From patchwork Tue Feb 28 21:54:34 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: =?utf-8?q?Heiko_St=C3=BCbner?= X-Patchwork-Id: 13155286 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 bombadil.infradead.org (bombadil.infradead.org [198.137.202.133]) (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 93543C64EC4 for ; Tue, 28 Feb 2023 21:54:59 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=lists.infradead.org; s=bombadil.20210309; h=Sender: Content-Transfer-Encoding:Content-Type:List-Subscribe:List-Help:List-Post: List-Archive:List-Unsubscribe:List-Id:MIME-Version:References:In-Reply-To: Message-Id:Date:Subject:Cc:To:From:Reply-To:Content-ID:Content-Description: Resent-Date:Resent-From:Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID: List-Owner; bh=gD10s0+HW542+58qRX9g3YfjKJvH5rm2f7tZ1nVAmNY=; b=RO8NZz+umUqmvY piDDVBKAXc0AO0JtL/2QyUPtj+x3GmX3qjRqLUsA9z3S8zrjxgjI0ZQkCed9c8mboh3JwCAMgtdWc IhirhGsc4YY2QjFxfSLy2fVryu97VHI7gGjjle2y4noyadWEhXdiuZiIeZ2n1jNJGDKA6guX0dMcF lC5DmvKMB3QPCizDHcyn+KTz0kkuGTDX37WRJTrKDBqXi1Fc6cQF/Uio+1abHjy9mxSzJCkkFRO8d DhqUBAkkkAL4j3ozJEQN/n2fdCcD9vKGh1+yu21hAxuT9QwtjwEOQzpl6PxIw8IGYHibmd3rsCGsn TwqcealvkSRk0Jwuwi4Q==; Received: from localhost ([::1] helo=bombadil.infradead.org) by bombadil.infradead.org with esmtp (Exim 4.94.2 #2 (Red Hat Linux)) id 1pX7vu-00EOZm-OW; Tue, 28 Feb 2023 21:54:50 +0000 Received: from gloria.sntech.de ([185.11.138.130]) by bombadil.infradead.org with esmtps (Exim 4.94.2 #2 (Red Hat Linux)) id 1pX7vr-00EOWt-KJ for linux-riscv@lists.infradead.org; Tue, 28 Feb 2023 21:54:49 +0000 Received: from ip5b412258.dynamic.kabel-deutschland.de ([91.65.34.88] helo=phil.lan) by gloria.sntech.de with esmtpsa (TLS1.3) tls TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384 (Exim 4.94.2) (envelope-from ) id 1pX7vm-0005hs-PZ; Tue, 28 Feb 2023 22:54:42 +0100 From: Heiko Stuebner To: palmer@dabbelt.com Cc: linux-riscv@lists.infradead.org, samuel@sholland.org, guoren@kernel.org, christoph.muellner@vrull.eu, heiko@sntech.de, conor.dooley@microchip.com, linux-kernel@vger.kernel.org, Heiko Stuebner Subject: [PATCH RFC 1/2] RISC-V: define the elements of the VCSR vector CSR Date: Tue, 28 Feb 2023 22:54:34 +0100 Message-Id: <20230228215435.3366914-2-heiko@sntech.de> X-Mailer: git-send-email 2.39.0 In-Reply-To: <20230228215435.3366914-1-heiko@sntech.de> References: <20230228215435.3366914-1-heiko@sntech.de> MIME-Version: 1.0 X-CRM114-Version: 20100106-BlameMichelson ( TRE 0.8.0 (BSD) ) MR-646709E3 X-CRM114-CacheID: sfid-20230228_135447_720269_3731618C X-CRM114-Status: UNSURE ( 8.80 ) X-CRM114-Notice: Please train this message. X-BeenThere: linux-riscv@lists.infradead.org