xref: /openbmc/qemu/target/i386/tcg/sysemu/excp_helper.c (revision 93eae3583256896dd91a4c2ca38dcbb8d4051cff)

/*
 *  x86 exception helpers - sysemu code
 *
 *  Copyright (c) 2003 Fabrice Bellard
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with this library; if not, see <http://www.gnu.org/licenses/>.
 */

#include "qemu/osdep.h"
#include "cpu.h"
#include "tcg/helper-tcg.h"

int get_pg_mode(CPUX86State *env)
{
    int pg_mode = 0;
    if (env->cr[0] & CR0_WP_MASK) {
        pg_mode |= PG_MODE_WP;
    }
    if (env->cr[4] & CR4_PAE_MASK) {
        pg_mode |= PG_MODE_PAE;
    }
    if (env->cr[4] & CR4_PSE_MASK) {
        pg_mode |= PG_MODE_PSE;
    }
    if (env->cr[4] & CR4_PKE_MASK) {
        pg_mode |= PG_MODE_PKE;
    }
    if (env->cr[4] & CR4_PKS_MASK) {
        pg_mode |= PG_MODE_PKS;
    }
    if (env->cr[4] & CR4_SMEP_MASK) {
        pg_mode |= PG_MODE_SMEP;
    }
    if (env->cr[4] & CR4_LA57_MASK) {
        pg_mode |= PG_MODE_LA57;
    }
    if (env->hflags & HF_LMA_MASK) {
        pg_mode |= PG_MODE_LMA;
    }
    if (env->efer & MSR_EFER_NXE) {
        pg_mode |= PG_MODE_NXE;
    }
    return pg_mode;
}

#define PG_ERROR_OK (-1)

typedef hwaddr (*MMUTranslateFunc)(CPUState *cs, hwaddr gphys, MMUAccessType access_type,
				int *prot);

#define GET_HPHYS(cs, gpa, access_type, prot)  \
	(get_hphys_func ? get_hphys_func(cs, gpa, access_type, prot) : gpa)

static int mmu_translate(CPUState *cs, hwaddr addr, MMUTranslateFunc get_hphys_func,
                         uint64_t cr3, int is_write1, int mmu_idx, int pg_mode,
                         hwaddr *xlat, int *page_size, int *prot)
{
    X86CPU *cpu = X86_CPU(cs);
    CPUX86State *env = &cpu->env;
    uint64_t ptep, pte;
    int32_t a20_mask;
    target_ulong pde_addr, pte_addr;
    int error_code = 0;
    int is_dirty, is_write, is_user;
    uint64_t rsvd_mask = PG_ADDRESS_MASK & ~MAKE_64BIT_MASK(0, cpu->phys_bits);
    uint32_t page_offset;
    uint32_t pkr;

    is_user = (mmu_idx == MMU_USER_IDX);
    is_write = is_write1 & 1;
    a20_mask = x86_get_a20_mask(env);

    if (!(pg_mode & PG_MODE_NXE)) {
        rsvd_mask |= PG_NX_MASK;
    }

    if (pg_mode & PG_MODE_PAE) {
        uint64_t pde, pdpe;
        target_ulong pdpe_addr;

#ifdef TARGET_X86_64
        if (pg_mode & PG_MODE_LMA) {
            bool la57 = pg_mode & PG_MODE_LA57;
            uint64_t pml5e_addr, pml5e;
            uint64_t pml4e_addr, pml4e;
            int32_t sext;

            /* test virtual address sign extension */
            sext = la57 ? (int64_t)addr >> 56 : (int64_t)addr >> 47;
            if (get_hphys_func && sext != 0 && sext != -1) {
                env->error_code = 0;
                cs->exception_index = EXCP0D_GPF;
                return 1;
            }

            if (la57) {
                pml5e_addr = ((cr3 & ~0xfff) +
                        (((addr >> 48) & 0x1ff) << 3)) & a20_mask;
                pml5e_addr = GET_HPHYS(cs, pml5e_addr, MMU_DATA_STORE, NULL);
                pml5e = x86_ldq_phys(cs, pml5e_addr);
                if (!(pml5e & PG_PRESENT_MASK)) {
                    goto do_fault;
                }
                if (pml5e & (rsvd_mask | PG_PSE_MASK)) {
                    goto do_fault_rsvd;
                }
                if (!(pml5e & PG_ACCESSED_MASK)) {
                    pml5e |= PG_ACCESSED_MASK;
                    x86_stl_phys_notdirty(cs, pml5e_addr, pml5e);
                }
                ptep = pml5e ^ PG_NX_MASK;
            } else {
                pml5e = cr3;
                ptep = PG_NX_MASK | PG_USER_MASK | PG_RW_MASK;
            }

