From patchwork Tue Jun 4 16:55:22 2013 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Kamal Mostafa X-Patchwork-Id: 248776 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Received: from huckleberry.canonical.com (huckleberry.canonical.com [91.189.94.19]) by ozlabs.org (Postfix) with ESMTP id C41542C00A2 for ; Wed, 5 Jun 2013 03:03:27 +1000 (EST) Received: from localhost ([127.0.0.1] helo=huckleberry.canonical.com) by huckleberry.canonical.com with esmtp (Exim 4.76) (envelope-from ) id 1Ujudk-0000sF-9u; Tue, 04 Jun 2013 17:03:20 +0000 Received: from youngberry.canonical.com ([91.189.89.112]) by huckleberry.canonical.com with esmtp (Exim 4.76) (envelope-from ) id 1Ujude-0000oe-3Q for kernel-team@lists.ubuntu.com; Tue, 04 Jun 2013 17:03:14 +0000 Received: from c-67-160-231-42.hsd1.ca.comcast.net ([67.160.231.42] helo=fourier) by youngberry.canonical.com with esmtpsa (TLS1.0:DHE_RSA_AES_128_CBC_SHA1:16) (Exim 4.71) (envelope-from ) id 1UjuXs-0003Dt-9D; Tue, 04 Jun 2013 16:57:16 +0000 Received: from kamal by fourier with local (Exim 4.80) (envelope-from ) id 1UjuXq-0002su-62; Tue, 04 Jun 2013 09:57:14 -0700 From: Kamal Mostafa To: linux-kernel@vger.kernel.org, stable@vger.kernel.org, kernel-team@lists.ubuntu.com Subject: [PATCH 38/75] ARM: 7626/1: arm/crypto: Make asm SHA-1 and AES code Thumb-2 compatible Date: Tue, 4 Jun 2013 09:55:22 -0700 Message-Id: <1370364959-9475-39-git-send-email-kamal@canonical.com> X-Mailer: git-send-email 1.8.1.2 In-Reply-To: <1370364959-9475-1-git-send-email-kamal@canonical.com> References: <1370364959-9475-1-git-send-email-kamal@canonical.com> X-Extended-Stable: 3.8 Cc: Russell King , Dave Martin , Kamal Mostafa X-BeenThere: kernel-team@lists.ubuntu.com X-Mailman-Version: 2.1.14 Precedence: list List-Id: Kernel team discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Errors-To: kernel-team-bounces@lists.ubuntu.com Sender: kernel-team-bounces@lists.ubuntu.com 3.8.13.2 -stable review patch. If anyone has any objections, please let me know. ------------------ From: Dave Martin commit 638591cd7b601d403ed703d55062b48c32ea8cfb upstream. This patch fixes aes-armv4.S and sha1-armv4-large.S to work natively in Thumb. This allows ARM/Thumb interworking workarounds to be removed. I also take the opportunity to convert some explicit assembler directives for exported functions to the standard ENTRY()/ENDPROC(). For the code itself: * In sha1_block_data_order, use of TEQ with sp is deprecated in ARMv7 and not supported in Thumb. For the branches back to .L_00_15 and .L_40_59, the TEQ is converted to a CMP, under the assumption that clobbering the C flag here will not cause incorrect behaviour. For the first branch back to .L_20_39_or_60_79 the C flag is important, so sp is moved temporarily into another register so that TEQ can be used for the comparison. * In the AES code, most forms of register-indexed addressing with shifts and rotates are not permitted for loads and stores in Thumb, so the address calculation is done using a separate instruction for the Thumb case. The resulting code is unlikely to be optimally scheduled, but it should not have a