X-Mailman-Version: 2.1.34 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: "linux-riscv" Errors-To: linux-riscv-bounces+linux-riscv=archiver.kernel.org@lists.infradead.org From: Heiko Stuebner The VCSR CSR contains two elements VXRM[2:1] and VXSAT[0]. Define constants for those to access the elements in a readable way. Signed-off-by: Heiko Stuebner Acked-by: Guo Ren Reviewed-by: Conor Dooley --- arch/riscv/include/asm/csr.h | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/arch/riscv/include/asm/csr.h b/arch/riscv/include/asm/csr.h index add51662b7c3..8b06f2472915 100644 --- a/arch/riscv/include/asm/csr.h +++ b/arch/riscv/include/asm/csr.h @@ -176,6 +176,11 @@ #define ENVCFG_CBIE_INV _AC(0x3, UL) #define ENVCFG_FIOM _AC(0x1, UL) +/* VCSR flags */ +#define VCSR_VXRM_MASK 3 +#define VCSR_VXRM_SHIFT 1 +#define VCSR_VXSAT_MASK 1 + /* symbolic CSR names: */ #define CSR_CYCLE 0xc00 #define CSR_TIME 0xc01 From patchwork Tue Feb 28 21:54:35 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: =?utf-8?q?Heiko_St=C3=BCbner?= X-Patchwork-Id: 13155288 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 bombadil.infradead.org (bombadil.infradead.org [198.137.202.133]) (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 11A4BC64EC7 for ; Tue, 28 Feb 2023 21:55:05 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=lists.infradead.org; s=bombadil.20210309; h=Sender: Content-Transfer-Encoding:Content-Type:List-Subscribe:List-Help:List-Post: List-Archive:List-Unsubscribe:List-Id:MIME-Version:References:In-Reply-To: Message-Id:Date:Subject:Cc:To:From:Reply-To:Content-ID:Content-Description: Resent-Date:Resent-From:Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID: List-Owner; bh=42T5RiBojetYxR7JYYdCUJVFXvWUly9PcoJGyJjTd8k=; b=Tl33uEmw2A/V3n 9i+gFwp6DKAnheEF4njiHiTU9NT2QTKSG37KL6diCQFfyl32uHrnks1p0VsnyoeOKNdhNReDIoh9H 4Uvum6eCsp5Tq0td+bIKTZdTfLyx0BpyjYmnYF8kQFs5dZ0vboO1omh2UusLi2Qo2jy4JaZVtYVpT wQrq0rU4hcz/yimhKhibKHt61+9173biJAF01nqVic+9sNFsFmRDXR3ELz9JEA3AMtIgZkIsjIicH hwW7LJ+VcBXSeTa+dVJTRuFc7OZ9GBpiYkf/1Ztjb3nZDK/xdXX23OQTLaCHdUYgRAiK6n1JgONzv Q+ca7h5x8lAg+G/R9mGA==; Received: from localhost ([::1] helo=bombadil.infradead.org) by bombadil.infradead.org with esmtp (Exim 4.94.2 #2 (Red Hat Linux)) id 1pX7w1-00EOcG-1U; Tue, 28 Feb 2023 21:54:57 +0000 Received: from gloria.sntech.de ([185.11.138.130]) by bombadil.infradead.org with esmtps (Exim 4.94.2 #2 (Red Hat Linux)) id 1pX7vr-00EOWu-KG for linux-riscv@lists.infradead.org; Tue, 28 Feb 2023 21:54:51 +0000 Received: from ip5b412258.dynamic.kabel-deutschland.de ([91.65.34.88] helo=phil.lan) by gloria.sntech.de with esmtpsa (TLS1.3) tls TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384 (Exim 4.94.2) (envelope-from ) id 1pX7vn-0005hs-3N; Tue, 28 Feb 2023 22:54:43 +0100 From: Heiko Stuebner To: palmer@dabbelt.com Cc: linux-riscv@lists.infradead.org, samuel@sholland.org, guoren@kernel.org, christoph.muellner@vrull.eu, heiko@sntech.de, conor.dooley@microchip.com, linux-kernel@vger.kernel.org, Heiko Stuebner Subject: [PATCH RFC 2/2] RISC-V: add T-Head vector errata handling Date: Tue, 28 Feb 2023 22:54:35 +0100 Message-Id: <20230228215435.3366914-3-heiko@sntech.de> X-Mailer: git-send-email 2.39.0 In-Reply-To: <20230228215435.3366914-1-heiko@sntech.de> References: <20230228215435.3366914-1-heiko@sntech.de> MIME-Version: 1.0 X-CRM114-Version: 20100106-BlameMichelson ( TRE 0.8.0 (BSD) ) MR-646709E3 X-CRM114-CacheID: sfid-20230228_135447_895255_B3CFC2D6 X-CRM114-Status: GOOD ( 25.12 ) X-BeenThere: linux-riscv@lists.infradead.org X-Mailman-Version: 2.1.34 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: "linux-riscv" Errors-To: linux-riscv-bounces+linux-riscv=archiver.kernel.org@lists.infradead.org From: Heiko Stuebner T-Head C9xx cores implement an older version (0.7.1) of the vector specification. Relevant changes concerning the kernel are: - different placement of the SR_VS bit for the vector unit status - different encoding of the vsetvli instruction - different instructions for loads and stores And a fixed VLEN of 128. The in-kernel access to vector instances is limited to the save and restore of process states so the above mentioned areas can simply be handled via the alternatives framework, similar to other T-Head specific issues. Signed-off-by: Heiko Stuebner --- arch/riscv/Kconfig.erratas | 13 +++ arch/riscv/errata/thead/errata.c | 32 ++++++ arch/riscv/include/asm/csr.h | 26 ++++- arch/riscv/include/asm/errata_list.h | 62 +++++++++++- arch/riscv/include/asm/vector.h | 139 +++++++++++++++++++++++++-- 5 files changed, 256 insertions(+), 16 deletions(-) diff --git a/arch/riscv/Kconfig.erratas b/arch/riscv/Kconfig.erratas index 69621ae6d647..624cefc9fcd7 100644 --- a/arch/riscv/Kconfig.erratas +++ b/arch/riscv/Kconfig.erratas @@ -79,4 +79,17 @@ config ERRATA_THEAD_PMU If you don't know what to do here, say "Y". +config ERRATA_THEAD_VECTOR + bool "Apply T-Head Vector errata" + depends on ERRATA_THEAD && RISCV_ISA_V + default y + help + The T-Head C9xx cores implement an earlier version 0.7.1 + of the vector extensions. + + This will apply the necessary errata to handle the non-standard + behaviour via when switch to and from vector mode for processes. + + If you don't know what to do here, say "Y". + endmenu # "CPU errata selection" diff --git a/arch/riscv/errata/thead/errata.c b/arch/riscv/errata/thead/errata.c