| Message ID | 20240204054228.651-3-zhiwei_liu@linux.alibaba.com |
|---|---|
| State | New |
| Headers | show |
| Series | target/riscv: Support mxstatus CSR for thead-c906 | expand |
On Sun, Feb 4, 2024 at 3:44 PM LIU Zhiwei <zhiwei_liu@linux.alibaba.com> wrote: > > This patch set fix the regression on kernel pointed by Björn Töpel in > https://www.mail-archive.com/qemu-devel@nongnu.org/msg1018232.html. > > thead-c906 uses some flags in pte [60-63] bits. It has history reasons that > SVPBMT didn't exist when thead-c906 came to wotrld. We named this feature as > xtheadmaee[1]. this feature is controlled by an custom CSR named mxstatus, > whose maee field encodes whether enable the pte [60-63] bits. > > [1]:https://github.com/T-head-Semi/thead-extension-spec/blob/master/xtheadmaee.adoc > > Signed-off-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com> > --- > v1->v2: > 1) Remove mxstatus user mode access > 2) Add reference documentation to the commit log > --- > target/riscv/cpu.c | 6 ++++ > target/riscv/cpu.h | 9 ++++++ > target/riscv/cpu_bits.h | 6 ++++ > target/riscv/cpu_cfg.h | 4 ++- > target/riscv/cpu_helper.c | 25 ++++++++------- > target/riscv/meson.build | 1 + > target/riscv/tcg/tcg-cpu.c | 7 +++- > target/riscv/xthead_csr.c | 65 ++++++++++++++++++++++++++++++++++++++ > 8 files changed, 110 insertions(+), 13 deletions(-) > create mode 100644 target/riscv/xthead_csr.c > > diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c > index 2dcbc9ff32..bfdbb0539a 100644 > --- a/target/riscv/cpu.c > +++ b/target/riscv/cpu.c > @@ -171,6 +171,7 @@ const RISCVIsaExtData isa_edata_arr[] = { > ISA_EXT_DATA_ENTRY(xtheadmemidx, PRIV_VERSION_1_11_0, ext_xtheadmemidx), > ISA_EXT_DATA_ENTRY(xtheadmempair, PRIV_VERSION_1_11_0, ext_xtheadmempair), > ISA_EXT_DATA_ENTRY(xtheadsync, PRIV_VERSION_1_11_0, ext_xtheadsync), > + ISA_EXT_DATA_ENTRY(xtheadmaee, PRIV_VERSION_1_11_0, ext_xtheadmaee), > ISA_EXT_DATA_ENTRY(xventanacondops, PRIV_VERSION_1_12_0, ext_XVentanaCondOps), > > DEFINE_PROP_END_OF_LIST(), > @@ -506,6 +507,7 @@ static void rv64_thead_c906_cpu_init(Object *obj) > > cpu->cfg.mvendorid = THEAD_VENDOR_ID; > #ifndef CONFIG_USER_ONLY > + cpu->cfg.ext_xtheadmaee = true; > set_satp_mode_max_supported(cpu, VM_1_10_SV39); > #endif > > @@ -949,6 +951,9 @@ static void riscv_cpu_reset_hold(Object *obj) > } > > pmp_unlock_entries(env); > + if (riscv_cpu_cfg(env)->ext_xtheadmaee) { > + env->th_mxstatus |= TH_MXSTATUS_MAEE; > + } > #endif > env->xl = riscv_cpu_mxl(env); > riscv_cpu_update_mask(env); > @@ -1439,6 +1444,7 @@ const RISCVCPUMultiExtConfig riscv_cpu_vendor_exts[] = { > MULTI_EXT_CFG_BOOL("xtheadmemidx", ext_xtheadmemidx, false), > MULTI_EXT_CFG_BOOL("xtheadmempair", ext_xtheadmempair, false), > MULTI_EXT_CFG_BOOL("xtheadsync", ext_xtheadsync, false), > + MULTI_EXT_CFG_BOOL("xtheadmaee", ext_xtheadmaee, false), > MULTI_EXT_CFG_BOOL("xventanacondops", ext_XVentanaCondOps, false), > > DEFINE_PROP_END_OF_LIST(), > diff --git a/target/riscv/cpu.h b/target/riscv/cpu.h > index 5f3955c38d..1bacf40355 100644 > --- a/target/riscv/cpu.h > +++ b/target/riscv/cpu.h > @@ -412,6 +412,14 @@ struct CPUArchState { > target_ulong cur_pmmask; > target_ulong cur_pmbase; > > + union { > + /* Custom CSR for Xuantie CPU */ > + struct { > +#ifndef CONFIG_USER_ONLY > + target_ulong th_mxstatus; > +#endif > + }; > + }; > /* Fields from here on are preserved across CPU reset. */ > QEMUTimer *stimer; /* Internal timer for S-mode interrupt */ > QEMUTimer *vstimer; /* Internal timer for VS-mode interrupt */ > @@ -799,6 +807,7 @@ void riscv_add_satp_mode_properties(Object *obj); > bool riscv_cpu_accelerator_compatible(RISCVCPU *cpu); > > /* CSR function table */ > +extern riscv_csr_operations th_csr_ops[CSR_TABLE_SIZE]; > extern riscv_csr_operations csr_ops[CSR_TABLE_SIZE]; > > extern const bool valid_vm_1_10_32[], valid_vm_1_10_64[]; > diff --git a/target/riscv/cpu_bits.h b/target/riscv/cpu_bits.h > index e116f6c252..67ebb1cefe 100644 > --- a/target/riscv/cpu_bits.h > +++ b/target/riscv/cpu_bits.h > @@ -897,4 +897,10 @@ typedef enum RISCVException { > /* JVT CSR bits */ > #define JVT_MODE 0x3F > #define JVT_BASE (~0x3F) > + > +/* Xuantie custom CSRs */ > +#define CSR_TH_MXSTATUS 0x7c0 > + > +#define TH_MXSTATUS_MAEE_SHIFT 21 > +#define TH_MXSTATUS_MAEE (0x1 << TH_MXSTATUS_MAEE_SHIFT) > #endif > diff --git a/target/riscv/cpu_cfg.h b/target/riscv/cpu_cfg.h > index 780ae6ef17..3735c69fd6 100644 > --- a/target/riscv/cpu_cfg.h > +++ b/target/riscv/cpu_cfg.h > @@ -136,6 +136,7 @@ struct RISCVCPUConfig { > bool ext_xtheadmemidx; > bool ext_xtheadmempair; > bool ext_xtheadsync; > + bool ext_xtheadmaee; > bool ext_XVentanaCondOps; > > uint32_t pmu_mask; > @@ -176,7 +177,8 @@ static inline bool has_xthead_p(const RISCVCPUConfig *cfg) > cfg->ext_xtheadcondmov || > cfg->ext_xtheadfmemidx || cfg->ext_xtheadfmv || > cfg->ext_xtheadmac || cfg->ext_xtheadmemidx || > - cfg->ext_xtheadmempair || cfg->ext_xtheadsync; > + cfg->ext_xtheadmempair || cfg->ext_xtheadsync || > + cfg->ext_xtheadmaee; > } > > #define MATERIALISE_EXT_PREDICATE(ext) \ > diff --git a/target/riscv/cpu_helper.c b/target/riscv/cpu_helper.c > index c7cc7eb423..5c1f380276 100644 > --- a/target/riscv/cpu_helper.c > +++ b/target/riscv/cpu_helper.c > @@ -812,6 +812,7 @@ static int get_physical_address(CPURISCVState *env, hwaddr *physical, > int napot_bits = 0; > target_ulong napot_mask; > > + bool skip_pte_check = env->th_mxstatus & TH_MXSTATUS_MAEE; > /* > * Check if we should use the background registers for the two > * stage translation. We don't need to check if we actually need > @@ -974,18 +975,19 @@ restart: > if (riscv_cpu_sxl(env) == MXL_RV32) { > ppn = pte >> PTE_PPN_SHIFT; > } else { > - if (pte & PTE_RESERVED) { > - return TRANSLATE_FAIL; > - } > + if (!skip_pte_check) { > + if (pte & PTE_RESERVED) { > + return TRANSLATE_FAIL; > + } > > - if (!pbmte && (pte & PTE_PBMT)) { > - return TRANSLATE_FAIL; > - } > + if (!pbmte && (pte & PTE_PBMT)) { > + return TRANSLATE_FAIL; > + } > > - if (!riscv_cpu_cfg(env)->ext_svnapot && (pte & PTE_N)) { > - return TRANSLATE_FAIL; > + if (!riscv_cpu_cfg(env)->ext_svnapot && (pte & PTE_N)) { > + return TRANSLATE_FAIL; > + } > } > - > ppn = (pte & (target_ulong)PTE_PPN_MASK) >> PTE_PPN_SHIFT; Unfortunately we won't be able to take this upstream. This is core QEMU RISC-V code that is now being changed against the spec. I think adding the CSR is fine, but we can't take this core change. A fix that works for everyone should be supporting the th_mxstatus CSR, but don't support setting the TH_MXSTATUS_MAEE bit. That way guests can detect that the bit isn't set and not use the reserved bits in the PTE. From my understanding the extra PTE bits are related to cache