large impact given the overall size of the code. I haven't run any benchmarks. Signed-off-by: Dave Martin Tested-by: David McCullough (ARM only) Acked-by: David McCullough Acked-by: Nicolas Pitre Signed-off-by: Russell King Signed-off-by: Kamal Mostafa --- arch/arm/crypto/aes-armv4.S | 64 ++++++++++++-------------------------- arch/arm/crypto/sha1-armv4-large.S | 24 ++++++-------- 2 files changed, 29 insertions(+), 59 deletions(-) diff --git a/arch/arm/crypto/aes-armv4.S b/arch/arm/crypto/aes-armv4.S index e59b1d5..19d6cd6 100644 --- a/arch/arm/crypto/aes-armv4.S +++ b/arch/arm/crypto/aes-armv4.S @@ -34,8 +34,9 @@ @ A little glue here to select the correct code below for the ARM CPU @ that is being targetted. +#include + .text -.code 32 .type AES_Te,%object .align 5 @@ -145,10 +146,8 @@ AES_Te: @ void AES_encrypt(const unsigned char *in, unsigned char *out, @ const AES_KEY *key) { -.global AES_encrypt -.type AES_encrypt,%function .align 5 -AES_encrypt: +ENTRY(AES_encrypt) sub r3,pc,#8 @ AES_encrypt stmdb sp!,{r1,r4-r12,lr} mov r12,r0 @ inp @@ -239,15 +238,8 @@ AES_encrypt: strb r6,[r12,#14] strb r3,[r12,#15] #endif -#if __ARM_ARCH__>=5 ldmia sp!,{r4-r12,pc} -#else - ldmia sp!,{r4-r12,lr} - tst lr,#1 - moveq pc,lr @ be binary compatible with V4, yet - .word 0xe12fff1e @ interoperable with Thumb ISA:-) -#endif -.size AES_encrypt,.-AES_encrypt +ENDPROC(AES_encrypt) .type _armv4_AES_encrypt,%function .align 2 @@ -386,10 +378,8 @@ _armv4_AES_encrypt: ldr pc,[sp],#4 @ pop and return .size _armv4_AES_encrypt,.-_armv4_AES_encrypt -.global private_AES_set_encrypt_key -.type private_AES_set_encrypt_key,%function .align 5 -private_AES_set_encrypt_key: +ENTRY(private_AES_set_encrypt_key) _armv4_AES_set_encrypt_key: sub r3,pc,#8 @ AES_set_encrypt_key teq r0,#0 @@ -658,15 +648,11 @@ _armv4_AES_set_encrypt_key: .Ldone: mov r0,#0 ldmia sp!,{r4-r12,lr} -.Labrt: tst lr,#1 - moveq pc,lr @ be binary compatible with V4, yet - .word 0xe12fff1e @ interoperable with Thumb ISA:-) -.size private_AES_set_encrypt_key,.-private_AES_set_encrypt_key +.Labrt: mov pc,lr +ENDPROC(private_AES_set_encrypt_key) -.global private_AES_set_decrypt_key -.type private_AES_set_decrypt_key,%function .align 5 -private_AES_set_decrypt_key: +ENTRY(private_AES_set_decrypt_key) str lr,[sp,#-4]! @ push lr #if 0 @ kernel does both of these in setkey so optimise this bit out by @@ -748,15 +734,8 @@ private_AES_set_decrypt_key: bne .Lmix mov r0,#0 -#if __ARM_ARCH__>=5 ldmia sp!,{r4-r12,pc} -#else - ldmia sp!,{r4-r12,lr} - tst lr,#1 - moveq pc,lr @ be binary compatible with V4, yet - .word 0xe12fff1e @ interoperable with Thumb ISA:-) -#endif -.size private_AES_set_decrypt_key,.-private_AES_set_decrypt_key +ENDPROC(private_AES_set_decrypt_key) .type AES_Td,%object .align 5 @@ -862,10 +841,8 @@ AES_Td: @ void AES_decrypt(const unsigned char *in, unsigned char *out, @ const AES_KEY *key) { -.global AES_decrypt -.type AES_decrypt,%function .align 5 -AES_decrypt: +ENTRY(AES_decrypt) sub r3,pc,#8 @ AES_decrypt stmdb sp!,{r1,r4-r12,lr} mov r12,r0 @ inp @@ -956,15 +933,8 @@ AES_decrypt: strb r6,[r12,#14] strb r3,[r12,#15] #endif -#if __ARM_ARCH__>=5 ldmia sp!,{r4-r12,pc} -#else - ldmia sp!,{r4-r12,lr} - tst lr,#1 - moveq pc,lr @ be binary compatible with V4, yet - .word 0xe12fff1e @ interoperable with Thumb ISA:-) -#endif -.size AES_decrypt,.