From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-6.7 required=3.0 tests=HEADER_FROM_DIFFERENT_DOMAINS, INCLUDES_PATCH,MAILING_LIST_MULTI,SIGNED_OFF_BY,SPF_HELO_NONE,SPF_PASS, URIBL_BLOCKED autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id A72A4C43613 for ; Thu, 20 Jun 2019 11:30:23 +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 6707E2082C for ; Thu, 20 Jun 2019 11:30:23 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 6707E2082C Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=redhat.com Authentication-Results: mail.kernel.org; spf=pass smtp.mailfrom=qemu-devel-bounces+qemu-devel=archiver.kernel.org@nongnu.org Received: from localhost ([::1]:46462 helo=lists.gnu.org) by lists.gnu.org with esmtp (Exim 4.86_2) (envelope-from ) id 1hdvGo-0003Im-Ik for qemu-devel@archiver.kernel.org; Thu, 20 Jun 2019 07:30:22 -0400 Received: from eggs.gnu.org ([209.51.188.92]:52851) by lists.gnu.org with esmtp (Exim 4.86_2) (envelope-from ) id 1hdv4K-0004QD-39 for qemu-devel@nongnu.org; Thu, 20 Jun 2019 07:17:30 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1hduz3-0002f6-8P for qemu-devel@nongnu.org; Thu, 20 Jun 2019 07:12:03 -0400 Received: from mail-wr1-f67.google.com ([209.85.221.67]:46812) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_128_CBC_SHA1:16) (Exim 4.71) (envelope-from ) id 1hduz2-0002eY-Ep for qemu-devel@nongnu.org; Thu, 20 Jun 2019 07:12:01 -0400 Received: by mail-wr1-f67.google.com with SMTP id n4so2547524wrw.13 for ; Thu, 20 Jun 2019 04:12:00 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:subject:to:cc:references:from:openpgp:message-id :date:user-agent:mime-version:in-reply-to:content-language :content-transfer-encoding; bh=SiJ+faFumrakUcH8DKoI0rkEIiTK8zkKQBqfmX752is=; b=pvFPRa3/OQIzRG+CuXYDhvoolw3mXNfWFtPm+yct5u1zfMztO8lK69DiZMWxGIgoAk 1IF8+nGJ0qaQpUq+qkTHO39zCtE2d40i+IMuStgYi2FHUEX9zbx0rNhKredsP2ygONIg JOlX7qVA+jB0rdkGGddhfdis0Ej8N0FMEUb7FXhXcpcLFCy+EUuwriYTFHwL2EH+jBzc Ze/J2uAlAYZ+KUjUjd/i9/ChO6GKw73jv4vT9xb0UeDnJVID2e+M4tOagF7tKchu7wyL dKg7R4iqjswUL5n6j6zsrAe64yIwXZ5bFVGNovnZbGsrRsNGJ5aDdjRNsR08HS8285G6 0Dbw== X-Gm-Message-State: APjAAAXQCCv0qgK/DhfJBQ5tupjayMUeCF4PUdrqbRZ6UDedzPuJTC/s rpelgPbynAE7V198ZvhLkNk4jA== X-Google-Smtp-Source: APXvYqy5YbziheGgsjxZSUKDt+V7hrYZ3hyvTG/hhN8BJM2u5u06CmSNcH0hUaqOBkv1g30DspulTg== X-Received: by 2002:adf:dcc2:: with SMTP id x2mr51158674wrm.55.1561029119307; Thu, 20 Jun 2019 04:11:59 -0700 (PDT) Received: from [192.168.1.38] (183.red-88-21-202.staticip.rima-tde.net. [88.21.202.183]) by smtp.gmail.com with ESMTPSA id i188sm5210803wma.27.2019.06.20.04.11.58 (version=TLS1_3 cipher=AEAD-AES128-GCM-SHA256 bits=128/128); Thu, 20 Jun 2019 04:11:58 -0700 (PDT) To: Aleksandar Markovic , qemu-devel@nongnu.org References: <1561022396-20649-1-git-send-email-aleksandar.markovic@rt-rk.com> <1561022396-20649-4-git-send-email-aleksandar.markovic@rt-rk.com> From: =?UTF-8?Q?Philippe_Mathieu-Daud=c3=a9?