            pml4e_addr = ((pml5e & PG_ADDRESS_MASK) +
                    (((addr >> 39) & 0x1ff) << 3)) & a20_mask;
            pml4e_addr = GET_HPHYS(cs, pml4e_addr, MMU_DATA_STORE, NULL);
            pml4e = x86_ldq_phys(cs, pml4e_addr);
            if (!(pml4e & PG_PRESENT_MASK)) {
                goto do_fault;
            }
            if (pml4e & (rsvd_mask | PG_PSE_MASK)) {
                goto do_fault_rsvd;
            }
            if (!(pml4e & PG_ACCESSED_MASK)) {
                pml4e |= PG_ACCESSED_MASK;
                x86_stl_phys_notdirty(cs, pml4e_addr, pml4e);
            }
            ptep &= pml4e ^ PG_NX_MASK;
            pdpe_addr = ((pml4e & PG_ADDRESS_MASK) + (((addr >> 30) & 0x1ff) << 3)) &
                a20_mask;
            pdpe_addr = GET_HPHYS(cs, pdpe_addr, MMU_DATA_STORE, NULL);
            pdpe = x86_ldq_phys(cs, pdpe_addr);
            if (!(pdpe & PG_PRESENT_MASK)) {
                goto do_fault;
            }
            if (pdpe & rsvd_mask) {
                goto do_fault_rsvd;
            }
            ptep &= pdpe ^ PG_NX_MASK;
            if (!(pdpe & PG_ACCESSED_MASK)) {
                pdpe |= PG_ACCESSED_MASK;
                x86_stl_phys_notdirty(cs, pdpe_addr, pdpe);
            }
            if (pdpe & PG_PSE_MASK) {
                /* 1 GB page */
                *page_size = 1024 * 1024 * 1024;
                pte_addr = pdpe_addr;
                pte = pdpe;
                goto do_check_protect;
            }
        } else
#endif
        {
            /* XXX: load them when cr3 is loaded ? */
            pdpe_addr = ((cr3 & ~0x1f) + ((addr >> 27) & 0x18)) &
                a20_mask;
            pdpe_addr = GET_HPHYS(cs, pdpe_addr, MMU_DATA_STORE, NULL);
            pdpe = x86_ldq_phys(cs, pdpe_addr);
            if (!(pdpe & PG_PRESENT_MASK)) {
                goto do_fault;
            }
            rsvd_mask |= PG_HI_USER_MASK;
            if (pdpe & (rsvd_mask | PG_NX_MASK)) {
                goto do_fault_rsvd;
            }
            ptep = PG_NX_MASK | PG_USER_MASK | PG_RW_MASK;
        }

        pde_addr = ((pdpe & PG_ADDRESS_MASK) + (((addr >> 21) & 0x1ff) << 3)) &
            a20_mask;
        pde_addr = GET_HPHYS(cs, pde_addr, MMU_DATA_STORE, NULL);
        pde = x86_ldq_phys(cs, pde_addr);
        if (!(pde & PG_PRESENT_MASK)) {
            goto do_fault;
        }
        if (pde & rsvd_mask) {
            goto do_fault_rsvd;
        }
        ptep &= pde ^ PG_NX_MASK;
        if (pde & PG_PSE_MASK) {
            /* 2 MB page */
            *page_size = 2048 * 1024;
            pte_addr = pde_addr;
            pte = pde;
            goto do_check_protect;
        }
        /* 4 KB page */
        if (!(pde & PG_ACCESSED_MASK)) {
            pde |= PG_ACCESSED_MASK;
            x86_stl_phys_notdirty(cs, pde_addr, pde);
        }
        pte_addr = ((pde & PG_ADDRESS_MASK) + (((addr >> 12) & 0x1ff) << 3)) &
            a20_mask;
        pte_addr = GET_HPHYS(cs, pte_addr, MMU_DATA_STORE, NULL);
        pte = x86_ldq_phys(cs, pte_addr);
        if (!(pte & PG_PRESENT_MASK)) {
            goto do_fault;
        }
        if (pte & rsvd_mask) {
            goto do_fault_rsvd;
        }
        /* combine pde and pte nx, user and rw protections */
        ptep &= pte ^ PG_NX_MASK;
        *page_size = 4096;
    } else {
        uint32_t pde;