-AES_decrypt +ENDPROC(AES_decrypt) .type _armv4_AES_decrypt,%function .align 2 @@ -1064,7 +1034,9 @@ _armv4_AES_decrypt: and r9,lr,r1,lsr#8 ldrb r7,[r10,r7] @ Td4[s1>>0] - ldrb r1,[r10,r1,lsr#24] @ Td4[s1>>24] + ARM( ldrb r1,[r10,r1,lsr#24] ) @ Td4[s1>>24] + THUMB( add r1,r10,r1,lsr#24 ) @ Td4[s1>>24] + THUMB( ldrb r1,[r1] ) ldrb r8,[r10,r8] @ Td4[s1>>16] eor r0,r7,r0,lsl#24 ldrb r9,[r10,r9] @ Td4[s1>>8] @@ -1077,7 +1049,9 @@ _armv4_AES_decrypt: ldrb r8,[r10,r8] @ Td4[s2>>0] and r9,lr,r2,lsr#16 - ldrb r2,[r10,r2,lsr#24] @ Td4[s2>>24] + ARM( ldrb r2,[r10,r2,lsr#24] ) @ Td4[s2>>24] + THUMB( add r2,r10,r2,lsr#24 ) @ Td4[s2>>24] + THUMB( ldrb r2,[r2] ) eor r0,r0,r7,lsl#8 ldrb r9,[r10,r9] @ Td4[s2>>16] eor r1,r8,r1,lsl#16 @@ -1090,7 +1064,9 @@ _armv4_AES_decrypt: and r9,lr,r3 @ i2 ldrb r9,[r10,r9] @ Td4[s3>>0] - ldrb r3,[r10,r3,lsr#24] @ Td4[s3>>24] + ARM( ldrb r3,[r10,r3,lsr#24] ) @ Td4[s3>>24] + THUMB( add r3,r10,r3,lsr#24 ) @ Td4[s3>>24] + THUMB( ldrb r3,[r3] ) eor r0,r0,r7,lsl#16 ldr r7,[r11,#0] eor r1,r1,r8,lsl#8 diff --git a/arch/arm/crypto/sha1-armv4-large.S b/arch/arm/crypto/sha1-armv4-large.S index 7050ab1..92c6eed 100644 --- a/arch/arm/crypto/sha1-armv4-large.S +++ b/arch/arm/crypto/sha1-armv4-large.S @@ -51,13 +51,12 @@ @ Profiler-assisted and platform-specific optimization resulted in 10% @ improvement on Cortex A8 core and 12.2 cycles per byte. -.text +#include -.global sha1_block_data_order -.type sha1_block_data_order,%function +.text .align 2 -sha1_block_data_order: +ENTRY(sha1_block_data_order) stmdb sp!,{r4-r12,lr} add r2,r1,r2,lsl#6 @ r2 to point at the end of r1 ldmia r0,{r3,r4,r5,r6,r7} @@ -194,7 +193,7 @@ sha1_block_data_order: eor r10,r10,r7,ror#2 @ F_00_19(B,C,D) str r9,[r14,#-4]! add r3,r3,r10 @ E+=F_00_19(B,C,D) - teq r14,sp + cmp r14,sp bne .L_00_15 @ [((11+4)*5+2)*3] #if __ARM_ARCH__<7 ldrb r10,[r1,#2] @@ -374,7 +373,9 @@ sha1_block_data_order: @ F_xx_xx add r3,r3,r9 @ E+=X[i] add r3,r3,r10 @ E+=F_20_39(B,C,D) - teq r14,sp @ preserve carry + ARM( teq r14,sp ) @ preserve carry + THUMB( mov r11,sp ) + THUMB( teq r14,r11 ) @ preserve carry bne .L_20_39_or_60_79 @ [+((12+3)*5+2)*4] bcs .L_done @ [+((12+3)*5+2)*4], spare 300 bytes @@ -466,7 +467,7 @@ sha1_block_data_order: add r3,r3,r9 @ E+=X[i] add r3,r3,r10 @ E+=F_40_59(B,C,D) add r3,r3,r11,ror#2 - teq r14,sp + cmp r14,sp bne .L_40_59 @ [+((12+5)*5+2)*4] ldr r8,.LK_60_79 @@ -485,19 +486,12 @@ sha1_block_data_order: teq r1,r2 bne .Lloop @ [+18], total 1307 -#if __ARM_ARCH__>=5 ldmia sp!,{r4-r12,pc} -#else - ldmia sp!,{r4-r12,lr} - tst lr,#1 - moveq pc,lr @ be binary compatible with V4, yet - .word 0xe12fff1e @ interoperable with Thumb ISA:-) -#endif .align 2 .LK_00_19: .word 0x5a827999 .LK_20_39: .word 0x6ed9eba1 .LK_40_59: .word 0x8f1bbcdc .LK_60_79: .word 0xca62c1d6 -.size sha1_block_data_order,.-sha1_block_data_order +ENDPROC(sha1_block_data_order) .asciz "SHA1 block transform for ARMv4, CRYPTOGAMS by " .align 2