From patchwork Fri Feb 28 16:43:41 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Taylor Simpson X-Patchwork-Id: 11412983 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 59F84138D for ; Fri, 28 Feb 2020 17:15:08 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id 21D60246A2 for ; Fri, 28 Feb 2020 17:15:08 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=fail reason="signature verification failed" (1024-bit key) header.d=quicinc.com header.i=@quicinc.com header.b="IkLwreMD" DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 21D60246A2 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=quicinc.com Authentication-Results: mail.kernel.org; spf=pass smtp.mailfrom=qemu-devel-bounces+patchwork-qemu-devel=patchwork.kernel.org@nongnu.org Received: from localhost ([::1]:51098 helo=lists1p.gnu.org) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1j7jEB-0001R4-4v for patchwork-qemu-devel@patchwork.kernel.org; Fri, 28 Feb 2020 12:15:07 -0500 Received: from eggs.gnu.org ([2001:470:142:3::10]:58321) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1j7ilz-0005Xj-6W for qemu-devel@nongnu.org; Fri, 28 Feb 2020 11:46:02 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1j7ilx-00077f-J9 for qemu-devel@nongnu.org; Fri, 28 Feb 2020 11:45:59 -0500 Received: from alexa-out-sd-02.qualcomm.com ([199.106.114.39]:27035) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1j7ilc-0005Ug-DD for qemu-devel@nongnu.org; Fri, 28 Feb 2020 11:45:57 -0500 DKIM-Signature: v=1; a=rsa-sha256; c=simple/simple; d=quicinc.com; i=@quicinc.com; q=dns/txt; s=qcdkim; t=1582908336; x=1614444336; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=62VJBm3dtMm6WcoVsEDttjWJhGMBNnVQvamLpNJf4bE=; b=IkLwreMDtsrLN1HMgqYFz4pt4shw9088WqMIISnbjBC4oCs8u6D6M+47 5IH3v0tm/kIXQhSz51ere5jzNUS6b7kpF8uGcr5tz4x5PXC+4jrom/GMu 6tINIMeKG3cmJqmYTegapBl6Wc0lJcWyChPksKfkW+RdsHEg/xjAsECE5 c=; Received: from unknown (HELO ironmsg01-sd.qualcomm.com) ([10.53.140.141]) by alexa-out-sd-02.qualcomm.com with ESMTP; 28 Feb 2020 08:44:33 -0800 Received: from vu-tsimpson-aus.qualcomm.com (HELO vu-tsimpson1-aus.qualcomm.com) ([10.222.150.1]) by ironmsg01-sd.qualcomm.com with ESMTP; 28 Feb 2020 08:44:31 -0800 Received: by vu-tsimpson1-aus.qualcomm.com (Postfix, from userid 47164) id AD7AB1110; Fri, 28 Feb 2020 10:44:31 -0600 (CST) From: Taylor Simpson To: qemu-devel@nongnu.org Subject: [RFC PATCH v2 45/67] Hexagon TCG generation - step 07 Date: Fri, 28 Feb 2020 10:43:41 -0600 Message-Id: <1582908244-304-46-git-send-email-tsimpson@quicinc.