= Openpgp: id=89C1E78F601EE86C867495CBA2A3FD6EDEADC0DE; url=http://pgp.mit.edu/pks/lookup?op=get&search=0xA2A3FD6EDEADC0DE Message-ID: <19e8abc0-005e-446e-7f64-ae6aad6ef05e@redhat.com> Date: Thu, 20 Jun 2019 13:11:58 +0200 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101 Thunderbird/60.6.1 MIME-Version: 1.0 In-Reply-To: <1561022396-20649-4-git-send-email-aleksandar.markovic@rt-rk.com> Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: 8bit X-detected-operating-system: by eggs.gnu.org: GNU/Linux 2.2.x-3.x [generic] [fuzzy] X-Received-From: 209.85.221.67 Subject: Re: [Qemu-devel] [PATCH 3/4] target/mips: Fix some space checkpatch errors in translate.c 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: arikalo@wavecomp.com, amarkovic@wavecomp.com Errors-To: qemu-devel-bounces+qemu-devel=archiver.kernel.org@nongnu.org Sender: "Qemu-devel" On 6/20/19 11:19 AM, Aleksandar Markovic wrote: > From: Aleksandar Markovic > > Remove some space-related checkpatch warning. > > Signed-off-by: Aleksandar Markovic Reviewed-by: Philippe Mathieu-Daudé Tested-by: Philippe Mathieu-Daudé > --- > target/mips/translate.c | 126 +++++++++++++++++++++++++----------------------- > 1 file changed, 65 insertions(+), 61 deletions(-) > > diff --git a/target/mips/translate.c b/target/mips/translate.c > index a3cf976..324b32a 100644 > --- a/target/mips/translate.c > +++ b/target/mips/translate.c > @@ -2617,7 +2617,7 @@ static const char * const mxuregnames[] = { > } while (0) > > /* General purpose registers moves. */ > -static inline void gen_load_gpr (TCGv t, int reg) > +static inline void gen_load_gpr(TCGv t, int reg) > { > if (reg == 0) > tcg_gen_movi_tl(t, 0); > @@ -2625,14 +2625,14 @@ static inline void gen_load_gpr (TCGv t, int reg) > tcg_gen_mov_tl(t, cpu_gpr[reg]); > } > > -static inline void gen_store_gpr (TCGv t, int reg) > +static inline void gen_store_gpr(TCGv t, int reg) > { > if (reg != 0) > tcg_gen_mov_tl(cpu_gpr[reg], t); > } > > /* Moves to/from shadow registers. */ > -static inline void gen_load_srsgpr (int from, int to) > +static inline void gen_load_srsgpr(int from, int to) > { > TCGv t0 = tcg_temp_new(); > > @@ -2839,7 +2839,7 @@ static void gen_store_fpr64(DisasContext *ctx, TCGv_i64 t, int reg) > } > } > > -static inline int get_fp_bit (int cc) > +static inline int get_fp_bit(int cc) > { > if (cc) > return 24 + cc; > @@ -2848,7 +2848,8 @@ static inline int get_fp_bit (int cc) > } > > /* Addresses computation */ > -static inline void gen_op_addr_add (DisasContext *ctx, TCGv ret, TCGv arg0, TCGv arg1) > +static inline void gen_op_addr_add(DisasContext *ctx, TCGv ret, TCGv arg0, > + TCGv arg1) > { > tcg_gen_add_tl(ret, arg0, arg1); > > @@ -3328,8 +3329,8 @@ OP_LD_ATOMIC(lld,ld64); > #endif > #undef OP_LD_ATOMIC > > -static void gen_base_offset_addr (DisasContext *ctx, TCGv addr, > - int base, int offset) > +static void gen_base_offset_addr(DisasContext *ctx, TCGv addr, > + int base, int offset) > { > if (base == 0) { > tcg_gen_movi_tl(addr, offset); > @@ -3341,7 +3342,7 @@ static void gen_base_offset_addr (DisasContext *ctx, TCGv addr, > } > } > > -static target_ulong pc_relative_pc (DisasContext *ctx) > +static target_ulong pc_relative_pc(DisasContext *ctx) > { > target_ulong pc = ctx->base.pc_next; > > @@ -3578,8 +3579,8 @@ static void gen_llwp(DisasContext *ctx, uint32_t base, int16_t offset, > } > > /* Store */ > -static void gen_st (DisasContext *ctx, uint32_t opc, int rt, > - int base, int offset) > +static void gen_st(DisasContext *ctx, uint32_t opc, int rt, > + int base, int offset) > { > TCGv t0 = tcg_temp_new(); > TCGv t1 = tcg_temp_new(); > @@ -3717,8 +3718,8 @@ static void gen_scwp(DisasContext *ctx, uint32_t base, int16_t offset, > } > > /* Load and store */ > -static void gen_flt_ldst (DisasContext *ctx, uint32_t opc, int