From patchwork Tue Mar 7 05:34:36 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Taylor Simpson X-Patchwork-Id: 13162815 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from lists.gnu.org (lists.gnu.org [209.51.188.17]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id 0768BC678D5 for ; Tue, 7 Mar 2023 05:37:44 +0000 (UTC) Received: from localhost ([::1] helo=lists1p.gnu.org) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1pZPyh-0006pg-9j; Tue, 07 Mar 2023 00:35:11 -0500 Received: from eggs.gnu.org ([2001:470:142:3::10]) by lists.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.90_1) (envelope-from ) id 1pZPyO-0006Xo-NN for qemu-devel@nongnu.org; Tue, 07 Mar 2023 00:34:53 -0500 Received: from mx0a-0031df01.pphosted.com ([205.220.168.131]) by eggs.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.90_1) (envelope-from ) id 1pZPyJ-0006ZK-W3 for qemu-devel@nongnu.org; Tue, 07 Mar 2023 00:34:52 -0500 Received: from pps.filterd (m0279862.ppops.net [127.0.0.1]) by mx0a-0031df01.pphosted.com (8.17.1.19/8.17.1.19) with ESMTP id 3274brpv007366; Tue, 7 Mar 2023 05:34:45 GMT DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=quicinc.com; h=from : to : cc : subject : date : message-id : in-reply-to : references : mime-version : content-type : content-transfer-encoding; s=qcppdkim1; bh=n6WTtQL9UabiUZaZXv3lbVkniWbBMIisnZyYXy1A02U=; b=oao81p0BSDM4CZw8MmEACs7Dx/BmZAMuDN4GL9JiUNYOSaePXQ+CS6loESIcU0vEqbiw /wXHDKOCeStuMGwR5C+/fJTlbVn0tQ78b9WCStOemPD7BIFBrOjE0uuW4lmsOhgL4MDX b1RDutYZna5EOB3487p4AI5Jpsu850QihPM719ZxcHmAXFqzfaXAo1u3VQzUEP3/JVDD +11L5nqGfXqOOzovIdXQReERgK//kAswdBWUyWEGlhgAe9V4qnt26GiVH+TTOTzpng6V pYXv4GfALiZOE9OTg/NKbQT7m05AlSuIk+aiG8x9QZ/7k1oD9a1Pykg+8AsY5BHUrp/n /A== Received: from nalasppmta03.qualcomm.com (Global_NAT1.qualcomm.com [129.46.96.20]) by mx0a-0031df01.pphosted.com (PPS) with ESMTPS id 3p417epx8u-1 (version=TLSv1.2 cipher=ECDHE-RSA-AES256-GCM-SHA384 bits=256 verify=NOT); Tue, 07 Mar 2023 05:34:45 +0000 Received: from pps.filterd (NALASPPMTA03.qualcomm.com [127.0.0.1]) by NALASPPMTA03.qualcomm.com (8.17.1.5/8.17.1.5) with ESMTP id 3275YikA020278; Tue, 7 Mar 2023 05:34:44 GMT Received: from pps.reinject (localhost [127.0.0.1]) by NALASPPMTA03.qualcomm.com (PPS) with ESMTPS id 3p4fjsvd1t-1 (version=TLSv1.2 cipher=ECDHE-RSA-AES256-GCM-SHA384 bits=256 verify=NOT); Tue, 07 Mar 2023 05:34:44 +0000 Received: from NALASPPMTA03.qualcomm.com (NALASPPMTA03.qualcomm.com [127.0.0.1]) by pps.reinject (8.17.1.5/8.17.1.5) with ESMTP id 3275YixK020263; Tue, 7 Mar 2023 05:34:44 GMT Received: from hu-devc-sd-u20-a-1.qualcomm.com (hu-tsimpson-lv.qualcomm.com [10.47.204.221]) by NALASPPMTA03.qualcomm.com (PPS) with ESMTPS id 3275YhID020245 (version=TLSv1.2 cipher=ECDHE-RSA-AES256-GCM-SHA384 bits=256 verify=NOT); Tue, 07 Mar 2023 05:34:44 +0000 Received: by hu-devc-sd-u20-a-1.qualcomm.com (Postfix, from userid 47164) id A9DFFA85; Mon, 6 Mar 2023 21:34:42 -0800 (PST) From: Taylor Simpson To: qemu-devel@nongnu.org Cc: tsimpson@quicinc.com, richard.henderson@linaro.org, philmd@linaro.org, peter.maydell@linaro.org, bcain@quicinc.com, quic_mathbern@quicinc.com, stefanha@redhat.com, ale@rev.ng, anjo@rev.ng Subject: [PULL 15/17] Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair] Date: Mon, 6 Mar 2023 21:34:36 -0800 Message-Id: <20230307053438.2416629-16-tsimpson@quicinc.com> X-Mailer: git-send-email 2.25.1 In-Reply-To: <20230307053438.2416629-1-tsimpson@quicinc.com> References: <20230307053438.2416629-1-tsimpson@quicinc.com> MIME-Version: 1.0 X-QCInternal: smtphost X-QCInternal: smtphost X-Proofpoint-Virus-Version: vendor=nai engine=6200 definitions=5800 signatures=585085 X-Proofpoint-Virus-Version: vendor=nai engine=6200 definitions=5800 signatures=585085 X-Proofpoint-ORIG-GUID: bJzm7l2_CEIITCYYl_bUTtS3Ol_iwY8N X-Proofpoint-GUID: bJzm7l2_CEIITCYYl_bUTtS3Ol_iwY8N X-Proofpoint-Virus-Version: vendor=baseguard engine=ICAP:2.0.219,Aquarius:18.0.942,Hydra:6.0.573,FMLib:17.11.170.22 definitions=2023-03-06_14,2023-03-06_01,2023-02-09_01 X-Proofpoint-Spam-Details: rule=outbound_notspam policy=outbound score=0 spamscore=0 mlxscore=0 impostorscore=0 clxscore=1015 suspectscore=0 lowpriorityscore=0 phishscore=0 adultscore=0 malwarescore=0 priorityscore=1501 mlxlogscore=662 bulkscore=0 classifier=spam adjust=0 reason=mlx scancount=1 engine=8.12.0-2212070000 definitions=main-2303070050 Received-SPF: pass client-ip=205.220.168.131; envelope-from=tsimpson@qualcomm.com; helo=mx0a-0031df01.pphosted.com X-Spam_score_int: -24 X-Spam_score: -2.5 X-Spam_bar: -- X-Spam_report: (-2.5 / 5.0 requ) BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, HEADER_FROM_DIFFERENT_DOMAINS=0.25, RCVD_IN_DNSWL_LOW=-0.7, SPF_HELO_NONE=0.001, SPF_PASS=-0.001 autolearn=ham autolearn_force=no X-Spam_action: no action X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: qemu-devel-bounces+qemu-devel=archiver.kernel.org@nongnu.org Sender: qemu-devel-bounces+qemu-devel=archiver.kernel.org@nongnu.org We assign the instruction destination register to hex_new_value[num] instead of a TCG temp that gets copied back to hex_new_value[num]. We introduce new functions get_result_gpr[_pair] to facilitate getting the proper destination register. Since we preload hex_new_value for predicated instructions, we don't need the check for slot_cancelled. So, we call gen_log_reg_write instead. We update the helper function generation and gen_tcg.h to maintain the disable-hexagon-idef-parser configuration. Here is a simple example of the differences in the TCG code generated: IN: 0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) } BEFORE ---- 00400094 mov_i32 slot_cancelled,$0x0 mov_i32 new_r2,r2 mov_i32 loc2,$0x0 and_i32 tmp0,p0,$0x1 brcond_i32 tmp0,$0x0,eq,$L1 and_i32 tmp0,r0,r1 mov_i32 loc2,tmp0 br $L2 set_label $L1 or_i32 slot_cancelled,slot_cancelled,$0x8 set_label $L2 and_i32 tmp0,slot_cancelled,$0x8 movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq mov_i32 r2,new_r2 AFTER ---- 00400094 mov_i32 slot_cancelled,$0x0 mov_i32 new_r2,r2 and_i32 tmp0,p0,$0x1 brcond_i32 tmp0,$0x0,eq,$L1 and_i32 tmp0,r0,r1 mov_i32 new_r2,tmp0 br $L2 set_label $L1 or_i32 slot_cancelled,slot_cancelled,$0x8 set_label $L2 mov_i32 r2,new_r2 We'll remove the unnecessary manipulation of slot_cancelled in a subsequent patch. Signed-off-by: Taylor Simpson Reviewed-by: Anton Johansson Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com> --- target/hexagon/gen_tcg.h | 6 +- target/hexagon/macros.h | 15 --- target/hexagon/genptr.c | 121 +++++--------------- target/hexagon/idef-parser/parser-helpers.c | 4 +- target/hexagon/gen_helper_funcs.py | 19 ++- target/hexagon/gen_helper_protos.py | 12 +- target/hexagon/gen_tcg_funcs.py | 76 +++++++----- target/hexagon/hex_common.py | 9 +- 8 files changed, 110 insertions(+), 152 deletions(-) diff --git a/target/hexagon/gen_tcg.h b/target/hexagon/gen_tcg.h index e72672e198..bcf0cf466a 100644 --- a/target/hexagon/gen_tcg.h +++ b/target/hexagon/gen_tcg.h @@ -332,8 +332,6 @@ tcg_gen_movi_tl(EA, 0); \ PRED; \ CHECK_NOSHUF_PRED(GET_EA, SIZE, LSB); \ - PRED_LOAD_CANCEL(LSB, EA); \ - tcg_gen_movi_tl(RdV, 0); \ tcg_gen_brcondi_tl(TCG_COND_EQ, LSB, 0, label); \ fLOAD(1, SIZE, SIGN, EA, RdV); \ gen_set_label(label); \ @@ -391,8 +389,6 @@ tcg_gen_movi_tl(EA, 0); \ PRED; \ CHECK_NOSHUF_PRED(GET_EA, 8, LSB); \ - PRED_LOAD_CANCEL(LSB, EA); \ - tcg_gen_movi_i64(RddV, 0); \ tcg_gen_brcondi_tl(TCG_COND_EQ, LSB, 0, label); \ fLOAD(1, 8, u, EA, RddV); \ gen_set_label(label); \ @@ -504,7 +500,7 @@ */ #define fGEN_TCG_SL2_return(SHORTCODE) \ do { \ - TCGv_i64 RddV = tcg_temp_new_i64(); \ + TCGv_i64 RddV = get_result_gpr_pair(ctx, HEX_REG_FP); \ gen_return(ctx, RddV, hex_gpr[HEX_REG_FP]); \ gen_log_reg_write_pair(HEX_REG_FP, RddV); \ } while (0) diff --git a/target/hexagon/macros.h b/target/hexagon/macros.h index 6974ef6c9e..1fa6057d27 100644 --- a/target/hexagon/macros.h +++ b/target/hexagon/macros.h @@ -210,21 +210,6 @@ static inline void gen_cancel(uint32_t slot) #define LOAD_CANCEL(EA) do { CANCEL; } while (0) -#ifdef QEMU_GENERATE -static inline void gen_pred_cancel(TCGv pred, uint32_t slot_num) - { - TCGv slot_mask = tcg_temp_new(); - TCGv tmp = tcg_temp_new(); - TCGv zero = tcg_constant_tl(0); - tcg_gen_ori_tl(slot_mask, hex_slot_cancelled, 1 << slot_num); - tcg_gen_andi_tl(tmp, pred, 1); - tcg_gen_movcond_tl(TCG_COND_EQ, hex_slot_cancelled, tmp, zero, - slot_mask, hex_slot_cancelled); -} -#define PRED_LOAD_CANCEL(PRED, EA) \ - gen_pred_cancel(PRED, insn->is_endloop ? 4 : insn->slot) -#endif - #define STORE_CANCEL(EA) { env->slot_cancelled |= (1 << slot); } #define fMAX(A, B) (((A) > (B)) ? (A) : (B)) diff --git a/target/hexagon/genptr.c b/target/hexagon/genptr.c index 61ab9af9e4..da5dbc83cf 100644 --- a/target/hexagon/genptr.c +++ b/target/hexagon/genptr.c @@ -68,26 +68,17 @@ static inline void gen_masked_reg_write(TCGv new_val, TCGv cur_val, } } -static inline void gen_log_predicated_reg_write(int rnum, TCGv val, - uint32_t slot) +static TCGv get_result_gpr(DisasContext *ctx, int rnum) { - TCGv zero = tcg_constant_tl(0); - TCGv slot_mask = tcg_temp_new(); + return hex_new_value[rnum]; +} - tcg_gen_andi_tl(slot_mask, hex_slot_cancelled, 1 << slot); - tcg_gen_movcond_tl(TCG_COND_EQ, hex_new_value[rnum], slot_mask, zero, - val, hex_new_value[rnum]); - if (HEX_DEBUG) { - /* - * Do this so HELPER(debug_commit_end) will know - * - * Note that slot_mask indicates the value is not written - * (i.e., slot was cancelled), so we create a true/false value before - * or'ing with hex_reg_written[rnum]. - */ - tcg_gen_setcond_tl(TCG_COND_EQ, slot_mask, slot_mask, zero); - tcg_gen_or_tl(hex_reg_written[rnum], hex_reg_written[rnum], slot_mask); - } +static TCGv_i64 get_result_gpr_pair(DisasContext *ctx, int rnum) +{ + TCGv_i64 result = tcg_temp_new_i64(); + tcg_gen_concat_i32_i64(result, hex_new_value[rnum], + hex_new_value[rnum + 1]); + return result; } void gen_log_reg_write(int rnum, TCGv val) @@ -102,39 +93,6 @@ void gen_log_reg_write(int rnum, TCGv val) } } -static