/** * Author......: See docs/credits.txt * License.....: MIT */ #define NEW_SIMD_CODE #ifdef KERNEL_STATIC #include "inc_vendor.h" #include "inc_types.h" #include "inc_common.cl" #include "inc_simd.cl" #include "inc_hash_sha1.cl" #endif #define COMPARE_S "inc_comp_single.cl" #define COMPARE_M "inc_comp_multi.cl" typedef struct pbkdf1_sha1_tmp { // pbkdf1-sha1 is limited to 160 bits u32 ipad[5]; u32 opad[5]; u32 out[5]; } pbkdf1_sha1_tmp_t; DECLSPEC static void hmac_sha1_run_V (u32x *w0, u32x *w1, u32x *w2, u32x *w3, u32x *ipad, u32x *opad, u32x *digest) { digest[0] = ipad[0]; digest[1] = ipad[1]; digest[2] = ipad[2]; digest[3] = ipad[3]; digest[4] = ipad[4]; sha1_transform_vector (w0, w1, w2, w3, digest); w0[0] = digest[0]; w0[1] = digest[1]; w0[2] = digest[2]; w0[3] = digest[3]; w1[0] = digest[4]; w1[1] = 0x80000000; w1[2] = 0; w1[3] = 0; w2[0] = 0; w2[1] = 0; w2[2] = 0; w2[3] = 0; w3[0] = 0; w3[1] = 0; w3[2] = 0; w3[3] = (64 + 20) * 8; digest[0] = opad[0]; digest[1] = opad[1]; digest[2] = opad[2]; digest[3] = opad[3]; digest[4] = opad[4]; sha1_transform_vector (w0, w1, w2, w3, digest); } KERNEL_FQ void m15100_init (KERN_ATTR_TMPS (pbkdf1_sha1_tmp_t)) { /** * base */ const u64 gid = get_global_id (0); if (gid >= gid_max) return; sha1_hmac_ctx_t sha1_hmac_ctx; sha1_hmac_init_global_swap (&sha1_hmac_ctx, pws[gid].i, pws[gid].pw_len); tmps[gid].ipad[0] = sha1_hmac_ctx.ipad.h[0]; tmps[gid].ipad[1] = sha1_hmac_ctx.ipad.h[1]; tmps[gid].ipad[2] = sha1_hmac_ctx.ipad.h[2]; tmps[gid].ipad[3] = sha1_hmac_ctx.ipad.h[3]; tmps[gid].ipad[4] = sha1_hmac_ctx.ipad.h[4]; tmps[gid].opad[0] = sha1_hmac_ctx.opad.h[0]; tmps[gid].opad[1] = sha1_hmac_ctx.opad.h[1]; tmps[gid].opad[2] = sha1_hmac_ctx.opad.h[2]; tmps[gid].opad[3] = sha1_hmac_ctx.opad.h[3]; tmps[gid].opad[4] = sha1_hmac_ctx.opad.h[4]; sha1_hmac_update_global_swap (&sha1_hmac_ctx, salt_bufs[salt_pos].salt_buf_pc, salt_bufs[salt_pos].salt_len_pc); sha1_hmac_final (&sha1_hmac_ctx); tmps[gid].out[0] = sha1_hmac_ctx.opad.h[0]; tmps[gid].out[1] = sha1_hmac_ctx.opad.h[1]; tmps[gid].out[2] = sha1_hmac_ctx.opad.h[2]; tmps[gid].out[3] = sha1_hmac_ctx.opad.h[3]; tmps[gid].out[4] = sha1_hmac_ctx.opad.h[4]; } KERNEL_FQ void m15100_loop (KERN_ATTR_TMPS (pbkdf1_sha1_tmp_t)) { const u64 gid = get_global_id (0); if ((gid * VECT_SIZE) >= gid_max) return; u32x ipad[5]; u32x opad[5]; ipad[0] = packv (tmps, ipad, gid, 0); ipad[1] = packv (tmps, ipad, gid, 1); ipad[2] = packv (tmps, ipad, gid, 2); ipad[3] = packv (tmps, ipad, gid, 3); ipad[4] = packv (tmps, ipad, gid, 4); opad[0] = packv (tmps, opad, gid, 0); opad[1] = packv (tmps, opad, gid, 1); opad[2] = packv (tmps, opad, gid, 2); opad[3] = packv (tmps, opad, gid, 3); opad[4] = packv (tmps, opad, gid, 4); u32x out[5]; out[0] = packv (tmps, out, gid, 0); out[1] = packv (tmps, out, gid, 1); out[2] = packv (tmps, out, gid, 2); out[3] = packv (tmps, out, gid, 3); out[4] = packv (tmps, out, gid, 4); for (u32 j = 0; j < loop_cnt; j++) { u32x w0[4]; u32x w1[4]; u32x w2[4]; u32x w3[4]; w0[0] = out[0]; w0[1] = out[1]; w0[2] = out[2]; w0[3] = out[3]; w1[0] = out[4]; w1[1] = 0x80000000; w1[2] = 0; w1[3] = 0; w2[0] = 0; w2[1] = 0; w2[2] = 0; w2[3] = 0; w3[0] = 0; w3[1] = 0; w3[2] = 0; w3[3] = (64 + 20) * 8; hmac_sha1_run_V (w0, w1, w2, w3, ipad, opad, out); } unpackv (tmps, out, gid, 0, out[0]); unpackv (tmps, out, gid, 1, out[1]); unpackv (tmps, out, gid, 2, out[2]); unpackv (tmps, out, gid, 3, out[3]); unpackv (tmps, out, gid, 4, out[4]); } KERNEL_FQ void m15100_comp (KERN_ATTR_TMPS (pbkdf1_sha1_tmp_t)) { /** * base */ const u64 gid = get_global_id (0); if (gid >= gid_max) return; const u64 lid = get_local_id (0); const u32 r0 = tmps[gid].out[DGST_R0]; const u32 r1 = tmps[gid].out[DGST_R1]; const u32 r2 = tmps[gid].out[DGST_R2]; const u32 r3 = tmps[gid].out[DGST_R3]; #define il_pos 0 #ifdef KERNEL_STATIC #include COMPARE_M #endif }