ft, > - TCGv t0) > +static void gen_flt_ldst(DisasContext *ctx, uint32_t opc, int ft, > + TCGv t0) > { > /* Don't do NOP if destination is zero: we must perform the actual > memory access. */ > @@ -5132,8 +5133,8 @@ static void gen_mul_txx9(DisasContext *ctx, uint32_t opc, > tcg_temp_free(t1); > } > > -static void gen_mul_vr54xx (DisasContext *ctx, uint32_t opc, > - int rd, int rs, int rt) > +static void gen_mul_vr54xx(DisasContext *ctx, uint32_t opc, > + int rd, int rs, int rt) > { > TCGv t0 = tcg_temp_new(); > TCGv t1 = tcg_temp_new(); > @@ -5196,8 +5197,8 @@ static void gen_mul_vr54xx (DisasContext *ctx, uint32_t opc, > tcg_temp_free(t1); > } > > -static void gen_cl (DisasContext *ctx, uint32_t opc, > - int rd, int rs) > +static void gen_cl(DisasContext *ctx, uint32_t opc, > + int rd, int rs) > { > TCGv t0; > > @@ -6188,8 +6189,8 @@ static void gen_compute_branch_nm(DisasContext *ctx, uint32_t opc, > > > /* special3 bitfield operations */ > -static void gen_bitops (DisasContext *ctx, uint32_t opc, int rt, > - int rs, int lsb, int msb) > +static void gen_bitops(DisasContext *ctx, uint32_t opc, int rt, > + int rs, int lsb, int msb) > { > TCGv t0 = tcg_temp_new(); > TCGv t1 = tcg_temp_new(); > @@ -6259,7 +6260,7 @@ fail: > tcg_temp_free(t1); > } > > -static void gen_bshfl (DisasContext *ctx, uint32_t op2, int rt, int rd) > +static void gen_bshfl(DisasContext *ctx, uint32_t op2, int rt, int rd) > { > TCGv t0; > > @@ -6502,7 +6503,7 @@ static inline void gen_mfhc0_load64(TCGv arg, target_ulong off, int shift) > tcg_temp_free_i64(t0); > } > > -static inline void gen_mfc0_load32 (TCGv arg, target_ulong off) > +static inline void gen_mfc0_load32(TCGv arg, target_ulong off) > { > TCGv_i32 t0 = tcg_temp_new_i32(); > > @@ -6511,13 +6512,13 @@ static inline void gen_mfc0_load32 (TCGv arg, target_ulong off) > tcg_temp_free_i32(t0); > } > > -static inline void gen_mfc0_load64 (TCGv arg, target_ulong off) > +static inline void gen_mfc0_load64(TCGv arg, target_ulong off) > { > tcg_gen_ld_tl(arg, cpu_env, off); > tcg_gen_ext32s_tl(arg, arg); > } > > -static inline void gen_mtc0_store32 (TCGv arg, target_ulong off) > +static inline void gen_mtc0_store32(TCGv arg, target_ulong off) > { > TCGv_i32 t0 = tcg_temp_new_i32(); > > @@ -10077,7 +10078,8 @@ die: > generate_exception_end(ctx, EXCP_RI); > } > > -static void gen_cp0 (CPUMIPSState *env, DisasContext *ctx, uint32_t opc, int rt, int rd) > +static void gen_cp0(CPUMIPSState *env, DisasContext *ctx, uint32_t opc, > + int rt, int rd) > { > const char *opn = "ldst"; > > @@ -10633,7 +10635,8 @@ enum r6_f_cmp_op { > R6_OPC_CMP_SUNE_D = FOP(26, FMT_L), > R6_OPC_CMP_SNE_D = FOP(27, FMT_L), > }; > -static void gen_cp1 (DisasContext *ctx, uint32_t opc, int rt, int fs) > + > +static void gen_cp1(DisasContext *ctx, uint32_t opc, int rt, int fs) > { > TCGv t0 = tcg_temp_new(); > > @@ -10714,7 +10717,7 @@ static void gen_cp1 (DisasContext *ctx, uint32_t opc, int rt, int fs) > tcg_temp_free(t0); > } > > -static void gen_movci (DisasContext *ctx, int rd, int rs, int cc, int tf) > +static void gen_movci(DisasContext *ctx, int rd, int rs, int cc, int tf) > { > TCGLabel *l1; > TCGCond cond; > @@ -10763,7 +10766,8 @@ static inline void gen_movcf_s(DisasContext *ctx, int fs, int fd, int cc, > tcg_temp_free_i32(t0); > } > > -static inline void gen_movcf_d (DisasContext *ctx, int fs, int fd, int