control in the hardware, which we don't need here Alistair > } > > @@ -998,7 +1000,8 @@ restart: > } > > /* Inner PTE, continue walking */ > - if (pte & (PTE_D | PTE_A | PTE_U | PTE_ATTR)) { > + if ((pte & (PTE_D | PTE_A | PTE_U)) || > + (!skip_pte_check && (pte & PTE_ATTR))) { > return TRANSLATE_FAIL; > } > base = ppn << PGSHIFT; > @@ -1012,7 +1015,7 @@ restart: > /* Misaligned PPN */ > return TRANSLATE_FAIL; > } > - if (!pbmte && (pte & PTE_PBMT)) { > + if (!skip_pte_check && !pbmte && (pte & PTE_PBMT)) { > /* Reserved without Svpbmt. */ > return TRANSLATE_FAIL; > } > diff --git a/target/riscv/meson.build b/target/riscv/meson.build > index a5e0734e7f..d7f675881d 100644 > --- a/target/riscv/meson.build > +++ b/target/riscv/meson.build > @@ -12,6 +12,7 @@ riscv_ss.add(files( > 'cpu.c', > 'cpu_helper.c', > 'csr.c', > + 'xthead_csr.c', > 'fpu_helper.c', > 'gdbstub.c', > 'op_helper.c', > diff --git a/target/riscv/tcg/tcg-cpu.c b/target/riscv/tcg/tcg-cpu.c > index 559bf373f3..4b1184c8ab 100644 > --- a/target/riscv/tcg/tcg-cpu.c > +++ b/target/riscv/tcg/tcg-cpu.c > @@ -763,7 +763,6 @@ void riscv_cpu_validate_set_extensions(RISCVCPU *cpu, Error **errp) > cpu->cfg.pmu_mask = 0; > cpu->pmu_avail_ctrs = 0; > } > - > /* > * Disable isa extensions based on priv spec after we > * validated and set everything we need. > @@ -871,12 +870,18 @@ static void riscv_cpu_validate_profiles(RISCVCPU *cpu) > } > } > > +static inline bool th_csr_p(const RISCVCPUConfig *cfg) > +{ > + return cfg->ext_xtheadmaee; > +} > + > void riscv_tcg_cpu_register_vendor_csr(RISCVCPU *cpu) > { > static const struct { > bool (*guard_func)(const RISCVCPUConfig *); > riscv_csr_operations *csr_ops; > } vendors[] = { > + { th_csr_p, th_csr_ops }, > }; > for (int i = 0; i < ARRAY_SIZE(vendors); ++i) { > if (!vendors[i].guard_func(&cpu->cfg)) { > diff --git a/target/riscv/xthead_csr.c b/target/riscv/xthead_csr.c > new file mode 100644 > index 0000000000..9f88fa50db > --- /dev/null > +++ b/target/riscv/xthead_csr.c > @@ -0,0 +1,65 @@ > +/* > + * Xuantie implementation for RISC-V Control and Status Registers. > + * > + * Copyright (c) 2024 Alibaba Group. All rights reserved. > + * > + * This program is free software; you can redistribute it and/or modify it > + * under the terms and conditions of the GNU General Public License, > + * version 2 or later, as published by the Free Software Foundation. > + * > + * This program is distributed in the hope it will be useful, but WITHOUT > + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or > + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for > + * more details. > + * > + * You should have received a copy of the GNU General Public License along with > + * this program. If not, see <http://www.gnu.org/licenses/>. > + */ > + > + > +#include "qemu/osdep.h" > +#include "qemu/log.h" > +#include "cpu.h" > +#include "tcg/tcg-cpu.h" > +#include "exec/exec-all.h" > +#include "exec/tb-flush.h" > +#include "qapi/error.h" > + > +#if !defined(CONFIG_USER_ONLY) > +static RISCVException th_maee_check(CPURISCVState *env, int csrno) > +{ > + if (riscv_cpu_cfg(env)->ext_xtheadmaee) { > + return RISCV_EXCP_ILLEGAL_INST; > + } > + return RISCV_EXCP_NONE; > +} > + > +static RISCVException > +read_th_mxstatus(CPURISCVState *env, int csrno, target_ulong *val) > +{ > + *val = env->th_mxstatus; > + return RISCV_EXCP_NONE; > +} > + > +static RISCVException > +write_th_mxstatus(CPURISCVState *env, int csrno, target_ulong val) > +{ > + uint64_t mxstatus = env->th_mxstatus; > + uint64_t mask = TH_MXSTATUS_MAEE; > + > + if ((val ^ mxstatus) & TH_MXSTATUS_MAEE) { > + tlb_flush(env_cpu(env)); > + } > + > + mxstatus = (mxstatus & ~mask) | (val & mask); > + env->th_mxstatus = mxstatus; > + return RISCV_EXCP_NONE; > +} > +#endif > + > +riscv_csr_operations th_csr_ops[CSR_TABLE_SIZE] = { > +#if !defined(CONFIG_USER_ONLY) > + [CSR_TH_MXSTATUS] = { "th_mxstatus", th_maee_check, read_th_mxstatus, > + write_th_mxstatus}, > +#endif /* !CONFIG_USER_ONLY */ > +}; > -- > 2.25.1 > >
On Mon, Feb 5, 2024 at 3:42 AM Alistair Francis <alistair23@gmail.com> wrote: > > On Sun, Feb 4, 2024 at 3:44 PM LIU Zhiwei <zhiwei_liu@linux.alibaba.com> wrote: > > > > This patch set fix the regression on kernel pointed by Björn Töpel in > > https://www.mail-archive.com/qemu-devel@nongnu.org/msg1018232.html. > > > > thead-c906 uses some flags in pte [60-63] bits. It has history reasons that > > SVPBMT didn't exist when thead-c906 came to wotrld. We named this feature as > > xtheadmaee[1]. this feature is controlled by an custom CSR named mxstatus, > > whose maee field encodes whether enable the pte [60-63] bits. > > > > [1]:https://github.com/T-head-Semi/thead-extension-spec/blob/master/xtheadmaee.adoc > > > > Signed-off-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com> > > --- > > v1->v2: > > 1) Remove mxstatus user mode access > > 2) Add reference documentation to the commit log > > --- > > target/riscv/cpu.c | 6 ++++ > > target/riscv/cpu.h | 9 ++++++ > > target/riscv/cpu_bits.h | 6 ++++ > > target/riscv/cpu_cfg.h | 4 ++- > > target/riscv/cpu_helper.c | 25 ++++++++------- > > target/riscv/meson.build | 1 + > > target/riscv/tcg/tcg-cpu.c | 7 +++- > > target/riscv/xthead_csr.c | 65 ++++++++++++++++++++++++++++++++++++++ > > 8 files changed, 110 insertions(+), 13 deletions(-) > > create mode 100644 target/riscv/xthead_csr.c > > > > diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c > > index 2dcbc9ff32..bfdbb0539a 100644 > > --- a/target/riscv/cpu.c > > +++ b/target/riscv/cpu.c > > @@ -171,6 +171,7 @@ const RISCVIsaExtData isa_edata_arr[] = { > > ISA_EXT_DATA_ENTRY(xtheadmemidx, PRIV_VERSION_1_11_0, ext_xtheadmemidx), > > ISA_EXT_DATA_ENTRY(xtheadmempair, PRIV_VERSION_1_11_0, ext_xtheadmempair), > > ISA_EXT_DATA_ENTRY(xtheadsync, PRIV_VERSION_1_11_0, ext_xtheadsync), > > + ISA_EXT_DATA_ENTRY(xtheadmaee, PRIV_VERSION_1_11_0, ext_xtheadmaee), > > ISA_EXT_DATA_ENTRY(xventanacondops, PRIV_VERSION_1_12_0, ext_XVentanaCondOps), > > > > DEFINE_PROP_END_OF_LIST(), > > @@ -506,6 +507,7 @@ static void rv64_thead_c906_cpu_init(Object *obj) > > > > cpu->cfg.mvendorid = THEAD_VENDOR_ID; > > #ifndef CONFIG_USER_ONLY > > + cpu->cfg.ext_xtheadmaee = true; > > set_satp_mode_max_supported(cpu, VM_1_10_SV39); > > #endif > > > > @@ -949,6 +951,9 @@ static void riscv_cpu_reset_hold(Object *obj) > > } > > > > pmp_unlock_entries(env); > > + if (riscv_cpu_cfg(env)->ext_xtheadmaee) { > > + env->th_mxstatus |= TH_MXSTATUS_MAEE; > > + } > > #endif > > env->xl = riscv_cpu_mxl(env); > > riscv_cpu_update_mask(env); > > @@ -1439,6 +1444,7 @@ const RISCVCPUMultiExtConfig