index fac5742d1c1e..55b3aaa2468a 100644 --- a/arch/riscv/errata/thead/errata.c +++ b/arch/riscv/errata/thead/errata.c @@ -12,6 +12,7 @@ #include #include #include +#include #include static bool errata_probe_pbmt(unsigned int stage, @@ -63,6 +64,34 @@ static bool errata_probe_pmu(unsigned int stage, return true; } +static bool errata_probe_vector(unsigned int stage, + unsigned long arch_id, unsigned long impid) +{ + if (!IS_ENABLED(CONFIG_ERRATA_THEAD_VECTOR)) + return false; + + /* target-c9xx cores report arch_id and impid as 0 */ + if (arch_id != 0 || impid != 0) + return false; + + if (stage == RISCV_ALTERNATIVES_EARLY_BOOT) { + /* + * Disable VECTOR to detect illegal usage of vector in kernel. + * This is normally done in _start_kernel but with the + * vector-1.0 SR_VS bits. VS is using [24:23] on T-Head's + * vector-0.7.1 and the vector-1.0-bits are unused there. + */ + csr_clear(CSR_STATUS, SR_VS_THEAD); + return false; + } + + /* let has_vector() return true and set the static vlen */ + static_branch_enable(&riscv_isa_ext_keys[RISCV_ISA_EXT_KEY_VECTOR]); + riscv_vsize = 128 / 8 * 32; + + return true; +} + static u32 thead_errata_probe(unsigned int stage, unsigned long archid, unsigned long impid) { @@ -77,6 +106,9 @@ static u32 thead_errata_probe(unsigned int stage, if (errata_probe_pmu(stage, archid, impid)) cpu_req_errata |= BIT(ERRATA_THEAD_PMU); + if (errata_probe_vector(stage, archid, impid)) + cpu_req_errata |= BIT(ERRATA_THEAD_VECTOR); + return cpu_req_errata; } diff --git a/arch/riscv/include/asm/csr.h b/arch/riscv/include/asm/csr.h index 8b06f2472915..8d16c11487aa 100644 --- a/arch/riscv/include/asm/csr.h +++ b/arch/riscv/include/asm/csr.h @@ -24,11 +24,27 @@ #define SR_FS_CLEAN _AC(0x00004000, UL) #define SR_FS_DIRTY _AC(0x00006000, UL) -#define SR_VS _AC(0x00000600, UL) /* Vector Status */ -#define SR_VS_OFF _AC(0x00000000, UL) -#define SR_VS_INITIAL _AC(0x00000200, UL) -#define SR_VS_CLEAN _AC(0x00000400, UL) -#define SR_VS_DIRTY _AC(0x00000600, UL) +#define SR_VS_OFF _AC(0x00000000, UL) + +#define SR_VS_1_0 _AC(0x00000600, UL) /* Vector Status */ +#define SR_VS_INITIAL_1_0 _AC(0x00000200, UL) +#define SR_VS_CLEAN_1_0 _AC(0x00000400, UL) +#define SR_VS_DIRTY_1_0 _AC(0x00000600, UL) + +#define SR_VS_THEAD _AC(0x01800000, UL) /* Vector Status */ +#define SR_VS_INITIAL_THEAD _AC(0x00800000, UL) +#define SR_VS_CLEAN_THEAD _AC(0x01000000, UL) +#define SR_VS_DIRTY_THEAD _AC(0x01800000, UL) + +/* + * Always default to vector-1.0 handling in assembly and let the broken + * implementations handle their case separately. + */ +#ifdef __ASSEMBLY__ +#define SR_VS SR_VS_1_0 +#else + +#endif #define SR_XS _AC(0x00018000, UL) /* Extension