Commit 015364ba authored by Andy Polyakov's avatar Andy Polyakov
Browse files

aesv8-armx.pl: inclrease interleave factor. 

This is to compensate for higher aes* instruction latency on Cortex-A57.
parent 0f777aeb

crypto/aes/asm/aesv8-armx.pl

+206 −245
Original line number Diff line number Diff line
@@ -11,16 +11,21 @@ 
# module is endian-agnostic in sense that it supports both big- and

# little-endian cases. As does it support both 32- and 64-bit modes

# of operation. Latter is achieved by limiting amount of utilized

# registers to 16, which implies additional instructions. This has

# no effect on mighty Apple A7, as results are literally equal to

# the theoretical estimates based on instruction latencies and issue

# rate. It remains to be seen how does it affect other platforms...

# registers to 16, which implies additional NEON load and integer

# instructions. This has no effect on mighty Apple A7, where results

# are literally equal to the theoretical estimates based on AES

# instruction latencies and issue rates. On Cortex-A53, an in-order

# execution core, this costs up to 10-15%, which is partially

# compensated by implementing dedicated code path for 128-bit

# CBC encrypt case. On Cortex-A57 parallelizable mode performance

# seems to be limited by sheer amount of NEON instructions...

#

# Performance in cycles per byte processed with 128-bit key:

#

#		CBC enc		CBC dec		CTR

# Apple A7	2.39		1.20		1.20

# Cortex-A5x	n/a		n/a		n/a

# Cortex-A53	2.45		1.87		1.94

# Cortex-A57	3.64		1.34		1.32



$flavour = shift;

open STDOUT,">".shift;
@@ -435,189 +440,166 @@ $code.=<<___; 


	vst1.8	{$ivec},[$out],#16

	b	.Lcbc_done



.align	5

.Lcbc_dec128:

	vld1.32	{$tmp0-$tmp1},[$key_]

	veor	$ivec,$ivec,$rndlast

	veor	$in0,$dat0,$rndlast

	mov	$step1,$step



.Loop2x_cbc_dec128:

	aesd	$dat0,q8

	aesd	$dat1,q8

	aesimc	$dat0,$dat0

	aesimc	$dat1,$dat1

	 subs	$len,$len,#32

	aesd	$dat0,q9

	aesd	$dat1,q9

	aesimc	$dat0,$dat0

	aesimc	$dat1,$dat1

	 cclr	$step,lo

	aesd	$dat0,$tmp0

	aesd	$dat1,$tmp0

	aesimc	$dat0,$dat0

	aesimc	$dat1,$dat1

	 cclr	$step1,ls

	aesd	$dat0,$tmp1

	aesd	$dat1,$tmp1

	aesimc	$dat0,$dat0

	aesimc	$dat1,$dat1

	aesd	$dat0,q10

	aesd	$dat1,q10

	aesimc	$dat0,$dat0

	aesimc	$dat1,$dat1

	aesd	$dat0,q11

	aesd	$dat1,q11

	aesimc	$dat0,$dat0

	aesimc	$dat1,$dat1

	aesd	$dat0,q12

	aesd	$dat1,q12

	aesimc	$dat0,$dat0

	aesimc	$dat1,$dat1

	aesd	$dat0,q13

	aesd	$dat1,q13

	aesimc	$dat0,$dat0

	aesimc	$dat1,$dat1

	aesd	$dat0,q14

	aesd	$dat1,q14

	aesimc	$dat0,$dat0

	aesimc	$dat1,$dat1

	aesd	$dat0,q15

	aesd	$dat1,q15



	veor	$ivec,$ivec,$dat0

	vld1.8	{$dat0},[$inp],$step

	veor	$in0,$in0,$dat1

	vld1.8	{$dat1},[$inp],$step1

	vst1.8	{$ivec},[$out],#16

	veor	$ivec,$in1,$rndlast

	vst1.8	{$in0},[$out],#16

	veor	$in0,$dat0,$rndlast

	vorr	$in1,$dat1,$dat1

	b.hs	.Loop2x_cbc_dec128



	adds	$len,$len,#32

	veor	$ivec,$ivec,$rndlast

	b.eq	.Lcbc_done

	veor	$in0,$in0,$rndlast

	b	.Lcbc_dec_tail



___

{

my ($dat2,$in2,$tmp2)=map("q$_",(10,11,9));