cc, int tf) > +static inline void gen_movcf_d(DisasContext *ctx, int fs, int fd, int cc, > + int tf) > { > int cond; > TCGv_i32 t0 = tcg_temp_new_i32(); > @@ -10886,8 +10890,8 @@ static void gen_sel_d(DisasContext *ctx, enum fopcode op1, int fd, int ft, > tcg_temp_free_i64(t1); > } > > -static void gen_farith (DisasContext *ctx, enum fopcode op1, > - int ft, int fs, int fd, int cc) > +static void gen_farith(DisasContext *ctx, enum fopcode op1, > + int ft, int fs, int fd, int cc) > { > uint32_t func = ctx->opcode & 0x3f; > switch (op1) { > @@ -12314,8 +12318,8 @@ static void gen_farith (DisasContext *ctx, enum fopcode op1, > } > > /* Coprocessor 3 (FPU) */ > -static void gen_flt3_ldst (DisasContext *ctx, uint32_t opc, > - int fd, int fs, int base, int index) > +static void gen_flt3_ldst(DisasContext *ctx, uint32_t opc, > + int fd, int fs, int base, int index) > { > TCGv t0 = tcg_temp_new(); > > @@ -12394,8 +12398,8 @@ static void gen_flt3_ldst (DisasContext *ctx, uint32_t opc, > tcg_temp_free(t0); > } > > -static void gen_flt3_arith (DisasContext *ctx, uint32_t opc, > - int fd, int fr, int fs, int ft) > +static void gen_flt3_arith(DisasContext *ctx, uint32_t opc, > + int fd, int fr, int fs, int ft) > { > switch (opc) { > case OPC_ALNV_PS: > @@ -13157,17 +13161,17 @@ enum { > RR_RY_CNVT_SEW = 0x6, > }; > > -static int xlat (int r) > +static int xlat(int r) > { > static int map[] = { 16, 17, 2, 3, 4, 5, 6, 7 }; > > return map[r]; > } > > -static void gen_mips16_save (DisasContext *ctx, > - int xsregs, int aregs, > - int do_ra, int do_s0, int do_s1, > - int framesize) > +static void gen_mips16_save(DisasContext *ctx, > + int xsregs, int aregs, > + int do_ra, int do_s0, int do_s1, > + int framesize) > { > TCGv t0 = tcg_temp_new(); > TCGv t1 = tcg_temp_new(); > @@ -13322,10 +13326,10 @@ static void gen_mips16_save (DisasContext *ctx, > tcg_temp_free(t2); > } > > -static void gen_mips16_restore (DisasContext *ctx, > - int xsregs, int aregs, > - int do_ra, int do_s0, int do_s1, > - int framesize) > +static void gen_mips16_restore(DisasContext *ctx, > + int xsregs, int aregs, > + int do_ra, int do_s0, int do_s1, > + int framesize) > { > int astatic; > TCGv t0 = tcg_temp_new(); > @@ -13428,8 +13432,8 @@ static void gen_mips16_restore (DisasContext *ctx, > tcg_temp_free(t2); > } > > -static void gen_addiupc (DisasContext *ctx, int rx, int imm, > - int is_64_bit, int extended) > +static void gen_addiupc(DisasContext *ctx, int rx, int imm, > + int is_64_bit, int extended) > { > TCGv t0; > > @@ -13459,9 +13463,9 @@ static void gen_cache_operation(DisasContext *ctx, uint32_t op, int base, > } > > #if defined(TARGET_MIPS64) > -static void decode_i64_mips16 (DisasContext *ctx, > - int ry, int funct, int16_t offset, > - int extended) > +static void decode_i64_mips16(DisasContext *ctx, > + int ry, int funct, int16_t offset, > + int extended) > { > switch (funct) { > case I64_LDSP: > @@ -13520,7 +13524,7 @@ static void decode_i64_mips16 (DisasContext *ctx, > } > #endif > > -static int decode_extended_mips16_opc (CPUMIPSState *env, DisasContext *ctx) > +static int decode_extended_mips16_opc(CPUMIPSState *env, DisasContext *ctx) > { > int extend = cpu_lduw_code(env, ctx->base.pc_next + 2); > int op, rx, ry, funct, sa; > @@ -13734,7 +13738,7 @@ static inline void gen_helper_do_semihosting(void *env) > } > #endif > > -static int