riscv_cpu_vendor_exts[] = { > > MULTI_EXT_CFG_BOOL("xtheadmemidx", ext_xtheadmemidx, false), > > MULTI_EXT_CFG_BOOL("xtheadmempair", ext_xtheadmempair, false), > > MULTI_EXT_CFG_BOOL("xtheadsync", ext_xtheadsync, false), > > + MULTI_EXT_CFG_BOOL("xtheadmaee", ext_xtheadmaee, false), > > MULTI_EXT_CFG_BOOL("xventanacondops", ext_XVentanaCondOps, false), > > > > DEFINE_PROP_END_OF_LIST(), > > diff --git a/target/riscv/cpu.h b/target/riscv/cpu.h > > index 5f3955c38d..1bacf40355 100644 > > --- a/target/riscv/cpu.h > > +++ b/target/riscv/cpu.h > > @@ -412,6 +412,14 @@ struct CPUArchState { > > target_ulong cur_pmmask; > > target_ulong cur_pmbase; > > > > + union { > > + /* Custom CSR for Xuantie CPU */ > > + struct { > > +#ifndef CONFIG_USER_ONLY > > + target_ulong th_mxstatus; > > +#endif > > + }; > > + }; > > /* Fields from here on are preserved across CPU reset. */ > > QEMUTimer *stimer; /* Internal timer for S-mode interrupt */ > > QEMUTimer *vstimer; /* Internal timer for VS-mode interrupt */ > > @@ -799,6 +807,7 @@ void riscv_add_satp_mode_properties(Object *obj); > > bool riscv_cpu_accelerator_compatible(RISCVCPU *cpu); > > > > /* CSR function table */ > > +extern riscv_csr_operations th_csr_ops[CSR_TABLE_SIZE]; > > extern riscv_csr_operations csr_ops[CSR_TABLE_SIZE]; > > > > extern const bool valid_vm_1_10_32[], valid_vm_1_10_64[]; > > diff --git a/target/riscv/cpu_bits.h b/target/riscv/cpu_bits.h > > index e116f6c252..67ebb1cefe 100644 > > --- a/target/riscv/cpu_bits.h > > +++ b/target/riscv/cpu_bits.h > > @@ -897,4 +897,10 @@ typedef enum RISCVException { > > /* JVT CSR bits */ > > #define JVT_MODE 0x3F > > #define JVT_BASE (~0x3F) > > + > > +/* Xuantie custom CSRs */ > > +#define CSR_TH_MXSTATUS 0x7c0 > > + > > +#define TH_MXSTATUS_MAEE_SHIFT 21 > > +#define TH_MXSTATUS_MAEE (0x1 << TH_MXSTATUS_MAEE_SHIFT) > > #endif > > diff --git a/target/riscv/cpu_cfg.h b/target/riscv/cpu_cfg.h > > index 780ae6ef17..3735c69fd6 100644 > > --- a/target/riscv/cpu_cfg.h > > +++ b/target/riscv/cpu_cfg.h > > @@ -136,6 +136,7 @@ struct RISCVCPUConfig { > > bool ext_xtheadmemidx; > > bool ext_xtheadmempair; > > bool ext_xtheadsync; > > + bool ext_xtheadmaee; > > bool ext_XVentanaCondOps; > > > > uint32_t pmu_mask; > > @@ -176,7 +177,8 @@ static inline bool has_xthead_p(const RISCVCPUConfig *cfg) > > cfg->ext_xtheadcondmov || > > cfg->ext_xtheadfmemidx || cfg->ext_xtheadfmv || > > cfg->ext_xtheadmac || cfg->ext_xtheadmemidx || > > - cfg->ext_xtheadmempair || cfg->ext_xtheadsync; > > + cfg->ext_xtheadmempair || cfg->ext_xtheadsync || > > + cfg->ext_xtheadmaee; > > } > > > > #define MATERIALISE_EXT_PREDICATE(ext) \ > > diff --git a/target/riscv/cpu_helper.c b/target/riscv/cpu_helper.c > > index c7cc7eb423..5c1f380276 100644 > > --- a/target/riscv/cpu_helper.c > > +++ b/target/riscv/cpu_helper.c > > @@ -812,6 +812,7 @@ static int get_physical_address(CPURISCVState *env, hwaddr *physical, > > int napot_bits = 0; > > target_ulong napot_mask; > > > > + bool skip_pte_check = env->th_mxstatus & TH_MXSTATUS_MAEE; > > /* > > * Check if we should use the background registers for the two > > * stage translation. We don't need to check if we actually need > > @@ -974,18 +975,19 @@ restart: > > if (riscv_cpu_sxl(env) == MXL_RV32) { > > ppn = pte >> PTE_PPN_SHIFT; > > } else { > > - if (pte & PTE_RESERVED) { > > - return TRANSLATE_FAIL; > > - } > > + if (!skip_pte_check) { > > + if (pte & PTE_RESERVED) { > > + return TRANSLATE_FAIL; > > + } > > > > - if (!pbmte && (pte & PTE_PBMT)) { > > - return TRANSLATE_FAIL; > > - } > > + if (!pbmte && (pte & PTE_PBMT)) { > > + return TRANSLATE_FAIL; > > + } > > > > - if (!riscv_cpu_cfg(env)->ext_svnapot && (pte & PTE_N)) { > > - return TRANSLATE_FAIL; > > + if (!riscv_cpu_cfg(env)->ext_svnapot && (pte & PTE_N)) { > > + return TRANSLATE_FAIL; > > + } > > } > > - > > ppn = (pte & (target_ulong)PTE_PPN_MASK) >> PTE_PPN_SHIFT; > > Unfortunately we won't be able to take this upstream. This is core > QEMU RISC-V code that is now being changed against the spec. I think > adding the CSR is fine, but we can't take this core change. > > A fix that works for everyone should be supporting the th_mxstatus > CSR, but don't support setting the TH_MXSTATUS_MAEE bit. That way > guests can detect that the bit isn't set and not use the reserved bits > in the PTE. From my understanding the extra PTE bits are related to > cache control in the hardware, which we don't need here Sounds good! Let me recap the overall plan: * QEMU does not emulate MAEE, but signals that MAEE is not available by setting TH_MXSTATUS_MAEE to 0. * Consequence: The c906 emulation does not enable any page-base memory attribute mechanism. * OpenSBI tests the TH_MXSTATUS_MAEE bit (M-mode only) and provides that information to user-space (e.g. DTB). * The current Linux errata code will be enhanced to not assume MAEE for each core with T-Head vendor ID, but also query the MAEE bit and ensure it is set. * Booting on a QEMU c906 will not enable MAEE, but will still boot. We never had a complete MAEE implementation upstream, because that's not needed. But until recently we could mess with reserved bits how we want. Now that QEMU is more restrictive about reserved bits in the PTEs, we need to address this in the kernel. The downside is, that the kernel now sees a c906 CPU without MAEE and such a CPU does not exist. But that's fine, because this does not require extra code to handle this case. Also, the additional check for the MAEE bit in the kernel errata probing code is likely needed anyway for future T-Head CPUs. BTW, what was the reason for QEMU to prevent setting reserved bits in PTEs? Thanks, Christoph > > Alistair > > > } > > > > @@ -998,7 +1000,8 @@ restart: > > } > > > > /* Inner PTE, continue walking */ > > - if (pte & (PTE_D | PTE_A | PTE_U | PTE_ATTR)) { > > + if ((pte & (PTE_D | PTE_A | PTE_U)) || > > + (!skip_pte_check && (pte & PTE_ATTR))) { > > return TRANSLATE_FAIL; > > } > > base = ppn << PGSHIFT; > > @@ -1012,7 +1015,7 @@ restart: > > /* Misaligned PPN */ > > return TRANSLATE_FAIL; > > } > > - if (!pbmte && (pte & PTE_PBMT)) { > > + if (!skip_pte_check && !pbmte && (pte & PTE_PBMT)) { > > /* Reserved without Svpbmt. */ > > return TRANSLATE_FAIL; > > } > > diff --git a/target/riscv/meson.build b/target/riscv/meson.build > > index a5e0734e7f..d7f675881d 100644 > > --- a/target/riscv/meson.build > > +++ b/target/riscv/meson.build > > @@ -12,6 +12,7 @@ riscv_ss.add(files( > > 'cpu.c', > > 'cpu_helper.c', > > 'csr.c', > > + 'xthead_csr.c', > > 'fpu_helper.c', > > 'gdbstub.c', > > 'op_helper.c', > > diff --git a/target/riscv/tcg/tcg-cpu.c b/target/riscv/tcg/tcg-cpu.c > > index 559bf373f3..4b1184c8ab 100644 > > --- a/target/riscv/tcg/tcg-cpu.c > > +++ b/target/riscv/tcg/tcg-cpu.c > > @@ -763,7 +763,6 @@ void riscv_cpu_validate_set_extensions(RISCVCPU *cpu, Error **errp) > > cpu->cfg.pmu_mask = 0; > > cpu->pmu_avail_ctrs = 0; > > } > > - > > /* > > * Disable isa extensions based on priv spec after we > > * validated and set everything we need. > > @@ -871,12 +870,18 @@ static void riscv_cpu_validate_profiles(RISCVCPU *cpu) > > } > > } > > > > +static inline bool th_csr_p(const RISCVCPUConfig *cfg) > > +{ > > + return cfg->ext_xtheadmaee; > > +} > > + > > void riscv_tcg_cpu_register_vendor_csr(RISCVCPU *cpu) > > { > > static const struct { > > bool (*guard_func)(const RISCVCPUConfig *); > > riscv_csr_operations *csr_ops; > > } vendors[] = { > > + { th_csr_p, th_csr_ops }, > > }; > > for (int i = 0; i < ARRAY_SIZE(vendors); ++i) { > > if (!vendors[i].guard_func(&cpu->cfg)) { > > diff --git a/target/riscv/xthead_csr.c b/target/riscv/xthead_csr.c > > new file mode 100644 > > index 0000000000..9f88fa50db > > --- /dev/null > > +++ b/target/riscv/xthead_csr.c > > @@ -0,0 +1,65 @@ > > +/* > > + * Xuantie implementation for RISC-V Control and Status Registers. > > + * > > + * Copyright (c) 2024 Alibaba Group. All rights reserved. > > + * > > + * This program is free software; you can redistribute it and/or modify it > > + * under the terms and conditions of the GNU General Public License, > > + * version 2 or later, as published by the Free Software Foundation. > > + * > > + * This program is distributed in the hope it will be useful, but WITHOUT > > + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or > > + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for > > + * more details. > > + * > > + * You should have received a copy of the GNU General Public License along with > > + * this program. If not, see <http://www.gnu.org/licenses/>. > > + */ > > + > > + > > +#include "qemu/osdep.h" > > +#include "qemu/log.h" > > +#include "cpu.h" > > +#include "tcg/tcg-cpu.h" > > +#include "exec/exec-all.h" > > +#include "exec/tb-flush.h" > > +#include "qapi/error.h" > > + > > +#if !defined(CONFIG_USER_ONLY) > > +static RISCVException th_maee_check(CPURISCVState *env, int csrno) > > +{ > > + if (riscv_cpu_cfg(env)->ext_xtheadmaee) { > > + return RISCV_EXCP_ILLEGAL_INST; > > + } > > + return RISCV_EXCP_NONE; > > +} > > + > > +static RISCVException > > +read_th_mxstatus(CPURISCVState *env, int csrno, target_ulong *val) > > +{ > > + *val = env->th_mxstatus; > > + return RISCV_EXCP_NONE; > > +} > > + > > +static RISCVException > > +write_th_mxstatus(CPURISCVState *env, int csrno, target_ulong val) > > +{ > > + uint64_t mxstatus = env->th_mxstatus; > > + uint64_t mask = TH_MXSTATUS_MAEE; > > + > > + if ((val ^ mxstatus) & TH_MXSTATUS_MAEE) { > > + tlb_flush(env_cpu(env)); > > + } > > + > > + mxstatus = (mxstatus & ~mask) | (val & mask); > > + env->th_mxstatus = mxstatus; > > + return RISCV_EXCP_NONE; > > +} > > +#endif > > + > > +riscv_csr_operations th_csr_ops[CSR_TABLE_SIZE] = { > > +#if !defined(CONFIG_USER_ONLY) > > + [CSR_TH_MXSTATUS] = { "th_mxstatus", th_maee_check, read_th_mxstatus, > > + write_th_mxstatus}, > > +#endif /* !CONFIG_USER_ONLY */ > > +}; > > -- > > 2.25.1 > > > >
On Mon, Feb 5, 2024 at 6:37 PM Christoph Müllner <christoph.muellner@vrull.eu> wrote: > > On Mon, Feb 5, 2024 at 3:42 AM Alistair Francis <alistair23@gmail.com> wrote: > > > > On Sun, Feb 4, 2024 at 3:44 PM LIU Zhiwei <zhiwei_liu@linux.alibaba.com> wrote: > > > > > > This patch set fix the regression on kernel pointed by Björn Töpel in > > > https://www.mail-archive.com/qemu-devel@nongnu.org/msg1018232.html. > > > > > > thead-c906 uses some flags in pte [60-63] bits. It has history reasons that > > > SVPBMT didn't exist when thead-c906 came to wotrld. We named this feature as > > > xtheadmaee[1]. this feature is controlled by an custom CSR named mxstatus, > > > whose maee field encodes whether enable the pte [60-63] bits. > > > > > > [1]:https://github.com/T-head-Semi/thead-extension-spec/blob/master/xtheadmaee.adoc > > > > > > Signed-off-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com> > > > --- > > > v1->v2: > > > 1) Remove mxstatus user mode access > > > 2) Add reference documentation to the commit log > > > --- > > > target/riscv/cpu.c | 6 ++++ > > > target/riscv/cpu.h | 9 ++++++ > > > target/riscv/cpu_bits.h | 6 ++++ > > > target/riscv/cpu_cfg.h | 4 ++- > > > target/riscv/cpu_helper.c | 25 ++++++++------- > > > target/riscv/meson.build | 1 + > > > target/riscv/tcg/tcg-cpu.c | 7 +++- > > > target/riscv/xthead_csr.c | 65 ++++++++++++++++++++++++++++++++++++++ > > > 8 files changed, 110 insertions(+), 13 deletions(-) > > > create mode 100644 target/riscv/xthead_csr.c > > > > > > diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c > > > index 2dcbc9ff32..bfdbb0539a 100644 > > > --- a/target/riscv/cpu.c > > > +++ b/target/riscv/cpu.c > > > @@ -171,6 +171,7 @@ const RISCVIsaExtData isa_edata_arr[] = { > > > ISA_EXT_DATA_ENTRY(xtheadmemidx, PRIV_VERSION_1_11_0, ext_xtheadmemidx), > > > ISA_EXT_DATA_ENTRY(xtheadmempair, PRIV_VERSION_1_11_0, ext_xtheadmempair), > > > ISA_EXT_DATA_ENTRY(xtheadsync, PRIV_VERSION_1_11_0, ext_xtheadsync), > > > + ISA_EXT_DATA_ENTRY(xtheadmaee, PRIV_VERSION_1_11_0, ext_xtheadmaee), > > > ISA_EXT_DATA_ENTRY(xventanacondops, PRIV_VERSION_1_12_0, ext_XVentanaCondOps), > > > > > > DEFINE_PROP_END_OF_LIST(), > > > @@ -506,6 +507,7 @@ static void rv64_thead_c906_cpu_init(Object *obj) > > > > > > cpu->cfg.mvendorid = THEAD_VENDOR_ID; > > > #ifndef CONFIG_USER_ONLY > > > + cpu->cfg.ext_xtheadmaee = true; > > > set_satp_mode_max_supported(cpu, VM_1_10_SV39); > > > #endif > > > > > > @@ -949,6 +951,9 @@ static void riscv_cpu_reset_hold(Object *obj) > > > } > > > > > > pmp_unlock_entries(env); > > > + if (riscv_cpu_cfg(env)->ext_xtheadmaee) { > > > + env->th_mxstatus |= TH_MXSTATUS_MAEE; > > > + } > > > #endif > > > env->xl = riscv_cpu_mxl(env); > > > riscv_cpu_update_mask(env); > > > @@ -1439,6 +1444,7 @@ const RISCVCPUMultiExtConfig riscv_cpu_vendor_exts[] = { > > > MULTI_EXT_CFG_BOOL("xtheadmemidx", ext_xtheadmemidx, false), > > > MULTI_EXT_CFG_BOOL("xtheadmempair", ext_xtheadmempair, false), > > > MULTI_EXT_CFG_BOOL("xtheadsync", ext_xtheadsync, false), > > > + MULTI_EXT_CFG_BOOL("xtheadmaee", ext_xtheadmaee, false), > > > MULTI_EXT_CFG_BOOL("xventanacondops", ext_XVentanaCondOps, false), > > > > > > DEFINE_PROP_END_OF_LIST(), > > > diff --git a/target/riscv/cpu.h b/target/riscv/cpu.h > > > index 5f3955c38d..1bacf40355 100644 > > > --- a/target/riscv/cpu.h > > > +++ b/target/riscv/cpu.h > > > @@ -412,6 +412,14 @@ struct CPUArchState { > > > target_ulong cur_pmmask; > > > target_ulong cur_pmbase; > > > > > > + union { > > > + /* Custom CSR for Xuantie CPU */ > > > + struct { > > > +#ifndef CONFIG_USER_ONLY > > > + target_ulong th_mxstatus; > > > +#endif > > > + }; > > > + }; > > > /* Fields from