$code.=<<___;

.align	5

.Lcbc_dec:

	subs	$len,$len,#16

	vorr	$in0,$dat,$dat

	vld1.8	{$dat2},[$inp],#16

	subs	$len,$len,#32		// bias

	add	$cnt,$rounds,#2

	vorr	$in1,$dat,$dat

	vorr	$dat1,$dat,$dat

	vorr	$in2,$dat2,$dat2

	b.lo	.Lcbc_dec_tail



	cclr	$step,eq

	cmp	$rounds,#2

	vld1.8	{$dat1},[$inp],$step

	vorr	$dat1,$dat2,$dat2

	vld1.8	{$dat2},[$inp],#16

	vorr	$in0,$dat,$dat

	vorr	$in1,$dat1,$dat1

	b.eq	.Lcbc_dec128

	vorr	$in2,$dat2,$dat2



.Loop2x_cbc_dec:

.Loop3x_cbc_dec:

	aesd	$dat0,q8

	aesd	$dat1,q8

	aesd	$dat2,q8

	vld1.32	{q8},[$key_],#16

	aesimc	$dat0,$dat0

	aesimc	$dat1,$dat1

	aesimc	$dat2,$dat2

	subs	$cnt,$cnt,#2

	aesd	$dat0,q9

	aesd	$dat1,q9

	aesd	$dat2,q9

	vld1.32	{q9},[$key_],#16

	aesimc	$dat0,$dat0

	aesimc	$dat1,$dat1

	b.gt	.Loop2x_cbc_dec

	aesimc	$dat2,$dat2

	b.gt	.Loop3x_cbc_dec



	aesd	$dat0,q8

	aesd	$dat1,q8

	aesd	$dat2,q8

	 veor	$tmp0,$ivec,$rndlast

	aesimc	$dat0,$dat0

	aesimc	$dat1,$dat1

	 veor	$tmp0,$ivec,$rndlast

	aesimc	$dat2,$dat2

	 veor	$tmp1,$in0,$rndlast

	aesd	$dat0,q9

	aesd	$dat1,q9

	aesd	$dat2,q9

	 veor	$tmp2,$in1,$rndlast

	 subs	$len,$len,#0x30

	aesimc	$dat0,$dat0

	aesimc	$dat1,$dat1

	 vorr	$ivec,$in1,$in1

	 subs	$len,$len,#32

	aesd	$dat0,q10

	aesd	$dat1,q10

	aesimc	$dat0,$dat0

	 cclr	$step,lo

	aesimc	$dat1,$dat1

	 mov	$key_,$key

	aesd	$dat0,q11

	aesd	$dat1,q11

	aesimc	$dat0,$dat0

	 vld1.8	{$in0},[$inp],$step

	aesimc	$dat1,$dat1

	 cclr	$step,ls

	aesimc	$dat2,$dat2

	 vorr	$ivec,$in2,$in2

	 mov.lo	x6,$len			// x6, $cnt, is zero at this point

	aesd	$dat0,q12

	aesd	$dat1,q12

	aesd	$dat2,q12

	 add	$inp,$inp,x6		// $inp is adjusted in such way that

					// at exit from the loop $dat1-$dat2

					// are loaded with last "words"

	aesimc	$dat0,$dat0

	aesimc	$dat1,$dat1

	 vld1.8	{$in1},[$inp],$step

	aesimc	$dat2,$dat2

	 mov	$key_,$key

	aesd	$dat0,q13

	aesd	$dat1,q13

	aesd	$dat2,q13

	 vld1.8	{$in0},[$inp],#16

	aesimc	$dat0,$dat0

	aesimc	$dat1,$dat1

	 vld1.32 {q8},[$key_],#16	// re-pre-load rndkey[0]

	aesimc	$dat2,$dat2

	 vld1.8	{$in1},[$inp],#16

	aesd	$dat0,q14

	aesd	$dat1,q14

	aesd	$dat2,q14

	 vld1.8	{$in2},[$inp],#16

	aesimc	$dat0,$dat0

	aesimc	$dat1,$dat1

	 vld1.32 {q9},[$key_],#16	// re-pre-load rndkey[1]