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1582908244-304-1-git-send-email-tsimpson@quicinc.com> References: <1582908244-304-1-git-send-email-tsimpson@quicinc.com> MIME-Version: 1.0 X-detected-operating-system: by eggs.gnu.org: FreeBSD 9.x [fuzzy] X-Received-From: 199.106.114.39 X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.23 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: riku.voipio@iki.fi, richard.henderson@linaro.org, laurent@vivier.eu, Taylor Simpson , philmd@redhat.com, aleksandar.m.mail@gmail.com Errors-To: qemu-devel-bounces+patchwork-qemu-devel=patchwork.kernel.org@nongnu.org Sender: "Qemu-devel" Override dczeroa, allocframe, and return instructions Signed-off-by: Taylor Simpson --- target/hexagon/helper_overrides.h | 209 ++++++++++++++++++++++++++++++++++++++ 1 file changed, 209 insertions(+) diff --git a/target/hexagon/helper_overrides.h b/target/hexagon/helper_overrides.h index 00647cb..1ac363e 100644 --- a/target/hexagon/helper_overrides.h +++ b/target/hexagon/helper_overrides.h @@ -991,4 +991,213 @@ #define fWRAP_L4_ior_memoph_io(GENHLPR, SHORTCODE) \ fWRAP_MEMOP(GENHLPR, SHORTCODE, 2, tcg_gen_ori_tl(tmp, tmp, 1 << UiV)) +/* dczeroa clears the 32 byte cache line at the address given */ +#define fWRAP_Y2_dczeroa(GENHLPR, SHORTCODE) SHORTCODE + +/* We have to brute force allocframe because it has C math in the semantics */ +#define fWRAP_S2_allocframe(GENHLPR, SHORTCODE) \ + do { \ + TCGv_i64 scramble_tmp = tcg_temp_new_i64(); \ + TCGv tmp = tcg_temp_new(); \ + { fEA_RI(RxV, -8); \ + fSTORE(1, 8, EA, fFRAME_SCRAMBLE((fCAST8_8u(fREAD_LR()) << 32) | \ + fCAST4_4u(fREAD_FP()))); \ + fWRITE_FP(EA); \ + fFRAMECHECK(EA - uiV, EA); \ + tcg_gen_subi_tl(RxV, EA, uiV); \ + } \ + tcg_temp_free_i64(scramble_tmp); \ + tcg_temp_free(tmp); \ + } while (0) + +#define fWRAP_SS2_allocframe(GENHLPR, SHORTCODE) \ + do { \ + TCGv_i64 scramble_tmp = tcg_temp_new_i64(); \ + TCGv tmp = tcg_temp_new(); \ + { fEA_RI(fREAD_SP(), -8); \ + fSTORE(1, 8, EA, fFRAME_SCRAMBLE((fCAST8_8u(fREAD_LR()) << 32) | \ + fCAST4_4u(fREAD_FP()))); \ + fWRITE_FP(EA); \ + fFRAMECHECK(EA - uiV, EA); \ + tcg_gen_subi_tl(tmp, EA, uiV); \ + fWRITE_SP(tmp); \ + } \ + tcg_temp_free_i64(scramble_tmp); \ + tcg_temp_free(tmp); \ + } while (0) + +/* Also have to brute force the deallocframe variants */ +#define fWRAP_L2_deallocframe(GENHLPR, SHORTCODE) \ + do { \ + TCGv tmp = tcg_temp_new(); \ + TCGv_i64 tmp_i64 = tcg_temp_new_i64(); \ + { \ + fEA_REG(RsV); \ + fLOAD(1, 8, u, EA, tmp_i64); \ + tcg_gen_mov_i64(RddV, fFRAME_UNSCRAMBLE(tmp_i64)); \ + tcg_gen_addi_tl(tmp, EA, 