here on are preserved across CPU reset. */ > > > QEMUTimer *stimer; /* Internal timer for S-mode interrupt */ > > > QEMUTimer *vstimer; /* Internal timer for VS-mode interrupt */ > > > @@ -799,6 +807,7 @@ void riscv_add_satp_mode_properties(Object *obj); > > > bool riscv_cpu_accelerator_compatible(RISCVCPU *cpu); > > > > > > /* CSR function table */ > > > +extern riscv_csr_operations th_csr_ops[CSR_TABLE_SIZE]; > > > extern riscv_csr_operations csr_ops[CSR_TABLE_SIZE]; > > > > > > extern const bool valid_vm_1_10_32[], valid_vm_1_10_64[]; > > > diff --git a/target/riscv/cpu_bits.h b/target/riscv/cpu_bits.h > > > index e116f6c252..67ebb1cefe 100644 > > > --- a/target/riscv/cpu_bits.h > > > +++ b/target/riscv/cpu_bits.h > > > @@ -897,4 +897,10 @@ typedef enum RISCVException { > > > /* JVT CSR bits */ > > > #define JVT_MODE 0x3F > > > #define JVT_BASE (~0x3F) > > > + > > > +/* Xuantie custom CSRs */ > > > +#define CSR_TH_MXSTATUS 0x7c0 > > > + > > > +#define TH_MXSTATUS_MAEE_SHIFT 21 > > > +#define TH_MXSTATUS_MAEE (0x1 << TH_MXSTATUS_MAEE_SHIFT) > > > #endif > > > diff --git a/target/riscv/cpu_cfg.h b/target/riscv/cpu_cfg.h > > > index 780ae6ef17..3735c69fd6 100644 > > > --- a/target/riscv/cpu_cfg.h > > > +++ b/target/riscv/cpu_cfg.h > > > @@ -136,6 +136,7 @@ struct RISCVCPUConfig { > > > bool ext_xtheadmemidx; > > > bool ext_xtheadmempair; > > > bool ext_xtheadsync; > > > + bool ext_xtheadmaee; > > > bool ext_XVentanaCondOps; > > > > > > uint32_t pmu_mask; > > > @@ -176,7 +177,8 @@ static inline bool has_xthead_p(const RISCVCPUConfig *cfg) > > > cfg->ext_xtheadcondmov || > > > cfg->ext_xtheadfmemidx || cfg->ext_xtheadfmv || > > > cfg->ext_xtheadmac || cfg->ext_xtheadmemidx || > > > - cfg->ext_xtheadmempair || cfg->ext_xtheadsync; > > > + cfg->ext_xtheadmempair || cfg->ext_xtheadsync || > > > + cfg->ext_xtheadmaee; > > > } > > > > > > #define MATERIALISE_EXT_PREDICATE(ext) \ > > > diff --git a/target/riscv/cpu_helper.c b/target/riscv/cpu_helper.c > > > index c7cc7eb423..5c1f380276 100644 > > > --- a/target/riscv/cpu_helper.c > > > +++ b/target/riscv/cpu_helper.c > > > @@ -812,6 +812,7 @@ static int get_physical_address(CPURISCVState *env, hwaddr *physical, > > > int napot_bits = 0; > > > target_ulong napot_mask; > > > > > > + bool skip_pte_check = env->th_mxstatus & TH_MXSTATUS_MAEE; > > > /* > > > * Check if we should use the background registers for the two > > > * stage translation. We don't need to check if we actually need > > > @@ -974,18 +975,19 @@ restart: > > > if (riscv_cpu_sxl(env) == MXL_RV32) { > > > ppn = pte >> PTE_PPN_SHIFT; > > > } else { > > > - if (pte & PTE_RESERVED) { > > > - return TRANSLATE_FAIL; > > > - } > > > + if (!skip_pte_check) { > > > + if (pte & PTE_RESERVED) { > > > + return TRANSLATE_FAIL; > > > + } > > > > > > - if (!pbmte && (pte & PTE_PBMT)) { > > > - return TRANSLATE_FAIL; > > > - } > > > + if (!pbmte && (pte & PTE_PBMT)) { > > > + return TRANSLATE_FAIL; > > > + } > > > > > > - if (!riscv_cpu_cfg(env)->ext_svnapot && (pte & PTE_N)) { > > > - return TRANSLATE_FAIL; > > > + if (!riscv_cpu_cfg(env)->ext_svnapot && (pte & PTE_N)) { > > > + return TRANSLATE_FAIL; > > > + } > > > } > > > - > > > ppn = (pte & (target_ulong)PTE_PPN_MASK) >> PTE_PPN_SHIFT; > > > > Unfortunately we won't be able to take this upstream. This is core > > QEMU RISC-V code that is now being changed against the spec. I think > > adding the CSR is fine, but we can't take this core change. > > > > A fix that works for everyone should be supporting the th_mxstatus > > CSR, but don't support setting the TH_MXSTATUS_MAEE bit. That way > > guests can detect that the bit isn't set and not use the reserved bits > > in the PTE. From my understanding the extra PTE bits are related to > > cache control in the hardware, which we don't need here > > Sounds good! Let me recap the overall plan: > * QEMU does not emulate MAEE, but signals that MAEE is not available > by setting TH_MXSTATUS_MAEE to 0. Yep! > * Consequence: The c906 emulation does not enable any page-base memory > attribute mechanism. Exactly > * OpenSBI tests the TH_MXSTATUS_MAEE bit (M-mode only) and provides > that information to user-space (e.g. DTB). To the kernel, but yep! > * The current Linux errata code will be enhanced to not assume MAEE > for each core with T-Head vendor ID, but also query the MAEE bit and > ensure it is set. I feel like it should already do that :) > * Booting on a QEMU c906 will not enable MAEE, but will still boot. That's the hope > > We never had a complete MAEE implementation upstream, because that's not needed. > But until recently we could mess with reserved bits how we want. > Now that QEMU is more restrictive about reserved bits in the PTEs, we > need to address this in the kernel. > > The downside is, that the kernel now sees a c906 CPU without MAEE and > such a CPU does not exist. Yeah, that is the downside. But in reality a CPU could exist that doesn't allow seeing MAEE, so I don't think it's that insane. > But that's fine, because this does not require extra code to handle this case. > Also, the additional check for the MAEE bit in the kernel errata > probing code is likely needed anyway for future T-Head CPUs. Exactly > > BTW, what was the reason for QEMU to prevent setting reserved bits in PTEs? I don't have it in front of me, but I thought that is what the spec said should happen Alistair > > Thanks, > Christoph > > > > > Alistair > > > > > } > > > > > > @@ -998,7 +1000,8 @@ restart: > > > } > > > > > > /* Inner PTE, continue walking */ > > > - if (pte & (PTE_D | PTE_A | PTE_U | PTE_ATTR)) { > > > + if ((pte & (PTE_D | PTE_A | PTE_U)) || > > > + (!skip_pte_check && (pte & PTE_ATTR))) { > > > return TRANSLATE_FAIL; > > > } > > > base = ppn << PGSHIFT; > > > @@ -1012,7 +1015,7 @@ restart: > > > /* Misaligned PPN */ > > > return TRANSLATE_FAIL; > > > } > > > - if (!pbmte && (pte & PTE_PBMT)) { > > > + if (!skip_pte_check && !pbmte && (pte & PTE_PBMT)) { > > > /* Reserved without Svpbmt. */ > > > return TRANSLATE_FAIL; > > > } > > > diff --git a/target/riscv/meson.build b/target/riscv/meson.build > > > index a5e0734e7f..d7f675881d 100644 > > > --- a/target/riscv/meson.build > > > +++ b/target/riscv/meson.build > > > @@ -12,6 +12,7 @@ riscv_ss.add(files( > > > 'cpu.c', > > > 'cpu_helper.c', > > > 'csr.c', > > > + 'xthead_csr.c', > > > 'fpu_helper.c', > > > 'gdbstub.c', > > > 'op_helper.c', > > > diff --git a/target/riscv/tcg/tcg-cpu.c b/target/riscv/tcg/tcg-cpu.c > > > index 559bf373f3..4b1184c8ab 100644 > > > --- a/target/riscv/tcg/tcg-cpu.c > > > +++ b/target/riscv/tcg/tcg-cpu.c > > > @@ -763,7 +763,6 @@ void riscv_cpu_validate_set_extensions(RISCVCPU *cpu, Error **errp) > > > cpu->cfg.pmu_mask = 0; > > > cpu->pmu_avail_ctrs = 0; > > > } > > > - > > > /* > > > * Disable isa extensions based on priv spec after we > > > * validated and set everything we need. > > > @@ -871,12 +870,18 @@ static void riscv_cpu_validate_profiles(RISCVCPU *cpu) > > > } > > > } > > > > > > +static inline bool th_csr_p(const RISCVCPUConfig *cfg) > > > +{ > > > + return cfg->ext_xtheadmaee; > > > +} > > > + > > > void riscv_tcg_cpu_register_vendor_csr(RISCVCPU *cpu) > > > { > > > static const struct { > > > bool (*guard_func)(const RISCVCPUConfig *); > > > riscv_csr_operations *csr_ops; > > > } vendors[] = { > > > + { th_csr_p, th_csr_ops }, > > > }; > > > for (int i = 0; i < ARRAY_SIZE(vendors); ++i) { > > > if (!vendors[i].guard_func(&cpu->cfg)) { > > > diff --git a/target/riscv/xthead_csr.c b/target/riscv/xthead_csr.c > > > new file mode 100644 > > > index 0000000000..9f88fa50db > > > --- /dev/null > > > +++ b/target/riscv/xthead_csr.c > > > @@ -0,0 +1,65 @@ > > > +/* > > > + * Xuantie implementation for RISC-V Control and Status Registers. > > > + * > > > + * Copyright (c) 2024 Alibaba Group. All rights reserved. > > > + * > > > + * This program is free software; you can redistribute it and/or modify it > > > + * under the terms and conditions of the GNU General Public License, > > > + * version 2 or later, as published by the Free Software Foundation. > > > + * > > > + * This program is distributed in the hope it will be useful, but WITHOUT > > > + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or > > > + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for > > > + * more details. > > > + * > > > + * You should have received a copy of the GNU General Public License along with > > > + * this program. If not, see <http://www.gnu.org/licenses/>. > > > + */ > > > + > > > + > > > +#include "qemu/osdep.h" > > > +#include "qemu/log.h" > > > +#include "cpu.h" > > > +#include "tcg/tcg-cpu.h" > > > +#include "exec/exec-all.h" > > > +#include "exec/tb-flush.h" > > > +#include "qapi/error.h" > > > + > > > +#if !defined(CONFIG_USER_ONLY) > > > +static RISCVException th_maee_check(CPURISCVState *env, int csrno) > > > +{ > > > + if (riscv_cpu_cfg(env)->ext_xtheadmaee) { > > > + return RISCV_EXCP_ILLEGAL_INST; > > > + } > > > + return RISCV_EXCP_NONE; > > > +} > > > + > > > +static RISCVException > > > +read_th_mxstatus(CPURISCVState *env, int csrno, target_ulong *val) > > > +{ > > > + *val = env->th_mxstatus; > > > + return RISCV_EXCP_NONE; > > > +} > > > + > > > +static RISCVException > > > +write_th_mxstatus(CPURISCVState *env, int csrno, target_ulong val) > > > +{ > > > + uint64_t mxstatus = env->th_mxstatus; > > > + uint64_t mask = TH_MXSTATUS_MAEE; > > > + > > > + if ((val ^ mxstatus) & TH_MXSTATUS_MAEE) { > > > + tlb_flush(env_cpu(env)); > > > + } > > > + > > > + mxstatus = (mxstatus & ~mask) | (val & mask); > > > + env->th_mxstatus = mxstatus; > > > + return RISCV_EXCP_NONE; > > > +} > > > +#endif > > > + > > > +riscv_csr_operations th_csr_ops[CSR_TABLE_SIZE] = { > > > +#if !defined(CONFIG_USER_ONLY) > > > + [CSR_TH_MXSTATUS] = { "th_mxstatus", th_maee_check, read_th_mxstatus, > > > + write_th_mxstatus}, > > > +#endif /* !CONFIG_USER_ONLY */ > > > +}; > > > -- > > > 2.25.1 > > > > > >
Christoph linked here on his submission to Linux of a fix for this, so I am reviving this to leave a couple comments :) On Thu, Feb 15, 2024 at 02:24:02PM +1000, Alistair Francis wrote: > On Mon, Feb 5, 2024 at 6:37 PM Christoph Müllner > <christoph.muellner@vrull.eu> wrote: > > On Mon, Feb 5, 2024 at 3:42 AM Alistair Francis <alistair23@gmail.com> wrote: > > > On Sun, Feb 4, 2024 at 3:44 PM LIU Zhiwei <zhiwei_liu@linux.alibaba.com> wrote: > > > > ppn = (pte & (target_ulong)PTE_PPN_MASK) >> PTE_PPN_SHIFT; > > > > > > Unfortunately we won't be able to take this upstream. This is core > > > QEMU RISC-V code that is now being changed against the spec. I think > > > adding the CSR is fine, but we can't take this core change. > > > > > > A fix that works for everyone should be supporting the th_mxstatus > > > CSR, but don't support setting the TH_MXSTATUS_MAEE bit. That way > > > guests can detect that the bit isn't set and not use the reserved bits > > > in the PTE. From my understanding the extra PTE bits are related to > > > cache control in the hardware, which we don't need here > > > > Sounds good! Let me recap the overall plan: > > * QEMU does not emulate MAEE, but signals that MAEE is not available > > by setting TH_MXSTATUS_MAEE to 0. > > Yep! > > > * Consequence: The c906 emulation does not enable any page-base memory > > attribute mechanism. > > Exactly > > > * OpenSBI tests the TH_MXSTATUS_MAEE bit (M-mode only) and provides > > that information to user-space (e.g. DTB). > > To the kernel, but yep! > > > * The current Linux errata code will be enhanced to not assume MAEE > > for each core with T-Head vendor ID, but also query the MAEE bit and > > ensure it is set. > > I feel like it should already do that :) It doesn't quite do this right now. It only makes the assumption for CPUs where marchid and mvendorid are zero. The c908, and I think Guo Ren confirmed it will be the case going forward, sets these to non-zero values. We should have always required a dt property be set, rather than using m*id, but we can't go back on that for these devices. Going forward, if there are more CPUs that want to use this e.g. C908 in MAEE mode (it can do svpbmt too) I'm gonna require it is explicitly set in DT. > > * Booting on a QEMU c906 will not enable MAEE, but will still boot. > > That's the hope > > > > > We never had a complete MAEE implementation upstream, because that's not needed. > > But until recently we could mess with reserved bits how we want. > > Now that QEMU is more restrictive about reserved bits in the PTEs, we > > need to address this in the kernel. > > > > The downside is, that the kernel now sees a c906 CPU without MAEE and > > such a CPU does not exist. > > Yeah, that is the downside. But in reality a CPU could exist that > doesn't allow seeing MAEE, so I don't think it's that insane. Aye, and the case of a new CPU disabling it but not setting m*id will be a pain. > > But that's fine, because this does not require extra code to handle this case. > > Also, the additional check for the MAEE bit in the kernel errata > > probing code is likely needed anyway for future T-Head CPUs. > > Exactly I don't really want to have extension detection side channels like this in Linux if it can be at all avoided, I'd rather get this information from DT or ACPI. The marchid == mvendorid == 0x0 case has a pretty high chance of needing some special handling in the future though, so something targeting those cases when there's some demonstrable problem seems fair to me. Cheers, Conor.
