2 * Flush routine for SHA1 multibuffer
4 * This file is provided under a dual BSD/GPLv2 license. When using or
5 * redistributing this file, you may do so under either license.
9 * Copyright(c) 2014 Intel Corporation.
11 * This program is free software; you can redistribute it and/or modify
12 * it under the terms of version 2 of the GNU General Public License as
13 * published by the Free Software Foundation.
15 * This program is distributed in the hope that it will be useful, but
16 * WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 * General Public License for more details.
20 * Contact Information:
21 * James Guilford <james.guilford@intel.com>
22 * Tim Chen <tim.c.chen@linux.intel.com>
26 * Copyright(c) 2014 Intel Corporation.
28 * Redistribution and use in source and binary forms, with or without
29 * modification, are permitted provided that the following conditions
32 * * Redistributions of source code must retain the above copyright
33 * notice, this list of conditions and the following disclaimer.
34 * * Redistributions in binary form must reproduce the above copyright
35 * notice, this list of conditions and the following disclaimer in
36 * the documentation and/or other materials provided with the
38 * * Neither the name of Intel Corporation nor the names of its
39 * contributors may be used to endorse or promote products derived
40 * from this software without specific prior written permission.
42 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
43 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
44 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
45 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
46 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
47 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
48 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
49 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
50 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
51 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
52 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
54 #include <linux/linkage.h>
55 #include <asm/frame.h>
56 #include "sha1_mb_mgr_datastruct.S"
61 # LINUX register definitions
70 # idx must be a register not clobbered by sha1_x8_avx2
72 #define DWORD_idx %r8d
74 #define unused_lanes %rbx
75 #define lane_data %rbx
81 #define size_offset %rax
83 #define start_offset %rax
87 #define extra_blocks %arg2
99 .macro SET_OFFSET _offset
104 # JOB* sha1_mb_mgr_flush_avx2(MB_MGR *state)
105 # arg 1 : rcx : state
106 ENTRY(sha1_mb_mgr_flush_avx2)
110 # If bit (32+3) is set, then all lanes are empty
111 mov _unused_lanes(state), unused_lanes
112 bt $32+3, unused_lanes
115 # find a lane with a non-null job
117 offset = (_ldata + 1 * _LANE_DATA_size + _job_in_lane)
118 cmpq $0, offset(state)
119 cmovne one(%rip), idx
120 offset = (_ldata + 2 * _LANE_DATA_size + _job_in_lane)
121 cmpq $0, offset(state)
122 cmovne two(%rip), idx
123 offset = (_ldata + 3 * _LANE_DATA_size + _job_in_lane)
124 cmpq $0, offset(state)
125 cmovne three(%rip), idx
126 offset = (_ldata + 4 * _LANE_DATA_size + _job_in_lane)
127 cmpq $0, offset(state)
128 cmovne four(%rip), idx
129 offset = (_ldata + 5 * _LANE_DATA_size + _job_in_lane)
130 cmpq $0, offset(state)
131 cmovne five(%rip), idx
132 offset = (_ldata + 6 * _LANE_DATA_size + _job_in_lane)
133 cmpq $0, offset(state)
134 cmovne six(%rip), idx
135 offset = (_ldata + 7 * _LANE_DATA_size + _job_in_lane)