	aesimc	$dat2,$dat2

	 vld1.32 {q8},[$key_],#16	// re-pre-load rndkey[0]

	aesd	$dat0,q15

	aesd	$dat1,q15

	aesd	$dat2,q15



	 mov	$cnt,$rounds

	 add	$cnt,$rounds,#2

	veor	$tmp0,$tmp0,$dat0

	veor	$tmp1,$tmp1,$dat1

	veor	$dat2,$dat2,$tmp2

	 vld1.32 {q9},[$key_],#16	// re-pre-load rndkey[1]

	 vorr	$dat0,$in0,$in0

	vst1.8	{$tmp0},[$out],#16

	 vorr	$dat1,$in1,$in1

	vst1.8	{$tmp1},[$out],#16

	b.hs	.Loop2x_cbc_dec

	vst1.8	{$dat2},[$out],#16

	 vorr	$dat2,$in2,$in2

	b.hs	.Loop3x_cbc_dec



	adds	$len,$len,#32

	cmn	$len,#0x30

	b.eq	.Lcbc_done

	nop



.Lcbc_dec_tail:

	aesd	$dat,q8

	aesd	$dat1,q8

	aesd	$dat2,q8

	vld1.32	{q8},[$key_],#16

	aesimc	$dat,$dat

	aesimc	$dat1,$dat1

	aesimc	$dat2,$dat2

	subs	$cnt,$cnt,#2

	aesd	$dat,q9

	aesd	$dat1,q9

	aesd	$dat2,q9

	vld1.32	{q9},[$key_],#16

	aesimc	$dat,$dat

	aesimc	$dat1,$dat1

	aesimc	$dat2,$dat2

	b.gt	.Lcbc_dec_tail



	aesd	$dat,q8

	aesimc	$dat,$dat

	aesd	$dat,q9

	aesimc	$dat,$dat

	 veor	$tmp,$ivec,$rndlast

	aesd	$dat,q10

	aesimc	$dat,$dat

	 vorr	$ivec,$in0,$in0

	aesd	$dat,q11

	aesimc	$dat,$dat

	aesd	$dat,q12

	aesimc	$dat,$dat

	aesd	$dat,q13

	aesimc	$dat,$dat

	aesd	$dat,q14

	aesimc	$dat,$dat

	aesd	$dat,q15



	veor	$tmp,$tmp,$dat

	vst1.8	{$tmp},[$out],#16

	aesd	$dat1,q8

	aesd	$dat2,q8

	aesimc	$dat1,$dat1

	aesimc	$dat2,$dat2

	aesd	$dat1,q9

	aesd	$dat2,q9

	aesimc	$dat1,$dat1

	aesimc	$dat2,$dat2

	aesd	$dat1,q12

	aesd	$dat2,q12

	aesimc	$dat1,$dat1

	aesimc	$dat2,$dat2

	 cmn	$len,#0x20

	aesd	$dat1,q13

	aesd	$dat2,q13

	aesimc	$dat1,$dat1

	aesimc	$dat2,$dat2

	 veor	$tmp1,$ivec,$rndlast

	aesd	$dat1,q14

	aesd	$dat2,q14

	aesimc	$dat1,$dat1

	aesimc	$dat2,$dat2

	 veor	$tmp2,$in1,$rndlast

	aesd	$dat1,q15

	aesd	$dat2,q15

	b.eq	.Lcbc_dec_one

	veor	$tmp1,$tmp1,$dat1

	veor	$tmp2,$tmp2,$dat2

	 vorr	$ivec,$in2,$in2

	vst1.8	{$tmp1},[$out],#16

	vst1.8	{$tmp2},[$out],#16

	b	.Lcbc_done



.Lcbc_dec_one:

	veor	$tmp1,$tmp1,$dat2

	 vorr	$ivec,$in2,$in2

	vst1.8	{$tmp1},[$out],#16



.Lcbc_done:

	vst1.8	{$ivec},[$ivp]

.Lcbc_abort:

___

}

$code.=<<___	if ($flavour !~ /64/);

	vldmia	sp!,{d8-d15}

	ldmia	sp!,{r4-r8,pc}
@@ -632,8 +614,12 @@ ___ 
}}}

{{{

my ($inp,$out,$len,$key,$ivp)=map("x$_",(0..4));

my ($rounds,$cnt,$key_,$ctr,$tctr,$tctr1)=("w5","w6","x7","w8","w9","w10");

my ($rounds,$cnt,$key_)=("w5","w6","x7");

my ($ctr,$tctr0,$tctr1,$tctr2)=map("w$_",(8..10,12));

my $step="x12";		# aliases with $tctr2



my ($dat0,$dat1,$in0,$in1,$tmp0,$tmp1,$ivec,$rndlast)=map("q$_",(0..7));

my ($dat2,$in2,$tmp2)=map("q$_",(10,11,9));



my ($dat,$tmp)=($dat0,$tmp0);
@@ -662,139 +648,149 @@ $code.=<<___; 
	vld1.32		{$dat0},[$ivp]



	vld1.32		{q8-q9},[$key]		// load key schedule...

	sub		$rounds,$rounds,#6

	add		$key_,$key,x5,lsl#4	// pointer to last 7 round keys

	sub		$rounds,$rounds,#4

	mov		$step,#16

	cmp		$len,#2

	add		$key_,$key,x5,lsl#4	// pointer to last 5 round keys

	sub		$rounds,$rounds,#2

	vld1.32		{q10-q11},[$key_],#32

	vld1.32		{q12-q13},[$key_],#32

	vld1.32		{q14-q15},[$key_],#32

	vld1.32		{$rndlast},[$key_]



	add		$key_,$key,#32

	mov		$cnt,$rounds



	subs		$len,$len,#2

	b.lo		.Lctr32_tail



	cclr		$step,lo

#ifndef __ARMEB__

	rev		$ctr, $ctr

#endif

	vorr		$dat1,$dat0,$dat0

	add		$ctr, $ctr, #1

	add		$tctr1, $ctr, #1

	vorr		$dat2,$dat0,$dat0

	add		$ctr, $ctr, #2

	vorr		$ivec,$dat0,$dat0

	rev		$tctr1, $ctr

	cmp		$rounds,#2

	rev		$tctr1, $tctr1

	vmov.32		${dat1}[3],$tctr1

	b.eq		.Lctr32_128

	b.ls		.Lctr32_tail

	rev		$tctr2, $ctr

	sub		$len,$len,#3		// bias

	vmov.32		${dat2}[3],$tctr2

	b		.Loop3x_ctr32



.Loop2x_ctr32:

.align	4

.Loop3x_ctr32:

	aese		$dat0,q8

	aese		$dat1,q8

	aese		$dat2,q8

	vld1.32		{q8},[$key_],#16

	aesmc		$dat0,$dat0

	aesmc		$dat1,$dat1

	aesmc		$dat2,$dat2

	subs		$cnt,$cnt,#2

	aese		$dat0,q9

	aese		$dat1,q9

	aese		$dat2,q9

	vld1.32		{q9},[$key_],#16

	aesmc		$dat0,$dat0

	aesmc		$dat1,$dat1

	b.gt		.Loop2x_ctr32

	aesmc		$dat2,$dat2

	b.gt		.Loop3x_ctr32



	aese		$dat0,q8

	aese		$dat1,q8

	aese		$dat2,q8

	 mov		$key_,$key

	aesmc		$tmp0,$dat0

	 vorr		$dat0,$ivec,$ivec

	 vld1.8		{$in0},[$inp],#16

	aesmc		$tmp1,$dat1

	 vorr		$dat1,$ivec,$ivec

	aesmc		$dat2,$dat2

	 vorr		$dat0,$ivec,$ivec

	aese		$tmp0,q9

	aese		$tmp1,q9

	 vld1.8		{$in0},[$inp],#16

	aesmc		$tmp0,$tmp0

	 vld1.8		{$in1},[$inp],#16

	aesmc		$tmp1,$tmp1

	 add		$ctr,$ctr,#1

	aese		$tmp0,q10

	aese		$tmp1,q10

	 rev		$tctr,$ctr

	aesmc		$tmp0,$tmp0

	aesmc		$tmp1,$tmp1

	 add		$ctr,$ctr,#1

	aese		$tmp0,q11

	aese		$tmp1,q11

	 veor		$in0,$in0,$rndlast

	 rev		$tctr1,$ctr

	aese		$tmp1,q9

	aese		$dat2,q9

	 vorr		$dat1,$ivec,$ivec

	aesmc		$tmp0,$tmp0

	 vld1.8		{$in2},[$inp],#16

	aesmc		$tmp1,$tmp1

	 veor		$in1,$in1,$rndlast

	 mov		$key_,$key

	aesmc		$tmp2,$dat2

	 vorr		$dat2,$ivec,$ivec

	 add		$tctr0,$ctr,#1

	aese		$tmp0,q12

	aese		$tmp1,q12

	 subs		$len,$len,#2

	aese		$tmp2,q12

	 veor		$in0,$in0,$rndlast

	 add		$tctr1,$ctr,#2

	aesmc		$tmp0,$tmp0

	aesmc		$tmp1,$tmp1

	 vld1.32	 {q8-q9},[$key_],#32	// re-pre-load rndkey[0-1]

	aesmc		$tmp2,$tmp2

	 veor		$in1,$in1,$rndlast

	 add		$ctr,$ctr,#3

	aese		$tmp0,q13

	aese		$tmp1,q13

	aese		$tmp2,q13

	 veor		$in2,$in2,$rndlast

	 rev		$tctr0,$tctr0

	aesmc		$tmp0,$tmp0

	 vld1.32	 {q8},[$key_],#16	// re-pre-load rndkey[0]

	aesmc		$tmp1,$tmp1

	aesmc		$tmp2,$tmp2

	 vmov.32	${dat0}[3], $tctr0

	 rev		$tctr1,$tctr1

	aese		$tmp0,q14

	aese		$tmp1,q14

	 vmov.32	${dat0}[3], $tctr

	aesmc		$tmp0,$tmp0

	aese		$tmp2,q14

	 vmov.32	${dat1}[3], $tctr1

	 rev		$tctr2,$ctr

	aesmc		$tmp0,$tmp0

	aesmc		$tmp1,$tmp1

	aesmc		$tmp2,$tmp2

	 vmov.32	${dat2}[3], $tctr2

	 subs		$len,$len,#3

	aese		$tmp0,q15

	aese		$tmp1,q15

	aese		$tmp2,q15



	 mov		$cnt,$rounds

	veor		$in0,$in0,$tmp0

	veor		$in1,$in1,$tmp1

	veor		$in2,$in2,$tmp2

	 vld1.32	 {q9},[$key_],#16	// re-pre-load rndkey[1]

	vst1.8		{$in0},[$out],#16

	vst1.8		{$in1},[$out],#16

	b.hs		.Loop2x_ctr32

	vst1.8		{$in2},[$out],#16

	b.hs		.Loop3x_ctr32



	adds		$len,$len,#2

	adds		$len,$len,#3

	b.eq		.Lctr32_done

	b		.Lctr32_tail



.Lctr32_128:

	vld1.32		{$tmp0-$tmp1},[$key_]

	cmp		$len,#1

	mov		$step,#16

	cclr		$step,eq



.Loop2x_ctr32_128:

.Lctr32_tail:

	aese		$dat0,q8

	aese		$dat1,q8

	vld1.32		{q8},[$key_],#16

	aesmc		$dat0,$dat0

	 vld1.8		{$in0},[$inp],#16

	aesmc		$dat1,$dat1

	 vld1.8		{$in1},[$inp],#16

	subs		$cnt,$cnt,#2

	aese		$dat0,q9

	aese		$dat1,q9

	 add		$ctr,$ctr,#1

	aesmc		$dat0,$dat0

	aesmc		$dat1,$dat1

	 rev		$tctr,$ctr

	aese		$dat0,$tmp0

	aese		$dat1,$tmp0

	 add		$ctr,$ctr,#1

	aesmc		$dat0,$dat0

	aesmc		$dat1,$dat1

	 rev		$tctr1,$ctr

	aese		$dat0,$tmp1

	aese		$dat1,$tmp1

	 subs		$len,$len,#2

	vld1.32		{q9},[$key_],#16

	aesmc		$dat0,$dat0

	aesmc		$dat1,$dat1

	aese		$dat0,q10

	aese		$dat1,q10

	b.gt		.Lctr32_tail



	aese		$dat0,q8

	aese		$dat1,q8

	aesmc		$dat0,$dat0

	aesmc		$dat1,$dat1

	aese		$dat0,q11

	aese		$dat1,q11

	aese		$dat0,q9

	aese		$dat1,q9

	aesmc		$dat0,$dat0

	aesmc		$dat1,$dat1

	 vld1.8		{$in0},[$inp],$step

	aese		$dat0,q12

	aese		$dat1,q12

	 vld1.8		{$in1},[$inp]

	aesmc		$dat0,$dat0

	aesmc		$dat1,$dat1

	aese		$dat0,q13
@@ -803,56 +799,19 @@ $code.=<<___; 
	aesmc		$dat1,$dat1

	aese		$dat0,q14

	aese		$dat1,q14

	 veor		$in0,$in0,$rndlast

	aesmc		$dat0,$dat0

	aesmc		$dat1,$dat1

	 veor		$in0,$in0,$rndlast

	aese		$dat0,q15

	 veor		$in1,$in1,$rndlast

	aese		$dat0,q15

	aese		$dat1,q15



	cmp		$len,#1

	veor		$in0,$in0,$dat0

	vorr		$dat0,$ivec,$ivec

	veor		$in1,$in1,$dat1

	vorr		$dat1,$ivec,$ivec

	vst1.8		{$in0},[$out],#16

	vmov.32		${dat0}[3], $tctr

	vst1.8		{$in1},[$out],#16

	vmov.32		${dat1}[3], $tctr1

	b.hs		.Loop2x_ctr32_128



	adds		$len,$len,#2

	b.eq		.Lctr32_done



.Lctr32_tail:

	aese		$dat,q8

	vld1.32		{q8},[$key_],#16

	aesmc		$dat,$dat

	subs		$cnt,$cnt,#2

	aese		$dat,q9

	vld1.32		{q9},[$key_],#16

	aesmc		$dat,$dat

	b.gt		.Lctr32_tail



	aese		$dat,q8

	aesmc		$dat,$dat

	aese		$dat,q9

	aesmc		$dat,$dat

	 vld1.8		{$in0},[$inp]

	aese		$dat,q10

	aesmc		$dat,$dat

	aese		$dat,q11

	aesmc		$dat,$dat

	aese		$dat,q12

	aesmc		$dat,$dat

	aese		$dat,q13

	aesmc		$dat,$dat

	aese		$dat,q14

	aesmc		$dat,$dat

	 veor		$in0,$in0,$rndlast

	aese		$dat,q15



	veor		$in0,$in0,$dat

	vst1.8		{$in0},[$out]

	vst1.8		{$in1},[$out]



.Lctr32_done:

___
@@ -894,6 +853,7 @@ if ($flavour =~ /64/) { ######## 64-bit code 


	#s/[v]?(aes\w+)\s+([qv].*)/unaes($1,$2)/geo	or

	s/cclr\s+([wx])([^,]+),\s*([a-z]+)/csel	$1$2,$1zr,$1$2,$3/o	or

	s/mov\.([a-z]+)\s+([wx][0-9]+),\s*([wx][0-9]+)/csel	$2,$3,$2,$1/o	or

	s/vmov\.i8/movi/o	or	# fix up legacy mnemonics

	s/vext\.8/ext/o		or

	s/vrev32\.8/rev32/o	or
@@ -971,6 +931,7 @@ if ($flavour =~ /64/) { ######## 64-bit code 
	s/vdup\.32\s+(.*)/unvdup32($1)/geo		or

	s/vmov\.32\s+(.*)/unvmov32($1)/geo		or

	s/^(\s+)b\./$1b/o				or

	s/^(\s+)mov\./$1mov/o				or

	s/^(\s+)ret/$1bx\tlr/o;



	print $_,"\n";