On Wed, Mar 27, 2024 at 9:19 PM Conor Dooley <conor@kernel.org> wrote: > > Christoph linked here on his submission to Linux of a fix for this, so I > am reviving this to leave a couple comments :) > > On Thu, Feb 15, 2024 at 02:24:02PM +1000, Alistair Francis wrote: > > On Mon, Feb 5, 2024 at 6:37 PM Christoph Müllner > > <christoph.muellner@vrull.eu> wrote: > > > On Mon, Feb 5, 2024 at 3:42 AM Alistair Francis <alistair23@gmail.com> wrote: > > > > On Sun, Feb 4, 2024 at 3:44 PM LIU Zhiwei <zhiwei_liu@linux.alibaba.com> wrote: > > > > > > ppn = (pte & (target_ulong)PTE_PPN_MASK) >> PTE_PPN_SHIFT; > > > > > > > > Unfortunately we won't be able to take this upstream. This is core > > > > QEMU RISC-V code that is now being changed against the spec. I think > > > > adding the CSR is fine, but we can't take this core change. > > > > > > > > A fix that works for everyone should be supporting the th_mxstatus > > > > CSR, but don't support setting the TH_MXSTATUS_MAEE bit. That way > > > > guests can detect that the bit isn't set and not use the reserved bits > > > > in the PTE. From my understanding the extra PTE bits are related to > > > > cache control in the hardware, which we don't need here > > > > > > Sounds good! Let me recap the overall plan: > > > * QEMU does not emulate MAEE, but signals that MAEE is not available > > > by setting TH_MXSTATUS_MAEE to 0. > > > > Yep! > > > > > * Consequence: The c906 emulation does not enable any page-base memory > > > attribute mechanism. > > > > Exactly > > > > > * OpenSBI tests the TH_MXSTATUS_MAEE bit (M-mode only) and provides > > > that information to user-space (e.g. DTB). > > > > To the kernel, but yep! > > > > > * The current Linux errata code will be enhanced to not assume MAEE > > > for each core with T-Head vendor ID, but also query the MAEE bit and > > > ensure it is set. > > > > I feel like it should already do that :) > > It doesn't quite do this right now. It only makes the assumption for > CPUs where marchid and mvendorid are zero. The c908, and I think Guo Ren > confirmed it will be the case going forward, sets these to non-zero > values. We should have always required a dt property be set, rather than > using m*id, but we can't go back on that for these devices. Going > forward, if there are more CPUs that want to use this e.g. C908 in MAEE > mode (it can do svpbmt too) I'm gonna require it is explicitly set in > DT. A DT node that we don't set also works fine for us Alistair
On Thu, Mar 28, 2024 at 2:19 AM Alistair Francis <alistair23@gmail.com> wrote: > > On Wed, Mar 27, 2024 at 9:19 PM Conor Dooley <conor@kernel.org> wrote: > > > > Christoph linked here on his submission to Linux of a fix for this, so I > > am reviving this to leave a couple comments :) > > > > On Thu, Feb 15, 2024 at 02:24:02PM +1000, Alistair Francis wrote: > > > On Mon, Feb 5, 2024 at 6:37 PM Christoph Müllner > > > <christoph.muellner@vrull.eu> wrote: > > > > On Mon, Feb 5, 2024 at 3:42 AM Alistair Francis <alistair23@gmail.com> wrote: > > > > > On Sun, Feb 4, 2024 at 3:44 PM LIU Zhiwei <zhiwei_liu@linux.alibaba.com> wrote: > > > > > > > > ppn = (pte & (target_ulong)PTE_PPN_MASK) >> PTE_PPN_SHIFT; > > > > > > > > > > Unfortunately we won't be able to take this upstream. This is core > > > > > QEMU RISC-V code that is now being changed against the spec. I think > > > > > adding the CSR is fine, but we can't take this core change. > > > > > > > > > > A fix that works for everyone should be supporting the th_mxstatus > > > > > CSR, but don't support setting the TH_MXSTATUS_MAEE bit. That way > > > > > guests can detect that the bit isn't set and not use the reserved bits > > > > > in the PTE. From my understanding the extra PTE bits are related to > > > > > cache control in the hardware, which we don't need here > > > > > > > > Sounds good! Let me recap the overall plan: > > > > * QEMU does not emulate MAEE, but signals that MAEE is not available > > > > by setting TH_MXSTATUS_MAEE to 0. > > > > > > Yep! > > > > > > > * Consequence: The c906 emulation does not enable any page-base memory > > > > attribute mechanism. > > > > > > Exactly > > > > > > > * OpenSBI tests the TH_MXSTATUS_MAEE bit (M-mode only) and provides > > > > that information to user-space (e.g. DTB). > > > > > > To the kernel, but yep! > > > > > > > * The current Linux errata code will be enhanced to not assume MAEE > > > > for each core with T-Head vendor ID, but also query the MAEE bit and > > > > ensure it is set. > > > > > > I feel like it should already do that :) > > > > It doesn't quite do this right now. It only makes the assumption for > > CPUs where marchid and mvendorid are zero. The c908, and I think Guo Ren > > confirmed it will be the case going forward, sets these to non-zero > > values. We should have always required a dt property be set, rather than > > using m*id, but we can't go back on that for these devices. Going > > forward, if there are more CPUs that want to use this e.g. C908 in MAEE > > mode (it can do svpbmt too) I'm gonna require it is explicitly set in > > A DT node that we don't set also works fine for us I would really like to do that, but given the page table is set up so early in the boot process probing via CSR is much easier to realize in the kernel than parsing the DT. Therefore, I think th.sxstatus emulation + probing is the best way to move forward. Thanks, Christoph
diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c index 2dcbc9ff32..bfdbb0539a 100644 --- a/target/riscv/cpu.c +++ b/target/riscv/cpu.c @@ -171,6 +171,7 @@ const RISCVIsaExtData isa_edata_arr[] = { ISA_EXT_DATA_ENTRY(xtheadmemidx, PRIV_VERSION_1_11_0, ext_xtheadmemidx), ISA_EXT_DATA_ENTRY(xtheadmempair, PRIV_VERSION_1_11_0, ext_xtheadmempair), ISA_EXT_DATA_ENTRY(xtheadsync, PRIV_VERSION_1_11_0, ext_xtheadsync), + ISA_EXT_DATA_ENTRY(xtheadmaee, PRIV_VERSION_1_11_0, ext_xtheadmaee), ISA_EXT_DATA_ENTRY(xventanacondops, PRIV_VERSION_1_12_0, ext_XVentanaCondOps), DEFINE_PROP_END_OF_LIST(), @@ -506,6 +507,7 @@ static void rv64_thead_c906_cpu_init(Object *obj) cpu->cfg.mvendorid = THEAD_VENDOR_ID; #ifndef CONFIG_USER_ONLY + cpu->cfg.ext_xtheadmaee = true; set_satp_mode_max_supported(cpu, VM_1_10_SV39); #endif @@ -949,6 +951,9 @@ static void riscv_cpu_reset_hold(Object *obj) } pmp_unlock_entries(env); + if (riscv_cpu_cfg(env)->ext_xtheadmaee) { + env->th_mxstatus |= TH_MXSTATUS_MAEE; + } #endif env->xl = riscv_cpu_mxl(env); riscv_cpu_update_mask(env); @@ -1439,6 +1444,7 @@ const RISCVCPUMultiExtConfig riscv_cpu_vendor_exts[] = { MULTI_EXT_CFG_BOOL("xtheadmemidx", ext_xtheadmemidx, false), MULTI_EXT_CFG_BOOL("xtheadmempair", ext_xtheadmempair, false), MULTI_EXT_CFG_BOOL("xtheadsync", ext_xtheadsync, false), + MULTI_EXT_CFG_BOOL("xtheadmaee", ext_xtheadmaee, false), MULTI_EXT_CFG_BOOL("xventanacondops", ext_XVentanaCondOps, false), DEFINE_PROP_END_OF_LIST(), diff --git a/target/riscv/cpu.h b/target/riscv/cpu.h index 5f3955c38d..1bacf40355 100644 --- a/target/riscv/cpu.h +++ b/target/riscv/cpu.h @@ -412,6 +412,14 @@ struct CPUArchState { target_ulong cur_pmmask; target_ulong cur_pmbase; + union { + /* Custom CSR for Xuantie CPU */ + struct { +#ifndef CONFIG_USER_ONLY + target_ulong th_mxstatus; +#endif + }; + }; /* Fields