8); \ + fWRITE_SP(tmp); \ + } \ + tcg_temp_free(tmp); \ + tcg_temp_free_i64(tmp_i64); \ + } while (0) + +#define fWRAP_SL2_deallocframe(GENHLPR, SHORTCODE) \ + do { \ + TCGv WORD = tcg_temp_new(); \ + TCGv tmp = tcg_temp_new(); \ + TCGv_i64 tmp_i64 = tcg_temp_new_i64(); \ + { \ + fEA_REG(fREAD_FP()); \ + fLOAD(1, 8, u, EA, tmp_i64); \ + fFRAME_UNSCRAMBLE(tmp_i64); \ + fWRITE_LR(fGETWORD(1, tmp_i64)); \ + fWRITE_FP(fGETWORD(0, tmp_i64)); \ + tcg_gen_addi_tl(tmp, EA, 8); \ + fWRITE_SP(tmp); \ + } \ + tcg_temp_free(WORD); \ + tcg_temp_free(tmp); \ + tcg_temp_free_i64(tmp_i64); \ + } while (0) + +#define fWRAP_L4_return(GENHLPR, SHORTCODE) \ + do { \ + TCGv tmp = tcg_temp_new(); \ + TCGv_i64 tmp_i64 = tcg_temp_new_i64(); \ + TCGv WORD = tcg_temp_new(); \ + { \ + fEA_REG(RsV); \ + fLOAD(1, 8, u, EA, tmp_i64); \ + tcg_gen_mov_i64(RddV, fFRAME_UNSCRAMBLE(tmp_i64)); \ + tcg_gen_addi_tl(tmp, EA, 8); \ + fWRITE_SP(tmp); \ + fJUMPR(REG_LR, fGETWORD(1, RddV), COF_TYPE_JUMPR);\ + } \ + tcg_temp_free(tmp); \ + tcg_temp_free_i64(tmp_i64); \ + tcg_temp_free(WORD); \ + } while (0) + +#define fWRAP_SL2_return(GENHLPR, SHORTCODE) \ + do { \ + TCGv tmp = tcg_temp_new(); \ + TCGv_i64 tmp_i64 = tcg_temp_new_i64(); \ + TCGv WORD = tcg_temp_new(); \ + { \ + fEA_REG(fREAD_FP()); \ + fLOAD(1, 8, u, EA, tmp_i64); \ + fFRAME_UNSCRAMBLE(tmp_i64); \ + fWRITE_LR(fGETWORD(1, tmp_i64)); \ + fWRITE_FP(fGETWORD(0, tmp_i64)); \ + tcg_gen_addi_tl(tmp, EA, 8); \ + fWRITE_SP(tmp); \ + fJUMPR(REG_LR, fGETWORD(1, tmp_i64), COF_TYPE_JUMPR);\ + } \ + tcg_temp_free(tmp); \ + tcg_temp_free_i64(tmp_i64); \ + tcg_temp_free(WORD); \ + } while (0) + +/* + * Conditional returns follow the same predicate naming convention as + * predicated loads above + */ +#define fWRAP_COND_RETURN(PRED) \ + do { \ + TCGv LSB = tcg_temp_new(); \ + TCGv_i64 LSB_i64 = tcg_temp_new_i64(); \ + TCGv zero = tcg_const_tl(0); \ + TCGv_i64 zero_i64 = tcg_const_i64(0); \ + TCGv_i64 unscramble = tcg_temp_new_i64(); \ + TCGv WORD = tcg_temp_new(); \ + TCGv SP = tcg_temp_new(); \ + TCGv_i64 tmp_i64 = tcg_temp_new_i64(); \ + TCGv tmp = tcg_temp_new(); \ + fEA_REG(RsV); \ + PRED; \ + tcg_gen_extu_i32_i64(LSB_i64, LSB); \ + fLOAD(1, 8, u, EA, tmp_i64); \ + tcg_gen_mov_i64(unscramble, fFRAME_UNSCRAMBLE(tmp_i64)); \ + READ_REG_PAIR(RddV, HEX_REG_FP); \ + tcg_gen_movcond_i64(TCG_COND_NE, RddV, LSB_i64, zero_i64, \ + unscramble, RddV); \ + tcg_gen_mov_tl(SP, hex_gpr[HEX_REG_SP]); \ + tcg_gen_addi_tl(tmp, EA, 8); \ + tcg_gen_movcond_tl(TCG_COND_NE, SP, LSB, zero, tmp, SP); \ + fWRITE_SP(SP); \ + gen_cond_return(LSB, fGETWORD(1, RddV)); \ + tcg_temp_free(LSB); \ + tcg_temp_free_i64(LSB_i64); \ + tcg_temp_free(zero); \ + tcg_temp_free_i64(zero_i64); \ + tcg_temp_free_i64(unscramble); \ + tcg_temp_free(WORD); \ + tcg_temp_free(SP); \ + tcg_temp_free_i64(tmp_i64); \ + tcg_temp_free(tmp); \ + } while (0) + +#define fWRAP_L4_return_t(GENHLPR, SHORTCODE) \ + fWRAP_COND_RETURN(fLSBOLD(PvV)) +#define fWRAP_L4_return_f(GENHLPR, SHORTCODE) \ + fWRAP_COND_RETURN(fLSBOLDNOT(PvV)) +#define fWRAP_L4_return_tnew_pt(GENHLPR, SHORTCODE) \ + fWRAP_COND_RETURN(fLSBNEW(PvN)) +#define fWRAP_L4_return_fnew_pt(GENHLPR, SHORTCODE) \ + fWRAP_COND_RETURN(fLSBNEWNOT(PvN)) +#define fWRAP_L4_return_tnew_pnt(GENHLPR, SHORTCODE) \ + fWRAP_COND_RETURN(fLSBNEW(PvN)) +#define fWRAP_L4_return_tnew_pnt(GENHLPR, SHORTCODE) \ + fWRAP_COND_RETURN(fLSBNEW(PvN)) +#define fWRAP_L4_return_fnew_pnt(GENHLPR, SHORTCODE) \ + fWRAP_COND_RETURN(fLSBNEWNOT(PvN)) + +#define fWRAP_COND_RETURN_SUBINSN(PRED) \ + do { \ + TCGv LSB = tcg_temp_new(); \ + TCGv_i64 LSB_i64 = tcg_temp_new_i64(); \ + TCGv zero = tcg_const_tl(0); \ + TCGv_i64 zero_i64 = tcg_const_i64(0); \ + TCGv_i64 unscramble = tcg_temp_new_i64(); \ + TCGv_i64 RddV = tcg_temp_new_i64(); \ + TCGv WORD = tcg_temp_new(); \ + TCGv SP = tcg_temp_new(); \ + TCGv_i64 tmp_i64 = tcg_temp_new_i64(); \ + TCGv tmp = tcg_temp_new(); \ + fEA_REG(fREAD_FP()); \ + PRED; \ + tcg_gen_extu_i32_i64(LSB_i64, LSB); \ + fLOAD(1, 8, u, EA, tmp_i64); \ + tcg_gen_mov_i64(unscramble, fFRAME_UNSCRAMBLE(tmp_i64)); \ + READ_REG_PAIR(RddV, HEX_REG_FP); \ + tcg_gen_movcond_i64(TCG_COND_NE, RddV, LSB_i64, zero_i64, \ + unscramble, RddV); \ + tcg_gen_mov_tl(SP, hex_gpr[HEX_REG_SP]); \ + tcg_gen_addi_tl(tmp, EA, 8); \ + tcg_gen_movcond_tl(TCG_COND_NE, SP, LSB, zero, tmp, SP); \ + fWRITE_SP(SP); \ + WRITE_REG_PAIR(HEX_REG_FP, RddV); \ + gen_cond_return(LSB, fGETWORD(1, RddV)); \ + tcg_temp_free(LSB); \ + tcg_temp_free_i64(LSB_i64); \ + tcg_temp_free(zero); \ + tcg_temp_free_i64(zero_i64); \ + tcg_temp_free_i64(unscramble); \ + tcg_temp_free_i64(RddV); \ + tcg_temp_free(WORD); \ + tcg_temp_free(SP); \ + tcg_temp_free_i64(tmp_i64); \ + tcg_temp_free(tmp); \ + } while (0) + +#define fWRAP_SL2_return_t(GENHLPR, SHORTCODE) \ + fWRAP_COND_RETURN_SUBINSN(fLSBOLD(fREAD_P0())) +#define fWRAP_SL2_return_f(GENHLPR, SHORTCODE) \ + fWRAP_COND_RETURN_SUBINSN(fLSBOLDNOT(fREAD_P0())) +#define fWRAP_SL2_return_tnew(GENHLPR, SHORTCODE) \ + fWRAP_COND_RETURN_SUBINSN(fLSBNEW0) +#define fWRAP_SL2_return_fnew(GENHLPR, SHORTCODE) \ + fWRAP_COND_RETURN_SUBINSN(fLSBNEW0NOT) + #endif