136 cmpq $0, offset(state)
137 cmovne seven(%rip), idx
139 # copy idx to empty lanes
141 offset = (_args + _data_ptr)
142 mov offset(state,idx,8), tmp
146 offset = (_ldata + I * _LANE_DATA_size + _job_in_lane)
147 cmpq $0, offset(state)
150 offset = (_args + _data_ptr + 8*I)
151 mov tmp, offset(state)
152 offset = (_lens + 4*I)
153 movl $0xFFFFFFFF, offset(state)
160 vmovdqu _lens+0*16(state), %xmm0
161 vmovdqu _lens+1*16(state), %xmm1
163 vpminud %xmm1, %xmm0, %xmm2 # xmm2 has {D,C,B,A}
164 vpalignr $8, %xmm2, %xmm3, %xmm3 # xmm3 has {x,x,D,C}
165 vpminud %xmm3, %xmm2, %xmm2 # xmm2 has {x,x,E,F}
166 vpalignr $4, %xmm2, %xmm3, %xmm3 # xmm3 has {x,x,x,E}
167 vpminud %xmm3, %xmm2, %xmm2 # xmm2 has min value in low dword
169 vmovd %xmm2, DWORD_idx
175 vpand clear_low_nibble(%rip), %xmm2, %xmm2
176 vpshufd $0, %xmm2, %xmm2
178 vpsubd %xmm2, %xmm0, %xmm0
179 vpsubd %xmm2, %xmm1, %xmm1
181 vmovdqu %xmm0, _lens+0*16(state)
182 vmovdqu %xmm1, _lens+1*16(state)
184 # "state" and "args" are the same address, arg1
187 # state and idx are intact
191 # process completed job "idx"
192 imul $_LANE_DATA_size, idx, lane_data
193 lea _ldata(state, lane_data), lane_data
195 mov _job_in_lane(lane_data), job_rax
196 movq $0, _job_in_lane(lane_data)
197 movl $STS_COMPLETED, _status(job_rax)
198 mov _unused_lanes(state), unused_lanes
201 mov unused_lanes, _unused_lanes(state)
203 movl $0xFFFFFFFF, _lens(state, idx, 4)
205 vmovd _args_digest(state , idx, 4) , %xmm0
206 vpinsrd $1, _args_digest+1*32(state, idx, 4), %xmm0, %xmm0
207 vpinsrd $2, _args_digest+2*32(state, idx, 4), %xmm0, %xmm0
208 vpinsrd $3, _args_digest+3*32(state, idx, 4), %xmm0, %xmm0
209 movl _args_digest+4*32(state, idx, 4), tmp2_w
211 vmovdqu %xmm0, _result_digest(job_rax)
212 offset = (_result_digest + 1*16)
213 mov tmp2_w, offset(job_rax)
223 ENDPROC(sha1_mb_mgr_flush_avx2)
226 #################################################################
229 ENTRY(sha1_mb_mgr_get_comp_job_avx2)
232 ## if bit 32+3 is set, then all lanes are empty
233 mov _unused_lanes(state), unused_lanes
234 bt $(32+3), unused_lanes
238 vmovdqu _lens(state), %xmm0
239 vmovdqu _lens+1*16(state), %xmm1
241 vpminud %xmm1, %xmm0, %xmm2 # xmm2 has {D,C,B,A}
242 vpalignr $8, %xmm2, %xmm3, %xmm3 # xmm3 has {x,x,D,C}
243 vpminud %xmm3, %xmm2, %xmm2 # xmm2 has {x,x,E,F}
244 vpalignr $4, %xmm2, %xmm3, %xmm3 # xmm3 has {x,x,x,E}
245 vpminud %xmm3, %xmm2, %xmm2 # xmm2 has min value in low dword
247 vmovd %xmm2, DWORD_idx
251 # process completed job "idx"
252 imul $_LANE_DATA_size, idx, lane_data
253 lea _ldata(state, lane_data), lane_data
255 mov _job_in_lane(lane_data), job_rax
256 movq $0, _job_in_lane(lane_data)
257 movl $STS_COMPLETED, _status(job_rax)
258 mov _unused_lanes(state), unused_lanes
261 mov unused_lanes, _unused_lanes(state)
263 movl $0xFFFFFFFF, _lens(state, idx, 4)
265 vmovd _args_digest(state, idx, 4), %xmm0
266 vpinsrd $1, _args_digest+1*32(state, idx, 4), %xmm0, %xmm0
267 vpinsrd $2, _args_digest+2*32(state, idx, 4), %xmm0, %xmm0
268 vpinsrd $3, _args_digest+3*32(state, idx, 4), %xmm0, %xmm0
269 movl _args_digest+4*32(state, idx, 4), tmp2_w
271 vmovdqu %xmm0, _result_digest(job_rax)
272 movl tmp2_w, _result_digest+1*16(job_rax)
282 ENDPROC(sha1_mb_mgr_get_comp_job_avx2)
284 .section .rodata.cst16.clear_low_nibble, "aM", @progbits, 16
287 .octa 0x000000000000000000000000FFFFFFF0
289 .section .rodata.cst8, "aM", @progbits, 8