decode_mips16_opc (CPUMIPSState *env, DisasContext *ctx) > +static int decode_mips16_opc(CPUMIPSState *env, DisasContext *ctx) > { > int rx, ry; > int sa; > @@ -13957,7 +13961,7 @@ static int decode_mips16_opc (CPUMIPSState *env, DisasContext *ctx) > case M16_OPC_LWPC: > gen_ld(ctx, OPC_LWPC, rx, 0, ((uint8_t)ctx->opcode) << 2); > break; > -#if defined (TARGET_MIPS64) > +#if defined(TARGET_MIPS64) > case M16_OPC_LWU: > check_insn(ctx, ISA_MIPS3); > check_mips_64(ctx); > @@ -14061,7 +14065,7 @@ static int decode_mips16_opc (CPUMIPSState *env, DisasContext *ctx) > case RR_SRAV: > gen_shift(ctx, OPC_SRAV, ry, rx, ry); > break; > -#if defined (TARGET_MIPS64) > +#if defined(TARGET_MIPS64) > case RR_DSRL: > check_insn(ctx, ISA_MIPS3); > check_mips_64(ctx); > @@ -14124,7 +14128,7 @@ static int decode_mips16_opc (CPUMIPSState *env, DisasContext *ctx) > case RR_MFLO: > gen_HILO(ctx, OPC_MFLO, 0, rx); > break; > -#if defined (TARGET_MIPS64) > +#if defined(TARGET_MIPS64) > case RR_DSRA: > check_insn(ctx, ISA_MIPS3); > check_mips_64(ctx); > @@ -14158,7 +14162,7 @@ static int decode_mips16_opc (CPUMIPSState *env, DisasContext *ctx) > case RR_DIVU: > gen_muldiv(ctx, OPC_DIVU, 0, rx, ry); > break; > -#if defined (TARGET_MIPS64) > +#if defined(TARGET_MIPS64) > case RR_DMULT: > check_insn(ctx, ISA_MIPS3); > check_mips_64(ctx); > @@ -14802,7 +14806,7 @@ enum { > ADDIUR1SP = 0x1 > }; > > -static int mmreg (int r) > +static int mmreg(int r) > { > static const int map[] = { 16, 17, 2, 3, 4, 5, 6, 7 }; > > @@ -14810,7 +14814,7 @@ static int mmreg (int r) > } > > /* Used for 16-bit store instructions. */ > -static int mmreg2 (int r) > +static int mmreg2(int r) > { > static const int map[] = { 0, 17, 2, 3, 4, 5, 6, 7 }; > > @@ -14885,8 +14889,8 @@ static void gen_andi16(DisasContext *ctx) > gen_logic_imm(ctx, OPC_ANDI, rd, rs, decoded_imm[encoded]); > } > > -static void gen_ldst_multiple (DisasContext *ctx, uint32_t opc, int reglist, > - int base, int16_t offset) > +static void gen_ldst_multiple(DisasContext *ctx, uint32_t opc, int reglist, > + int base, int16_t offset) > { > TCGv t0, t1; > TCGv_i32 t2; > @@ -15159,7 +15163,7 @@ static void gen_pool16c_r6_insn(DisasContext *ctx) > } > } > > -static void gen_ldxs (DisasContext *ctx, int base, int index, int rd) > +static void gen_ldxs(DisasContext *ctx, int base, int index, int rd) > { > TCGv t0 = tcg_temp_new(); > TCGv t1 = tcg_temp_new(); > @@ -15179,8 +15183,8 @@ static void gen_ldxs (DisasContext *ctx, int base, int index, int rd) > tcg_temp_free(t1); > } > > -static void gen_ldst_pair (DisasContext *ctx, uint32_t opc, int rd, > - int base, int16_t offset) > +static void gen_ldst_pair(DisasContext *ctx, uint32_t opc, int rd, > + int base, int16_t offset) > { > TCGv t0, t1; > > @@ -15270,7 +15274,7 @@ static void gen_sync(int stype) > tcg_gen_mb(tcg_mo); > } > > -static void gen_pool32axf (CPUMIPSState *env, DisasContext *ctx, int rt, int rs) > +static void gen_pool32axf(CPUMIPSState *env, DisasContext *ctx, int rt, int rs) > { > int extension = (ctx->opcode >> 6) & 0x3f; > int minor = (ctx->opcode >> 12) & 0xf; > @@ -17234,7 +17238,7 @@ static void decode_micromips32_opc(CPUMIPSState *env, DisasContext *ctx) > } > } > > -static int decode_micromips_opc (CPUMIPSState *env, DisasContext *ctx) > +static int decode_micromips_opc(CPUMIPSState *env, DisasContext *ctx) > { > uint32_t op; > >