from here on are preserved across CPU reset. */ QEMUTimer *stimer; /* Internal timer for S-mode interrupt */ QEMUTimer *vstimer; /* Internal timer for VS-mode interrupt */ @@ -799,6 +807,7 @@ void riscv_add_satp_mode_properties(Object *obj); bool riscv_cpu_accelerator_compatible(RISCVCPU *cpu); /* CSR function table */ +extern riscv_csr_operations th_csr_ops[CSR_TABLE_SIZE]; extern riscv_csr_operations csr_ops[CSR_TABLE_SIZE]; extern const bool valid_vm_1_10_32[], valid_vm_1_10_64[]; diff --git a/target/riscv/cpu_bits.h b/target/riscv/cpu_bits.h index e116f6c252..67ebb1cefe 100644 --- a/target/riscv/cpu_bits.h +++ b/target/riscv/cpu_bits.h @@ -897,4 +897,10 @@ typedef enum RISCVException { /* JVT CSR bits */ #define JVT_MODE 0x3F #define JVT_BASE (~0x3F) + +/* Xuantie custom CSRs */ +#define CSR_TH_MXSTATUS 0x7c0 + +#define TH_MXSTATUS_MAEE_SHIFT 21 +#define TH_MXSTATUS_MAEE (0x1 << TH_MXSTATUS_MAEE_SHIFT) #endif diff --git a/target/riscv/cpu_cfg.h b/target/riscv/cpu_cfg.h index 780ae6ef17..3735c69fd6 100644 --- a/target/riscv/cpu_cfg.h +++ b/target/riscv/cpu_cfg.h @@ -136,6 +136,7 @@ struct RISCVCPUConfig { bool ext_xtheadmemidx; bool ext_xtheadmempair; bool ext_xtheadsync; + bool ext_xtheadmaee; bool ext_XVentanaCondOps; uint32_t pmu_mask; @@ -176,7 +177,8 @@ static inline bool has_xthead_p(const RISCVCPUConfig *cfg) cfg->ext_xtheadcondmov || cfg->ext_xtheadfmemidx || cfg->ext_xtheadfmv || cfg->ext_xtheadmac || cfg->ext_xtheadmemidx || - cfg->ext_xtheadmempair || cfg->ext_xtheadsync; + cfg->ext_xtheadmempair || cfg->ext_xtheadsync || + cfg->ext_xtheadmaee; } #define MATERIALISE_EXT_PREDICATE(ext) \ diff --git a/target/riscv/cpu_helper.c b/target/riscv/cpu_helper.c index c7cc7eb423..5c1f380276 100644 --- a/target/riscv/cpu_helper.c +++ b/target/riscv/cpu_helper.c @@ -812,6 +812,7 @@ static int get_physical_address(CPURISCVState *env, hwaddr *physical, int napot_bits = 0; target_ulong napot_mask; + bool skip_pte_check = env->th_mxstatus & TH_MXSTATUS_MAEE; /* * Check if we should use the background registers for the two * stage translation. We don't need to check if we actually need @@ -974,18 +975,19 @@ restart: if (riscv_cpu_sxl(env) == MXL_RV32) { ppn = pte >> PTE_PPN_SHIFT; } else { - if (pte & PTE_RESERVED) { - return TRANSLATE_FAIL; - } + if (!skip_pte_check) { + if (pte & PTE_RESERVED) { + return TRANSLATE_FAIL; + } - if (!pbmte && (pte & PTE_PBMT)) { - return TRANSLATE_FAIL; - } + if (!pbmte && (pte & PTE_PBMT)) { + return TRANSLATE_FAIL; + } - if (!riscv_cpu_cfg(env)->ext_svnapot && (pte & PTE_N)) { - return TRANSLATE_FAIL; + if (!riscv_cpu_cfg(env)->ext_svnapot && (pte & PTE_N)) { + return TRANSLATE_FAIL; + } } - ppn = (pte & (target_ulong)PTE_PPN_MASK) >> PTE_PPN_SHIFT; } @@ -998,7 +1000,8 @@ restart: } /* Inner PTE, continue walking */ - if (pte & (PTE_D | PTE_A | PTE_U | PTE_ATTR)) { + if ((pte & (PTE_D | PTE_A | PTE_U)) || + (!skip_pte_check && (pte & PTE_ATTR))) { return TRANSLATE_FAIL; } base = ppn << PGSHIFT; @@ -1012,7 +1015,7 @@ restart: /* Misaligned PPN */ return TRANSLATE_FAIL; } - if (!pbmte && (pte & PTE_PBMT)) { + if (!skip_pte_check && !pbmte && (pte & PTE_PBMT)) { /* Reserved without Svpbmt. */ return TRANSLATE_FAIL; } diff --git a/target/riscv/meson.build b/target/riscv/meson.build index a5e0734e7f..d7f675881d 100644 --- a/target/riscv/meson.build +++ b/target/riscv/meson.build @@ -12,6 +12,7 @@ riscv_ss.add(files( 'cpu.c', 'cpu_helper.c', 'csr.c', + 'xthead_csr.c', 'fpu_helper.c', 'gdbstub.c', 'op_helper.c', diff --git a/target/riscv/tcg/tcg-cpu.c b/target/riscv/tcg/tcg-cpu.c index 559bf373f3..4b1184c8ab 100644 --- a/target/riscv/tcg/tcg-cpu.c +++ b/target/riscv/tcg/tcg-cpu.c @@ -763,7 +763,6 @@ void riscv_cpu_validate_set_extensions(RISCVCPU *cpu, Error **errp) cpu->cfg.pmu_mask = 0; cpu->pmu_avail_ctrs = 0; } - /* * Disable isa extensions based on priv spec after we * validated and set everything we need. @@ -871,12 +870,18 @@ static void riscv_cpu_validate_profiles(RISCVCPU *cpu) } } +static inline bool th_csr_p(const RISCVCPUConfig *cfg) +{ + return cfg->ext_xtheadmaee; +} + void riscv_tcg_cpu_register_vendor_csr(RISCVCPU *cpu) { static const struct { bool (*guard_func)(const RISCVCPUConfig *); riscv_csr_operations *csr_ops; } vendors[] = { + { th_csr_p, th_csr_ops }, }; for (int i = 0; i < ARRAY_SIZE(vendors); ++i) { if (!vendors[i].guard_func(&cpu->cfg)) { diff --git a/target/riscv/xthead_csr.c b/target/riscv/xthead_csr.c new file mode 100644 index 0000000000..9f88fa50db --- /dev/null +++ b/target/riscv/xthead_csr.c @@ -0,0 +1,65 @@ +/* + * Xuantie implementation for RISC-V Control and Status Registers. + * + * Copyright (c) 2024 Alibaba Group. All rights reserved. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms and conditions of the GNU General Public License, + * version 2 or later, as published by the Free Software Foundation. + * + * This program is distributed in the hope it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program. If not, see <http://www.gnu.org/licenses/>. + */ + + +#include "qemu/osdep.h" +#include "qemu/log.h" +#include "cpu.h" +#include "tcg/tcg-cpu.h" +#include "exec/exec-all.h" +#include "exec/tb-flush.h" +#include "qapi/error.h" + +#if !defined(CONFIG_USER_ONLY) +static RISCVException th_maee_check(CPURISCVState *env, int csrno) +{ + if (riscv_cpu_cfg(env)->ext_xtheadmaee) { + return RISCV_EXCP_ILLEGAL_INST; + } + return RISCV_EXCP_NONE; +} + +static RISCVException +read_th_mxstatus(CPURISCVState *env, int csrno, target_ulong *val) +{ + *val = env->th_mxstatus; + return RISCV_EXCP_NONE; +} + +static RISCVException +write_th_mxstatus(CPURISCVState *env, int csrno, target_ulong val) +{ + uint64_t mxstatus = env->th_mxstatus; + uint64_t mask = TH_MXSTATUS_MAEE; + + if ((val ^ mxstatus) & TH_MXSTATUS_MAEE) { + tlb_flush(env_cpu(env)); + } + + mxstatus = (mxstatus & ~mask) | (val & mask); + env->th_mxstatus = mxstatus; + return RISCV_EXCP_NONE; +} +#endif + +riscv_csr_operations th_csr_ops[CSR_TABLE_SIZE] = { +#if !defined(CONFIG_USER_ONLY) + [CSR_TH_MXSTATUS] = { "th_mxstatus", th_maee_check, read_th_mxstatus, + write_th_mxstatus}, +#endif /* !CONFIG_USER_ONLY */ +};
This patch set fix the regression on kernel pointed by Björn Töpel in https://www.mail-archive.com/qemu-devel@nongnu.org/msg1018232.html. thead-c906 uses some flags in pte [60-63] bits. It has history reasons that SVPBMT didn't exist when thead-c906 came to wotrld. We named this feature as xtheadmaee[1]. this feature is controlled by an custom CSR named mxstatus, whose maee field encodes whether enable the pte [60-63] bits. [1]:https://github.com/T-head-Semi/thead-extension-spec/blob/master/xtheadmaee.adoc Signed-off-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com> --- v1->v2: 1) Remove mxstatus user mode access 2) Add reference documentation to the commit log --- target/riscv/cpu.c | 6 ++++ target/riscv/cpu.h | 9 ++++++ target/riscv/cpu_bits.h | 6 ++++ target/riscv/cpu_cfg.h | 4 ++- target/riscv/cpu_helper.c | 25 ++++++++------- target/riscv/meson.build | 1 + target/riscv/tcg/tcg-cpu.c | 7 +++- target/riscv/xthead_csr.c | 65 ++++++++++++++++++++++++++++++++++++++ 8 files changed, 110 insertions(+), 13 deletions(-) create mode 100644 target/riscv/xthead_csr.c