. For example, 128-bit shuffle mask 0x000102030405060708090A0B0C0D0E0F exists in at least half a dozen places. There is a way to let linker merge them and use just one copy. The rules are as follows: mergeable objects of different sizes should not share sections. You can't put them all in one .rodata section, they will lose "mergeability". GCC puts its mergeable constants in ".rodata.cstSIZE" sections, or ".rodata.cstSIZE.<object_name>" if -fdata-sections is used. This patch does the same: .section .rodata.cst16.SHUF_MASK, "aM", @progbits, 16 It is important that all data in such section consists of 16-byte elements, not larger ones, and there are no implicit use of one element from another. When this is not the case, use non-mergeable section: .section .rodata[.VAR_NAME], "a", @progbits This reduces .data by ~15 kbytes: text data bss dec hex filename 11097415 2705840 2630712 16433967 fac32f vmlinux-prev.o 11112095 2690672 2630712 16433479 fac147 vmlinux.o Merged objects are visible in System.map: ffffffff81a28810 r POLY ffffffff81a28810 r POLY ffffffff81a28820 r TWOONE ffffffff81a28820 r TWOONE ffffffff81a28830 r PSHUFFLE_BYTE_FLIP_MASK <- merged regardless of ffffffff81a28830 r SHUF_MASK <------------- the name difference ffffffff81a28830 r SHUF_MASK ffffffff81a28830 r SHUF_MASK .. ffffffff81a28d00 r K512 <- merged three identical 640-byte tables ffffffff81a28d00 r K512 ffffffff81a28d00 r K512 Use of object names in section name suffixes is not strictly necessary, but might help if someday link stage will use garbage collection to eliminate unused sections (ld --gc-sections). Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com> CC: Herbert Xu <herbert@gondor.apana.org.au> CC: Josh Poimboeuf <jpoimboe@redhat.com> CC: Xiaodong Liu <xiaodong.liu@intel.com> CC: Megha Dey <megha.dey@intel.com> CC: linux-crypto@vger.kernel.org CC: x86@kernel.org CC: linux-kernel@vger.kernel.org Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Diffstat (limited to 'arch/x86/crypto/aesni-intel_avx-x86_64.S')
-rw-r--r--arch/x86/crypto/aesni-intel_avx-x86_64.S32
1 files changed, 24 insertions, 8 deletions
diff --git a/arch/x86/crypto/aesni-intel_avx-x86_64.S b/arch/x86/crypto/aesni-intel_avx-x86_64.S
index 522ab68d1c88..d664382c6e56 100644
--- a/arch/x86/crypto/aesni-intel_avx-x86_64.S
+++ b/arch/x86/crypto/aesni-intel_avx-x86_64.S
@@ -122,23 +122,39 @@
#include <linux/linkage.h>
#include <asm/inst.h>
-.data
+# constants in mergeable sections, linker can reorder and merge
+.section .rodata.cst16.POLY, "aM", @progbits, 16
.align 16
-
POLY: .octa 0xC2000000000000000000000000000001
+
+.section .rodata.cst16.POLY2, "aM", @progbits, 16
+.align 16
POLY2: .octa 0xC20000000000000000000001C2000000
-TWOONE: .octa 0x00000001000000000000000000000001
-# order of these constants should not change.
-# more specifically, ALL_F should follow SHIFT_MASK, and ZERO should follow ALL_F
+.section .rodata.cst16.TWOONE, "aM", @progbits, 16
+.align 16
+TWOONE: .octa 0x00000001000000000000000000000001
+.section .rodata.cst16.SHUF_MASK, "aM", @progbits, 16
+.align 16
SHUF_MASK: .octa 0x000102030405060708090A0B0C0D0E0F
-SHIFT_MASK: .octa 0x0f0e0d0c0b0a09080706050403020100
-ALL_F: .octa 0xffffffffffffffffffffffffffffffff
-ZERO: .octa 0x00000000000000000000000000000000
+
+.section .rodata.cst16.ONE, "aM", @progbits, 16
+.align 16
ONE: .octa 0x00000000000000000000000000000001
+
+.section .rodata.cst16.ONEf, "aM", @progbits, 16
+.align 16
ONEf: .octa 0x01000000000000000000000000000000
+# order of these constants should not change.
+# more specifically, ALL_F should follow SHIFT_MASK, and zero should follow ALL_F
+.section .rodata, "a", @progbits
+.align 16
+SHIFT_MASK: .octa 0x0f0e0d0c0b0a09080706050403020100
+ALL_F: .octa 0xffffffffffffffffffffffffffffffff
+ .octa 0x00000000000000000000000000000000
+
.text