18c2ecf20Sopenharmony_ci;; SPDX-License-Identifier: GPL-2.0-or-later 28c2ecf20Sopenharmony_ci;; Copyright 2010 Free Software Foundation, Inc. 38c2ecf20Sopenharmony_ci;; Contributed by Bernd Schmidt <bernds@codesourcery.com>. 48c2ecf20Sopenharmony_ci;; 58c2ecf20Sopenharmony_ci 68c2ecf20Sopenharmony_ci#include <linux/linkage.h> 78c2ecf20Sopenharmony_ci 88c2ecf20Sopenharmony_ci ;; ABI considerations for the divide functions 98c2ecf20Sopenharmony_ci ;; The following registers are call-used: 108c2ecf20Sopenharmony_ci ;; __c6xabi_divi A0,A1,A2,A4,A6,B0,B1,B2,B4,B5 118c2ecf20Sopenharmony_ci ;; __c6xabi_divu A0,A1,A2,A4,A6,B0,B1,B2,B4 128c2ecf20Sopenharmony_ci ;; __c6xabi_remi A1,A2,A4,A5,A6,B0,B1,B2,B4 138c2ecf20Sopenharmony_ci ;; __c6xabi_remu A1,A4,A5,A7,B0,B1,B2,B4 148c2ecf20Sopenharmony_ci ;; 158c2ecf20Sopenharmony_ci ;; In our implementation, divu and remu are leaf functions, 168c2ecf20Sopenharmony_ci ;; while both divi and remi call into divu. 178c2ecf20Sopenharmony_ci ;; A0 is not clobbered by any of the functions. 188c2ecf20Sopenharmony_ci ;; divu does not clobber B2 either, which is taken advantage of 198c2ecf20Sopenharmony_ci ;; in remi. 208c2ecf20Sopenharmony_ci ;; divi uses B5 to hold the original return address during 218c2ecf20Sopenharmony_ci ;; the call to divu. 228c2ecf20Sopenharmony_ci ;; remi uses B2 and A5 to hold the input values during the 238c2ecf20Sopenharmony_ci ;; call to divu. It stores B3 in on the stack. 248c2ecf20Sopenharmony_ci 258c2ecf20Sopenharmony_ci .text 268c2ecf20Sopenharmony_ciENTRY(__c6xabi_divu) 278c2ecf20Sopenharmony_ci ;; We use a series of up to 31 subc instructions. First, we find 288c2ecf20Sopenharmony_ci ;; out how many leading zero bits there are in the divisor. This 298c2ecf20Sopenharmony_ci ;; gives us both a shift count for aligning (shifting) the divisor 308c2ecf20Sopenharmony_ci ;; to the, and the number of times we have to execute subc. 318c2ecf20Sopenharmony_ci 328c2ecf20Sopenharmony_ci ;; At the end, we have both the remainder and most of the quotient 338c2ecf20Sopenharmony_ci ;; in A4. The top bit of the quotient is computed first and is 348c2ecf20Sopenharmony_ci ;; placed in A2. 358c2ecf20Sopenharmony_ci 368c2ecf20Sopenharmony_ci ;; Return immediately if the dividend is zero. 378c2ecf20Sopenharmony_ci mv .s2x A4, B1 388c2ecf20Sopenharmony_ci [B1] lmbd .l2 1, B4, B1 398c2ecf20Sopenharmony_ci|| [!B1] b .s2 B3 ; RETURN A 408c2ecf20Sopenharmony_ci|| [!B1] mvk .d2 1, B4 418c2ecf20Sopenharmony_ci mv .l1x B1, A6 428c2ecf20Sopenharmony_ci|| shl .s2 B4, B1, B4 438c2ecf20Sopenharmony_ci 448c2ecf20Sopenharmony_ci ;; The loop performs a maximum of 28 steps, so we do the 458c2ecf20Sopenharmony_ci ;; first 3 here. 468c2ecf20Sopenharmony_ci cmpltu .l1x A4, B4, A2 478c2ecf20Sopenharmony_ci [!A2] sub .l1x A4, B4, A4 488c2ecf20Sopenharmony_ci|| shru .s2 B4, 1, B4 498c2ecf20Sopenharmony_ci|| xor .s1 1, A2, A2 508c2ecf20Sopenharmony_ci 518c2ecf20Sopenharmony_ci shl .s1 A2, 31, A2 528c2ecf20Sopenharmony_ci|| [B1] subc .l1x A4,B4,A4 538c2ecf20Sopenharmony_ci|| [B1] add .s2 -1, B1, B1 548c2ecf20Sopenharmony_ci [B1] subc .l1x A4,B4,A4 558c2ecf20Sopenharmony_ci|| [B1] add .s2 -1, B1, B1 568c2ecf20Sopenharmony_ci 578c2ecf20Sopenharmony_ci ;; RETURN A may happen here (note: must happen before the next branch) 588c2ecf20Sopenharmony_ci_divu_loop: 598c2ecf20Sopenharmony_ci cmpgt .l2 B1, 7, B0 608c2ecf20Sopenharmony_ci|| [B1] subc .l1x A4,B4,A4 618c2ecf20Sopenharmony_ci|| [B1] add .s2 -1, B1, B1 628c2ecf20Sopenharmony_ci [B1] subc .l1x A4,B4,A4 638c2ecf20Sopenharmony_ci|| [B1] add .s2 -1, B1, B1 648c2ecf20Sopenharmony_ci|| [B0] b .s1 _divu_loop 658c2ecf20Sopenharmony_ci [B1] subc .l1x A4,B4,A4 668c2ecf20Sopenharmony_ci|| [B1] add .s2 -1, B1, B1 678c2ecf20Sopenharmony_ci [B1] subc .l1x A4,B4,A4 688c2ecf20Sopenharmony_ci|| [B1] add .s2 -1, B1, B1 698c2ecf20Sopenharmony_ci [B1] subc .l1x A4,B4,A4 708c2ecf20Sopenharmony_ci|| [B1] add .s2 -1, B1, B1 718c2ecf20Sopenharmony_ci [B1] subc .l1x A4,B4,A4 728c2ecf20Sopenharmony_ci|| [B1] add .s2 -1, B1, B1 738c2ecf20Sopenharmony_ci [B1] subc .l1x A4,B4,A4 748c2ecf20Sopenharmony_ci|| [B1] add .s2 -1, B1, B1 758c2ecf20Sopenharmony_ci ;; loop backwards branch happens here 768c2ecf20Sopenharmony_ci 778c2ecf20Sopenharmony_ci ret .s2 B3 788c2ecf20Sopenharmony_ci|| mvk .s1 32, A1 798c2ecf20Sopenharmony_ci sub .l1 A1, A6, A6 808c2ecf20Sopenharmony_ci shl .s1 A4, A6, A4 818c2ecf20Sopenharmony_ci shru .s1 A4, 1, A4 828c2ecf20Sopenharmony_ci|| sub .l1 A6, 1, A6 838c2ecf20Sopenharmony_ci or .l1 A2, A4, A4 848c2ecf20Sopenharmony_ci shru .s1 A4, A6, A4 858c2ecf20Sopenharmony_ci nop 868c2ecf20Sopenharmony_ciENDPROC(__c6xabi_divu) 87