void gen_log_predicated_reg_write_pair(int rnum, TCGv_i64 val, - uint32_t slot) -{ - TCGv val32 = tcg_temp_new(); - TCGv zero = tcg_constant_tl(0); - TCGv slot_mask = tcg_temp_new(); - - tcg_gen_andi_tl(slot_mask, hex_slot_cancelled, 1 << slot); - /* Low word */ - tcg_gen_extrl_i64_i32(val32, val); - tcg_gen_movcond_tl(TCG_COND_EQ, hex_new_value[rnum], - slot_mask, zero, - val32, hex_new_value[rnum]); - /* High word */ - tcg_gen_extrh_i64_i32(val32, val); - tcg_gen_movcond_tl(TCG_COND_EQ, hex_new_value[rnum + 1], - slot_mask, zero, - val32, hex_new_value[rnum + 1]); - if (HEX_DEBUG) { - /* - * Do this so HELPER(debug_commit_end) will know - * - * Note that slot_mask indicates the value is not written - * (i.e., slot was cancelled), so we create a true/false value before - * or'ing with hex_reg_written[rnum]. - */ - tcg_gen_setcond_tl(TCG_COND_EQ, slot_mask, slot_mask, zero); - tcg_gen_or_tl(hex_reg_written[rnum], hex_reg_written[rnum], slot_mask); - tcg_gen_or_tl(hex_reg_written[rnum + 1], hex_reg_written[rnum + 1], - slot_mask); - } -} - static void gen_log_reg_write_pair(int rnum, TCGv_i64 val) { const target_ulong reg_mask_low = reg_immut_masks[rnum]; @@ -290,11 +248,12 @@ static inline void gen_write_ctrl_reg_pair(DisasContext *ctx, int reg_num, TCGv_i64 val) { if (reg_num == HEX_REG_P3_0_ALIASED) { + TCGv result = get_result_gpr(ctx, reg_num + 1); TCGv val32 = tcg_temp_new(); tcg_gen_extrl_i64_i32(val32, val); gen_write_p3_0(ctx, val32); tcg_gen_extrh_i64_i32(val32, val); - gen_log_reg_write(reg_num + 1, val32); + tcg_gen_mov_tl(result, val32); } else { gen_log_reg_write_pair(reg_num, val); if (reg_num == HEX_REG_QEMU_PKT_CNT) { @@ -673,31 +632,28 @@ static void gen_jumpr(DisasContext *ctx, TCGv new_pc) static void gen_call(DisasContext *ctx, int pc_off) { - TCGv next_PC = - tcg_constant_tl(ctx->pkt->pc + ctx->pkt->encod_pkt_size_in_bytes); - gen_log_reg_write(HEX_REG_LR, next_PC); + TCGv lr = get_result_gpr(ctx, HEX_REG_LR); + tcg_gen_movi_tl(lr, ctx->next_PC); gen_write_new_pc_pcrel(ctx, pc_off, TCG_COND_ALWAYS, NULL); } static void gen_callr(DisasContext *ctx, TCGv new_pc) { - TCGv next_PC = tcg_constant_tl(ctx->next_PC); - gen_log_reg_write(HEX_REG_LR, next_PC); + TCGv lr = get_result_gpr(ctx, HEX_REG_LR); + tcg_gen_movi_tl(lr, ctx->next_PC); gen_write_new_pc_addr(ctx, new_pc, TCG_COND_ALWAYS, NULL); } static void gen_cond_call(DisasContext *ctx, TCGv pred, TCGCond cond, int pc_off) { - TCGv next_PC; + TCGv lr = get_result_gpr(ctx, HEX_REG_LR); TCGv lsb = tcg_temp_new(); TCGLabel *skip = gen_new_label(); tcg_gen_andi_tl(lsb, pred, 1); gen_write_new_pc_pcrel(ctx, pc_off, cond, lsb); tcg_gen_brcondi_tl(cond, lsb, 0, skip); - next_PC = - tcg_constant_tl(ctx->pkt->pc + ctx->pkt->encod_pkt_size_in_bytes); - gen_log_reg_write(HEX_REG_LR, next_PC); + tcg_gen_movi_tl(lr, ctx->next_PC); gen_set_label(skip); } @@ -728,8 +684,7 @@ static void gen_load_frame(DisasContext *ctx, TCGv_i64 frame, TCGv EA) tcg_gen_qemu_ld64(frame, EA, ctx->mem_idx); } -static void gen_return_base(DisasContext *ctx, TCGv_i64 dst, TCGv src, - TCGv r29) +static void gen_return(DisasContext *ctx, TCGv_i64 dst, TCGv src) { /* * frame = *src @@ -739,6 +694,7 @@ static void gen_return_base(DisasContext *ctx, TCGv_i64 dst, TCGv src, */ TCGv_i64 frame = tcg_temp_new_i64(); TCGv r31 = tcg_temp_new(); + TCGv r29 = get_result_gpr(ctx, HEX_REG_SP); gen_load_frame(ctx, frame, src); gen_frame_unscramble(frame); @@ -748,45 +704,25 @@ static void gen_return_base(DisasContext *ctx, TCGv_i64 dst, TCGv src, gen_jumpr(ctx, r31); } -static void gen_return(DisasContext *ctx, TCGv_i64 dst, TCGv src) -{ - TCGv r29 = tcg_temp_new(); - gen_return_base(ctx, dst, src, r29); - gen_log_reg_write(HEX_REG_SP, r29); -} - /* if (pred) dst = dealloc_return(src):raw */ static void gen_cond_return(DisasContext *ctx, TCGv_i64 dst, TCGv src, TCGv pred, TCGCond cond) { TCGv LSB = tcg_temp_new(); - TCGv mask = tcg_temp_new(); - TCGv r29 = tcg_temp_new(); TCGLabel *skip = gen_new_label(); tcg_gen_andi_tl(LSB, pred, 1); - /* Initialize the results in case the predicate is false */ - tcg_gen_movi_i64(dst, 0); - tcg_gen_movi_tl(r29, 0); - - /* Set the bit in hex_slot_cancelled if the predicate is flase */ - tcg_gen_movi_tl(mask, 1 << ctx->insn->slot); - tcg_gen_or_tl(mask, hex_slot_cancelled, mask); - tcg_gen_movcond_tl(cond, hex_slot_cancelled, LSB, tcg_constant_tl(0), - mask, hex_slot_cancelled); - tcg_gen_brcondi_tl(cond, LSB, 0, skip); - gen_return_base(ctx, dst, src, r29); + gen_return(ctx, dst, src); gen_set_label(skip); - gen_log_predicated_reg_write(HEX_REG_SP, r29, ctx->insn->slot); } /* sub-instruction version (no RddV, so handle it manually) */ static void gen_cond_return_subinsn(DisasContext *ctx, TCGCond cond, TCGv pred) { - TCGv_i64 RddV = tcg_temp_new_i64(); + TCGv_i64 RddV = get_result_gpr_pair(ctx, HEX_REG_FP); gen_cond_return(ctx, RddV, hex_gpr[HEX_REG_FP], pred, cond); - gen_log_predicated_reg_write_pair(HEX_REG_FP, RddV, ctx->insn->slot); + gen_log_reg_write_pair(HEX_REG_FP, RddV); } static void gen_endloop0(DisasContext *ctx) @@ -836,9 +772,9 @@ static void gen_endloop0(DisasContext *ctx) TCGLabel *label3 = gen_new_label(); tcg_gen_brcondi_tl(TCG_COND_LEU, hex_gpr[HEX_REG_LC0], 1, label3); { + TCGv lc0 = get_result_gpr(ctx, HEX_REG_LC0); gen_jumpr(ctx, hex_gpr[HEX_REG_SA0]); - tcg_gen_subi_tl(hex_new_value[HEX_REG_LC0], - hex_gpr[HEX_REG_LC0], 1); + tcg_gen_subi_tl(lc0, hex_gpr[HEX_REG_LC0], 1); } gen_set_label(label3); } @@ -855,8 +791,9 @@ static void gen_endloop1(DisasContext *ctx) TCGLabel *label = gen_new_label(); tcg_gen_brcondi_tl(TCG_COND_LEU, hex_gpr[HEX_REG_LC1], 1, label); { + TCGv lc1 = get_result_gpr(ctx, HEX_REG_LC1); gen_jumpr(ctx, hex_gpr[HEX_REG_SA1]); - tcg_gen_subi_tl(hex_new_value[HEX_REG_LC1], hex_gpr[HEX_REG_LC1], 1); + tcg_gen_subi_tl(lc1, hex_gpr[HEX_REG_LC1], 1); } gen_set_label(label); } @@ -909,15 +846,17 @@ static void gen_endloop01(DisasContext *ctx) */ tcg_gen_brcondi_tl(TCG_COND_LEU, hex_gpr[HEX_REG_LC0], 1, label3); { + TCGv lc0 = get_result_gpr(ctx, HEX_REG_LC0); gen_jumpr(ctx, hex_gpr[HEX_REG_SA0]); - tcg_gen_subi_tl(hex_new_value[HEX_REG_LC0], hex_gpr[HEX_REG_LC0], 1); + tcg_gen_subi_tl(lc0, hex_gpr[HEX_REG_LC0], 1); tcg_gen_br(done); } gen_set_label(label3); tcg_gen_brcondi_tl(TCG_COND_LEU, hex_gpr[HEX_REG_LC1], 1, done); { + TCGv lc1 = get_result_gpr(ctx, HEX_REG_LC1); gen_jumpr(ctx, hex_gpr[HEX_REG_SA1]); - tcg_gen_subi_tl(hex_new_value[HEX_REG_LC1], hex_gpr[HEX_REG_LC1], 1); + tcg_gen_subi_tl(lc1, hex_gpr[HEX_REG_LC1], 1); } gen_set_label(done); } diff --git a/target/hexagon/idef-parser/parser-helpers.c b/target/hexagon/idef-parser/parser-helpers.c index 