From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:35542) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ejRAD-0004Jh-TI for qemu-devel@nongnu.org; Wed, 07 Feb 2018 09:57:35 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1ejRA8-000203-Pp for qemu-devel@nongnu.org; Wed, 07 Feb 2018 09:57:33 -0500 Received: from mail-wm0-x242.google.com ([2a00:1450:400c:c09::242]:53155) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_128_CBC_SHA1:16) (Exim 4.71) (envelope-from ) id 1ejRA8-0001zt-Fj for qemu-devel@nongnu.org; Wed, 07 Feb 2018 09:57:28 -0500 Received: by mail-wm0-x242.google.com with SMTP id g1so3666102wmg.2 for ; Wed, 07 Feb 2018 06:57:28 -0800 (PST) References: <20180207111729.15737-1-ard.biesheuvel@linaro.org> <20180207111729.15737-6-ard.biesheuvel@linaro.org> From: Alex =?utf-8?Q?Benn=C3=A9e?= In-reply-to: <20180207111729.15737-6-ard.biesheuvel@linaro.org> Date: Wed, 07 Feb 2018 14:57:25 +0000 Message-ID: <87372cx2y2.fsf@linaro.org> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [PATCH v6 5/5] target/arm: enable user-mode SHA-3, SM3, SM4 and SHA-512 instruction support List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Ard Biesheuvel Cc: qemu-devel@nongnu.org, peter.maydell@linaro.org Ard Biesheuvel writes: > Add support for the new ARMv8.2 SHA-3, SM3, SM4 and SHA-512 instructions = to > AArch64 user mode emulation. So another problem I've come across is I can't turn this off. I ended up doing that in my FP16 series because otherwise existing RISU tests get broken. However having an optional flag for each little set of instructions seems overkill. Have you run any RISU tests? If you want you can add this to the aarch64.risu to generate some test patterns. # These are optional ARMv8.2 cryptographic extensions @v8.2,Cryptographic,CryptographicSHA # Cryptographic three-register SHA 512 # 31 21 20 16 15 14 13 12 11 10 9 5 4 0 # 11001110011 Rm 1 O 0 0 opcode Rn Rd SHA512H A64_V 1100 1110 011 rm:5 1 0 00 00 rn:5 rd:5 SHA512H2 A64_V 1100 1110 011 rm:5 1 0 00 01 rn:5 rd:5 SHA512SUI A64_V 1100 1110 011 rm:5 1 0 00 10 rn:5 rd:5 RAX1 A64_V 1100 1110 011 rm:5 1 0 00 11 rn:5 rd:5 SM3PARTW1 A64_V 1100 1110 011 rm:5 1 1 00 00 rn:5 rd:5 SM3PARTW2 A64_V 1100 1110 011 rm:5 1 1 00 01 rn:5 rd:5 SM4 A64_V 1100 1110 011 rm:5 1 1 00 10 rn:5 rd:5 # Cryptographic four-register # 31 23 22 21 20 16 15 14 10 9 5 4 0 # 1100 1110 0 Op0 Rm 0 Ra Rn Rd EOR3 A64_V 1110 1110 0 00 rm:5 0 ra:5 rn:5 rd:5 BCAX A64_V 1110 1110 0 01 rm:5 0 ra:5 rn:5 rd:5 SM3SS1 A64_V 1110 1110 0 10 rm:5 0 ra:5 rn:5 rd:5 # Cryptographic two-register SHA 512 # 31 12 11 10 9 5 4 0 # 1100 1110 1100 0000 1000 op Rn Rd SHA512SU0 A64_V 1100 1110 1100 0000 1000 00 rn:5 rd:5 SM4E A64_V 1100 1110 1100 0000 1000 01 rn:5 rd:5 @ > > Signed-off-by: Ard Biesheuvel > --- > linux-user/elfload.c | 19 +++++++++++++++++++ > target/arm/cpu64.c | 4 ++++ > 2 files changed, 23 insertions(+) > > diff --git a/linux-user/elfload.c b/linux-user/elfload.c > index 20f3d8c2c373..7922ab8eab79 100644 > --- a/linux-user/elfload.c > +++ b/linux-user/elfload.c > @@ -512,6 +512,21 @@ enum { > ARM_HWCAP_A64_SHA1 =3D 1 << 5, > ARM_HWCAP_A64_SHA2 =3D 1 << 6, > ARM_HWCAP_A64_CRC32 =3D 1 << 7, > + ARM_HWCAP_A64_ATOMICS =3D 1 << 8, > + ARM_HWCAP_A64_FPHP =3D 1 << 9, > + ARM_HWCAP_A64_ASIMDHP =3D 1 << 10, > + ARM_HWCAP_A64_CPUID =3D 1 << 11, > + ARM_HWCAP_A64_ASIMDRDM =3D 1 << 12, > + ARM_HWCAP_A64_JSCVT =3D 1 << 13, > + ARM_HWCAP_A64_FCMA =3D 1 << 14, > + ARM_HWCAP_A64_LRCPC =3D 1 << 15, > + ARM_HWCAP_A64_DCPOP =3D 1 << 16, > + ARM_HWCAP_A64_SHA3 =3D 1 << 17, > + ARM_HWCAP_A64_SM3 =3D 1 << 18, > + ARM_HWCAP_A64_SM4 =3D 1 << 19, > + ARM_HWCAP_A64_ASIMDDP =3D 1 << 20, > + ARM_HWCAP_A64_SHA512 =3D 1 << 21, > + ARM_HWCAP_A64_SVE =3D 1 << 22, > }; > > #define ELF_HWCAP get_elf_hwcap() > @@ -532,6 +547,10 @@ static uint32_t get_elf_hwcap(void) > GET_FEATURE(ARM_FEATURE_V8_SHA1, ARM_HWCAP_A64_SHA1); > GET_FEATURE(ARM_FEATURE_V8_SHA256, ARM_HWCAP_A64_SHA2); > GET_FEATURE(ARM_FEATURE_CRC, ARM_HWCAP_A64_CRC32); > + GET_FEATURE(ARM_FEATURE_V8_SHA3, ARM_HWCAP_A64_SHA3); > + GET_FEATURE(ARM_FEATURE_V8_SM3, ARM_HWCAP_A64_SM3); > + GET_FEATURE(ARM_FEATURE_V8_SM4, ARM_HWCAP_A64_SM4); > + GET_FEATURE(ARM_FEATURE_V8_SHA512, ARM_HWCAP_A64_SHA512); > #undef GET_FEATURE > > return hwcaps; > diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c > index 670c07ab6ed4..1c330adc281b 100644 > --- a/target/arm/cpu64.c > +++ b/target/arm/cpu64.c > @@ -224,6 +224,10 @@ static void aarch64_any_initfn(Object *obj) > set_feature(&cpu->env, ARM_FEATURE_V8_AES); > set_feature(&cpu->env, ARM_FEATURE_V8_SHA1); > set_feature(&cpu->env, ARM_FEATURE_V8_SHA256); > + set_feature(&cpu->env, ARM_FEATURE_V8_SHA512); > + set_feature(&cpu->env, ARM_FEATURE_V8_SHA3); > + set_feature(&cpu->env, ARM_FEATURE_V8_SM3); > + set_feature(&cpu->env, ARM_FEATURE_V8_SM4); > set_feature(&cpu->env, ARM_FEATURE_V8_PMULL); > set_feature(&cpu->env, ARM_FEATURE_CRC); > cpu->ctr =3D 0x80038003; /* 32 byte I and D cacheline size, VIPT ica= che */ -- Alex Benn=C3=A9e