Status */ #define SR_XS_OFF _AC(0x00000000, UL) diff --git a/arch/riscv/include/asm/errata_list.h b/arch/riscv/include/asm/errata_list.h index 95e626b7281e..3f93cdd1599f 100644 --- a/arch/riscv/include/asm/errata_list.h +++ b/arch/riscv/include/asm/errata_list.h @@ -19,7 +19,8 @@ #define ERRATA_THEAD_PBMT 0 #define ERRATA_THEAD_CMO 1 #define ERRATA_THEAD_PMU 2 -#define ERRATA_THEAD_NUMBER 3 +#define ERRATA_THEAD_VECTOR 3 +#define ERRATA_THEAD_NUMBER 4 #endif #define CPUFEATURE_SVPBMT 0 @@ -157,6 +158,65 @@ asm volatile(ALTERNATIVE( \ : "=r" (__ovl) : \ : "memory") +#ifdef CONFIG_ERRATA_THEAD_PBMT + +#define ALT_VS_SHIFT 61 +#define ALT_THEAD_VS_SHIFT 59 +#define ALT_THEAD_SR_VS(_val, _vs) \ +asm(ALTERNATIVE( "li %0, %1\t\nslli %0,%0,%3", \ + "li %0, %2\t\nslli %0,%0,%4", THEAD_VENDOR_ID, \ + ERRATA_THEAD_PBMT, CONFIG_ERRATA_THEAD_PBMT) \ + : "=r"(_val) \ + : "I"(prot##_MAIN >> ALT_SVPBMT_SHIFT), \ + "I"(prot##_THEAD >> ALT_THEAD_PBMT_SHIFT), \ + "I"(ALT_SVPBMT_SHIFT), \ + "I"(ALT_THEAD_PBMT_SHIFT)) +#else +#define ALT_THEAD_SR_VS(_val ## _MAIN) +#endif + +#ifdef CONFIG_ERRATA_THEAD_VECTOR + +#define THEAD_C9XX_CSR_VXSAT 0x9 +#define THEAD_C9XX_CSR_VXRM 0xa + +/* + * Vector 0.7.1 as used for example on T-Head Xuantie cores, uses an older + * encoding for vsetvli (ta, ma vs. d1), so provide an instruction for + * vsetvli t4, x0, e8, m8, d1 + */ +#define THEAD_VSETVLI_T4X0E8M8D1 ".long 0x00307ed7\n\t" + +/* + * While in theory, the vector-0.7.1 vsb.v and vlb.v result in the same + * encoding as the standard vse8.v and vle8.v, compilers seem to optimize + * the call resulting in a different encoding and then using a value for + * the "mop" field that is not part of vector-0.7.1 + * So encode specific variants for vstate_save and _restore. + */ +#define THEAD_VSB_V_V0T0 ".long 0x02028027\n\t" +#define THEAD_VSB_V_V8T0 ".long 0x02028427\n\t" +#define THEAD_VSB_V_V16T0 ".long 0x02028827\n\t" +#define THEAD_VSB_V_V24T0 ".long 0x02028c27\n\t" +#define THEAD_VLB_V_V0T0 ".long 0x012028007\n\t" +#define THEAD_VLB_V_V8T0 ".long 0x012028407\n\t" +#define THEAD_VLB_V_V16T0 ".long 0x012028807\n\t" +#define THEAD_VLB_V_V24T0 ".long 0x012028c07\n\t" + +#define ALT_SR_VS_VECTOR_1_0_SHIFT 9 +#define ALT_SR_VS_THEAD_SHIFT 23 + +#define ALT_SR_VS(_val, prot) \ +asm(ALTERNATIVE("li %0, %1\t\nslli %0,%0,%3", \ + "li %0, %2\t\nslli %0,%0,%4", THEAD_VENDOR_ID, \ + ERRATA_THEAD_VECTOR, CONFIG_ERRATA_THEAD_VECTOR) \ + : "=r"(_val) \ + : "I"(prot##_1_0 >> ALT_SR_VS_VECTOR_1_0_SHIFT), \ + "I"(prot##_THEAD >> ALT_SR_VS_THEAD_SHIFT), \ + "I"(ALT_SR_VS_VECTOR_1_0_SHIFT), \ + "I"(ALT_SR_VS_THEAD_SHIFT)) +#endif /* CONFIG_ERRATA_THEAD_VECTOR */ + #endif /* __ASSEMBLY__ */ #endif diff --git a/arch/riscv/include/asm/vector.h b/arch/riscv/include/asm/vector.h index ad9e6161dd89..ad91f783316e 100644 --- a/arch/riscv/include/asm/vector.h +++ b/arch/riscv/include/asm/vector.h @@ -15,6 +15,55 @@ #include #include #include +#include + +#ifdef CONFIG_ERRATA_THEAD_VECTOR + +static inline u32 riscv_sr_vs(void) +{ + u32 val; + + ALT_SR_VS(val, SR_VS); + return val; +} + +static inline u32 riscv_sr_vs_initial(void) +{ + u32 val; + + ALT_SR_VS(val, SR_VS_INITIAL); + return val; +} + +static inline u32 riscv_sr_vs_clean(void) +{ + u32 val; + + ALT_SR_VS(val, SR_VS_CLEAN); + return val; +} + +static inline u32 riscv_sr_vs_dirty(void) +{ + u32 val; + + ALT_SR_VS(val, SR_VS_DIRTY); + return val; +} + +#define SR_VS riscv_sr_vs() +#define SR_VS_INITIAL riscv_sr_vs_initial() +#define SR_VS_CLEAN riscv_sr_vs_clean() +#define SR_VS_DIRTY riscv_sr_vs_dirty() + +#else /* CONFIG_ERRATA_THEAD_VECTOR */ + +#define SR_VS SR_VS_1_0 +#define SR_VS_INITIAL SR_VS_INITIAL_1_0 +#define SR_VS_CLEAN SR_VS_CLEAN_1_0 +#define SR_VS_DIRTY SR_VS_DIRTY_1_0 + +#endif /* CONFIG_ERRATA_THEAD_VECTOR */ #define CSR_STR(x) __ASM_STR(x) @@ -29,6 +78,7 @@ static __always_inline bool has_vector(void) static inline void __vstate_clean(struct pt_regs *regs) { regs->status = (regs->status & ~(SR_VS)) | SR_VS_CLEAN; + } static inline void vstate_off(struct pt_regs *regs) @@ -58,30 +108,75 @@ static __always_inline void rvv_disable(void) static __always_inline void __vstate_csr_save(struct __riscv_v_state *dest) { - asm volatile ( + register u32 t1 asm("t1") = (SR_FS); + + /* + * CSR_VCSR is defined as + * [2:1] - vxrm[1:0] + * [0] - vxsat + * The earlier vector spec implemented by T-Head uses separate + * registers for the same bit-elements, so just combine those + * into the existing output field. + * + * Additionally T-Head cores need FS to be enabled when accessing + * the VXRM and VXSAT CSRs, otherwise ending in illegal instructions. + */ + asm volatile (ALTERNATIVE( "csrr %0, " CSR_STR(CSR_VSTART) "\n\t" "csrr %1, " CSR_STR(CSR_VTYPE) "\n\t" "csrr %2, " CSR_STR(CSR_VL) "\n\t" "csrr %3, " CSR_STR(CSR_VCSR) "\n\t" + __nops(5), + "csrs sstatus, t1\n\t" + "csrr %0, " CSR_STR(CSR_VSTART) "\n\t" + "csrr %1, " CSR_STR(CSR_VTYPE) "\n\t" + "csrr %2, " CSR_STR(CSR_VL) "\n\t" + "csrr %3, " CSR_STR(THEAD_C9XX_CSR_VXRM) "\n\t" + "slliw %3, %3, " CSR_STR(VCSR_VXRM_SHIFT) "\n\t" + "csrr t4, " CSR_STR(THEAD_C9XX_CSR_VXSAT) "\n\t" + "or %3, %3, t4\n\t" + "csrc sstatus, t1\n\t", + THEAD_VENDOR_ID, + ERRATA_THEAD_VECTOR, CONFIG_ERRATA_THEAD_VECTOR) : "=r" (dest->vstart), "=r" (dest->vtype), "=r" (dest->vl), - "=r" (dest->vcsr) : :); + "=r" (dest->vcsr) : "r"(t1) : "t4"); } static __always_inline void __vstate_csr_restore(struct __riscv_v_state *src) { - asm volatile ( + register u32 t1 asm("t1") = (SR_FS); + + /* + * Similar to __vstate_csr_save above, restore values for the + * separate VXRM and VXSAT CSRs from the vcsr variable. + */ + asm volatile (ALTERNATIVE( "vsetvl x0, %2, %1\n\t" "csrw " CSR_STR(CSR_VSTART) ", %0\n\t" "csrw " CSR_STR(CSR_VCSR) ", %3\n\t" + __nops(6), + "csrs sstatus, t1\n\t" + "vsetvl x0, %2, %1\n\t" + "csrw " CSR_STR(CSR_VSTART) ", %0\n\t" + "srliw t4, %3, " CSR_STR(VCSR_VXRM_SHIFT) "\n\t" + "andi t4, t4, " CSR_STR(VCSR_VXRM_MASK) "\n\t" + "csrw " CSR_STR(THEAD_C9XX_CSR_VXRM) ", t4\n\t" + "andi %3, %3, " CSR_STR(VCSR_VXSAT_MASK) "\n\t" + "csrw " CSR_STR(THEAD_C9XX_CSR_VXSAT) ", %3\n\t" + "csrc sstatus, t1\n\t", + THEAD_VENDOR_ID, + ERRATA_THEAD_VECTOR, CONFIG_ERRATA_THEAD_VECTOR) : : "r" (src->vstart), "r" (src->vtype), "r" (src->vl), - "r" (src->vcsr) :); + "r" (src->vcsr), "r"(t1): "t4"); } static inline void __vstate_save(struct __riscv_v_state *save_to, void *datap) { rvv_enable(); __vstate_csr_save(save_to); - asm volatile ( + + asm volatile (ALTERNATIVE( + "nop\n\t" "vsetvli t4, x0, e8, m8, ta, ma\n\t" "vse8.v v0, (%0)\n\t" "add %0, %0, t4\n\t" @@ -89,8 +184,18 @@ static inline void __vstate_save(struct __riscv_v_state *save_to, void *datap) "add %0, %0, t4\n\t" "vse8.v v16, (%0)\n\t" "add %0, %0, t4\n\t" - "vse8.v v24, (%0)\n\t" - : : "r" (datap) : "t4", "memory"); + "vse8.v v24, (%0)\n\t", + "mv t0, %0\n\t" + THEAD_VSETVLI_T4X0E8M8D1 + THEAD_VSB_V_V0T0 + "addi t0, t0, 128\n\t" + THEAD_VSB_V_V8T0 + "addi t0, t0, 128\n\t" + THEAD_VSB_V_V16T0 + "addi t0, t0, 128\n\t" + THEAD_VSB_V_V24T0, THEAD_VENDOR_ID, + ERRATA_THEAD_VECTOR, CONFIG_ERRATA_THEAD_VECTOR) + : : "r" (datap) : "t0", "t4", "memory"); rvv_disable(); } @@ -98,7 +203,9 @@ static inline void __vstate_restore(struct __riscv_v_state *restore_from, void *datap) { rvv_enable(); - asm volatile ( + + asm volatile (ALTERNATIVE( + "nop\n\t" "vsetvli t4, x0, e8, m8, ta, ma\n\t" "vle8.v v0, (%0)\n\t" "add %0, %0, t4\n\t" @@ -106,8 +213,20 @@ static inline void __vstate_restore(struct __riscv_v_state *restore_from, "add %0, %0, t4\n\t" "vle8.v v16, (%0)\n\t" "add %0, %0, t4\n\t" - "vle8.v v24, (%0)\n\t" - : : "r" (datap) : "t4"); + "vle8.v v24, (%0)\n\t", + + "mv t0, %0\n\t" + THEAD_VSETVLI_T4X0E8M8D1 + THEAD_VLB_V_V0T0 + "addi t0, t0, 128\n\t" + THEAD_VLB_V_V8T0 + "addi %0, %0, 128\n\t" + THEAD_VLB_V_V16T0 + "addi %0, %0, 128\n\t" + THEAD_VLB_V_V24T0, THEAD_VENDOR_ID, + ERRATA_THEAD_VECTOR, CONFIG_ERRATA_THEAD_VECTOR) + : : "r" (datap) : "t0", "t4"); + __vstate_csr_restore(restore_from); rvv_disable(); }