4e0c643cae..9720ef4ad1 100644 --- a/target/hexagon/idef-parser/parser-helpers.c +++ b/target/hexagon/idef-parser/parser-helpers.c @@ -1668,9 +1668,7 @@ void gen_inst_init_args(Context *c, YYLTYPE *locp) for (unsigned i = 0; i < c->inst.init_list->len; i++) { HexValue *val = &g_array_index(c->inst.init_list, HexValue, i); if (val->type == REGISTER_ARG) { - char reg_id[5]; - reg_compose(c, locp, &val->reg, reg_id); - EMIT_HEAD(c, "tcg_gen_movi_i%u(%s, 0);\n", val->bit_width, reg_id); + /* Nothing to do here */ } else if (val->type == PREDICATE) { char suffix = val->is_dotnew ? 'N' : 'V'; EMIT_HEAD(c, "tcg_gen_movi_i%u(P%c%c, 0);\n", val->bit_width, diff --git a/target/hexagon/gen_helper_funcs.py b/target/hexagon/gen_helper_funcs.py index 19e9883f4c..7a224b66e6 100755 --- a/target/hexagon/gen_helper_funcs.py +++ b/target/hexagon/gen_helper_funcs.py @@ -1,7 +1,7 @@ #!/usr/bin/env python3 ## -## Copyright(c) 2019-2022 Qualcomm Innovation Center, Inc. All Rights Reserved. +## Copyright(c) 2019-2023 Qualcomm Innovation Center, Inc. All Rights Reserved. ## ## This program is free software; you can redistribute it and/or modify ## it under the terms of the GNU General Public License as published by @@ -226,6 +226,14 @@ def gen_helper_function(f, tag, tagregs, tagimms): print("Bad register parse: ",regtype,regid,toss,numregs) i += 1 + ## For conditional instructions, we pass in the destination register + if 'A_CONDEXEC' in hex_common.attribdict[tag]: + for regtype, regid, toss, numregs in regs: + if (hex_common.is_writeonly(regid) and + not hex_common.is_hvx_reg(regtype)): + gen_helper_arg_opn(f, regtype, regid, i, tag) + i += 1 + ## Arguments to the helper function are the source regs and immediates for regtype,regid,toss,numregs in regs: if (hex_common.is_read(regid)): @@ -262,10 +270,11 @@ def gen_helper_function(f, tag, tagregs, tagimms): if hex_common.need_ea(tag): gen_decl_ea(f) ## Declare the return variable i=0 - for regtype,regid,toss,numregs in regs: - if (hex_common.is_writeonly(regid)): - gen_helper_dest_decl_opn(f,regtype,regid,i) - i += 1 + if 'A_CONDEXEC' not in hex_common.attribdict[tag]: + for regtype,regid,toss,numregs in regs: + if (hex_common.is_writeonly(regid)): + gen_helper_dest_decl_opn(f,regtype,regid,i) + i += 1 for regtype,regid,toss,numregs in regs: if (hex_common.is_read(regid)): diff --git a/target/hexagon/gen_helper_protos.py b/target/hexagon/gen_helper_protos.py index 674bf370fa..ddddc9e4f0 100755 --- a/target/hexagon/gen_helper_protos.py +++ b/target/hexagon/gen_helper_protos.py @@ -1,7 +1,7 @@ #!/usr/bin/env python3 ## -## Copyright(c) 2019-2022 Qualcomm Innovation Center, Inc. All Rights Reserved. +## Copyright(c) 2019-2023 Qualcomm Innovation Center, Inc. All Rights Reserved. ## ## This program is free software; you can redistribute it and/or modify ## it under the terms of the GNU General Public License as published by @@ -87,6 +87,7 @@ def gen_helper_prototype(f, tag, tagregs, tagimms): if hex_common.need_slot(tag): def_helper_size += 1 if hex_common.need_PC(tag): def_helper_size += 1 if hex_common.helper_needs_next_PC(tag): def_helper_size += 1 + if hex_common.need_condexec_reg(tag, regs): def_helper_size += 1 f.write('DEF_HELPER_%s(%s' % (def_helper_size, tag)) ## The return type is void f.write(', void' ) @@ -96,6 +97,7 @@ def gen_helper_prototype(f, tag, tagregs, tagimms): if hex_common.need_part1(tag): def_helper_size += 1 if hex_common.need_slot(tag): def_helper_size += 1 if hex_common.need_PC(tag): def_helper_size += 1 + if hex_common.need_condexec_reg(tag, regs): def_helper_size += 1 if hex_common.helper_needs_next_PC(tag): def_helper_size += 1 f.write('DEF_HELPER_%s(%s' % (def_helper_size, tag)) @@ -121,6 +123,14 @@ def gen_helper_prototype(f, tag, tagregs, tagimms): gen_def_helper_opn(f, tag, regtype, regid, toss, numregs, i) i += 1 + ## For conditional instructions, we pass in the destination register + if 'A_CONDEXEC' in hex_common.attribdict[tag]: + for regtype, regid, toss, numregs in regs: + if (hex_common.is_writeonly(regid) and + not hex_common.is_hvx_reg(regtype)): + gen_def_helper_opn(f, tag, regtype, regid, toss, numregs, i) + i += 1 + ## Generate the qemu type for each input operand (regs and immediates) for regtype,regid,toss,numregs in regs: if (hex_common.is_read(regid)): diff --git a/target/hexagon/gen_tcg_funcs.py b/target/hexagon/gen_tcg_funcs.py index 22bcb56917..9336528e50 100755 --- a/target/hexagon/gen_tcg_funcs.py +++ b/target/hexagon/gen_tcg_funcs.py @@ -30,23 +30,33 @@ def gen_decl_ea_tcg(f, tag): def genptr_decl_pair_writable(f, tag, regtype, regid, regno): regN="%s%sN" % (regtype,regid) - f.write(" TCGv_i64 %s%sV = tcg_temp_new_i64();\n" % \ - (regtype, regid)) - if (regtype == "C"): + if (regtype == "R"): + f.write(" const int %s = insn->regno[%d];\n" % (regN, regno)) + elif (regtype == "C"): f.write(" const int %s = insn->regno[%d] + HEX_REG_SA0;\n" % \ (regN, regno)) else: - f.write(" const int %s = insn->regno[%d];\n" % (regN, regno)) + print("Bad register parse: ", regtype, regid) + f.write(" TCGv_i64 %s%sV = get_result_gpr_pair(ctx, %s);\n" % \ + (regtype, regid, regN)) def genptr_decl_writable(f, tag, regtype, regid, regno): regN="%s%sN" % (regtype,regid) - f.write(" TCGv %s%sV = tcg_temp_new();\n" % \ - (regtype, regid)) - if (regtype == "C"): + if (regtype == "R"): + f.write(" const int %s = insn->regno[%d];\n" % (regN, regno)) + f.write(" TCGv %s%sV = get_result_gpr(ctx, %s);\n" % \ + (regtype, regid, regN)) + elif (regtype == "C"): f.write(" const int %s = insn->regno[%d] + HEX_REG_SA0;\n" % \ (regN, regno)) - else: + f.write(" TCGv %s%sV = get_result_gpr(ctx, %s);\n" % \ + (regtype, regid, regN)) + elif (regtype == "P"): f.write(" const int %s = insn->regno[%d];\n" % (regN, regno)) + f.write(" TCGv %s%sV = tcg_temp_new();\n" % \ + (regtype, regid)) + else: + print("Bad register parse: ", regtype, regid) def genptr_decl(f, tag, regtype, regid, regno): regN="%s%sN" % (regtype,regid) @@ -249,8 +259,12 @@ def genptr_src_read(f, tag, regtype, regid): f.write(" hex_gpr[%s%sN + 1]);\n" % \ (regtype, regid)) elif (regid in {"x", "y"}): - f.write(" tcg_gen_mov_tl(%s%sV, hex_gpr[%s%sN]);\n" % \ - (regtype,regid,regtype,regid)) + ## For read/write registers, we need to get the original value into + ## the result TCGv. For conditional instructions, this is done in + ## gen_start_packet. For unconditional instructions, we do it here. + if ('A_CONDEXEC' not in hex_common.attribdict[tag]): + f.write(" tcg_gen_mov_tl(%s%sV, hex_gpr[%s%sN]);\n" % \ + (regtype, regid, regtype, regid)) elif (regid not in {"s", "t", "u", "v"}): print("Bad register parse: ", regtype, regid) elif (regtype == "P"): @@ -360,25 +374,16 @@ def gen_helper_call_imm(f,immlett): f.write(", tcgv_%s" % hex_common.imm_name(immlett)) def genptr_dst_write_pair(f, tag, regtype, regid): - if ('A_CONDEXEC' in hex_common.attribdict[tag]): - f.write(" gen_log_predicated_reg_write_pair(%s%sN, %s%sV, insn->slot);\n" % \ - (regtype, regid, regtype, regid)) - else: - f.write(" gen_log_reg_write_pair(%s%sN, %s%sV);\n" % \ - (regtype, regid, regtype, regid)) + f.write(" gen_log_reg_write_pair(%s%sN, %s%sV);\n" % \ + (regtype, regid, regtype, regid)) def genptr_dst_write(f, tag, regtype, regid): if (regtype == "R"): if (regid in {"dd", "xx", "yy"}): genptr_dst_write_pair(f, tag, regtype, regid) elif (regid in {"d", "e", "x", "y"}): - if ('A_CONDEXEC' in hex_common.attribdict[tag]): - f.write(" gen_log_predicated_reg_write(%s%sN, %s%sV,\n" % \ - (regtype, regid, regtype, regid)) - f.write(" insn->slot);\n") - else: - f.write(" gen_log_reg_write(%s%sN, %s%sV);\n" % \ - (regtype, regid, regtype, regid)) + f.write(" gen_log_reg_write(%s%sN, %s%sV);\n" % \ + (regtype, regid, regtype, regid)) else: print("Bad register parse: ", regtype, regid) elif (regtype == "P"): @@ -462,14 +467,15 @@ def genptr_dst_write_opn(f,regtype, regid, tag): ## For A2_add: Rd32=add(Rs32,Rt32), { RdV=RsV+RtV;} ## We produce: ## static void generate_A2_add(DisasContext *ctx) -## { -## TCGv RdV = tcg_temp_new(); -## const int RdN = insn->regno[0]; -## TCGv RsV = hex_gpr[insn->regno[1]]; -## TCGv RtV = hex_gpr[insn->regno[2]]; -## -## gen_log_reg_write(RdN, RdV); -## } +## { +## Insn *insn __attribute__((unused)) = ctx->insn; +## const int RdN = insn->regno[0]; +## TCGv RdV = get_result_gpr(ctx, RdN); +## TCGv RsV = hex_gpr[insn->regno[1]]; +## TCGv RtV = hex_gpr[insn->regno[2]]; +## +## gen_log_reg_write(RdN, RdV); +## } ## ## where depends on hex_common.skip_qemu_helper(tag) ## if hex_common.skip_qemu_helper(tag) is True @@ -553,6 +559,14 @@ def gen_tcg_func(f, tag, regs, imms): if (i > 0): f.write(", ") f.write("cpu_env") i=1 + ## For conditional instructions, we pass in the destination register + if 'A_CONDEXEC' in hex_common.attribdict[tag]: + for regtype, regid, toss, numregs in regs: + if (hex_common.is_writeonly(regid) and + not hex_common.is_hvx_reg(regtype)): + gen_helper_call_opn(f, tag, regtype, regid, toss, \ + numregs, i) + i += 1 for regtype,regid,toss,numregs in regs: if (hex_common.is_written(regid)): if (not hex_common.is_hvx_reg(regtype)): diff --git a/target/hexagon/hex_common.py b/target/hexagon/hex_common.py index 76da362c11..0200a66cb6 100755 --- a/target/hexagon/hex_common.py +++ b/target/hexagon/hex_common.py @@ -1,7 +1,7 @@ #!/usr/bin/env python3 ## -## Copyright(c) 2019-2022 Qualcomm Innovation Center, Inc. All Rights Reserved. +## Copyright(c) 2019-2023 Qualcomm Innovation Center, Inc. All Rights Reserved. ## ## This program is free software; you can redistribute it and/or modify ## it under the terms of the GNU General Public License as published by @@ -237,6 +237,13 @@ def helper_needs_next_PC(tag): def need_pkt_has_multi_cof(tag): return 'A_COF' in attribdict[tag] +def need_condexec_reg(tag, regs): + if 'A_CONDEXEC' in attribdict[tag]: + for regtype, regid, toss, numregs in regs: + if is_writeonly(regid) and not is_hvx_reg(regtype): + return True + return False + def skip_qemu_helper(tag): return tag in overrides.keys()