        /* page directory entry */
        pde_addr = ((cr3 & ~0xfff) + ((addr >> 20) & 0xffc)) &
            a20_mask;
        pde_addr = GET_HPHYS(cs, pde_addr, MMU_DATA_STORE, NULL);
        pde = x86_ldl_phys(cs, pde_addr);
        if (!(pde & PG_PRESENT_MASK)) {
            goto do_fault;
        }
        ptep = pde | PG_NX_MASK;

        /* if PSE bit is set, then we use a 4MB page */
        if ((pde & PG_PSE_MASK) && (pg_mode & PG_MODE_PSE)) {
            *page_size = 4096 * 1024;
            pte_addr = pde_addr;

            /* Bits 20-13 provide bits 39-32 of the address, bit 21 is reserved.
             * Leave bits 20-13 in place for setting accessed/dirty bits below.
             */
            pte = pde | ((pde & 0x1fe000LL) << (32 - 13));
            rsvd_mask = 0x200000;
            goto do_check_protect_pse36;
        }

        if (!(pde & PG_ACCESSED_MASK)) {
            pde |= PG_ACCESSED_MASK;
            x86_stl_phys_notdirty(cs, pde_addr, pde);
        }

        /* page directory entry */
        pte_addr = ((pde & ~0xfff) + ((addr >> 10) & 0xffc)) &
            a20_mask;
        pte_addr = GET_HPHYS(cs, pte_addr, MMU_DATA_STORE, NULL);
        pte = x86_ldl_phys(cs, pte_addr);
        if (!(pte & PG_PRESENT_MASK)) {
            goto do_fault;
        }
        /* combine pde and pte user and rw protections */
        ptep &= pte | PG_NX_MASK;
        *page_size = 4096;
        rsvd_mask = 0;
    }

do_check_protect:
    rsvd_mask |= (*page_size - 1) & PG_ADDRESS_MASK & ~PG_PSE_PAT_MASK;
do_check_protect_pse36:
    if (pte & rsvd_mask) {
        goto do_fault_rsvd;
    }
    ptep ^= PG_NX_MASK;

    /* can the page can be put in the TLB?  prot will tell us */
    if (is_user && !(ptep & PG_USER_MASK)) {
        goto do_fault_protect;
    }

    *prot = 0;
    if (mmu_idx != MMU_KSMAP_IDX || !(ptep & PG_USER_MASK)) {
        *prot |= PAGE_READ;
        if ((ptep & PG_RW_MASK) || !(is_user || (pg_mode & PG_MODE_WP))) {
            *prot |= PAGE_WRITE;
        }
    }
    if (!(ptep & PG_NX_MASK) &&
        (mmu_idx == MMU_USER_IDX ||
         !((pg_mode & PG_MODE_SMEP) && (ptep & PG_USER_MASK)))) {
        *prot |= PAGE_EXEC;
    }

    if (!(pg_mode & PG_MODE_LMA)) {
        pkr = 0;
    } else if (ptep & PG_USER_MASK) {
        pkr = pg_mode & PG_MODE_PKE ? env->pkru : 0;
    } else {
        pkr = pg_mode & PG_MODE_PKS ? env->pkrs : 0;
    }
    if (pkr) {
        uint32_t pk = (pte & PG_PKRU_MASK) >> PG_PKRU_BIT;
        uint32_t pkr_ad = (pkr >> pk * 2) & 1;
        uint32_t pkr_wd = (pkr >> pk * 2) & 2;
        uint32_t pkr_prot = PAGE_READ | PAGE_WRITE | PAGE_EXEC;

        if (pkr_ad) {
            pkr_prot &= ~(PAGE_READ | PAGE_WRITE);
        } else if (pkr_wd && (is_user || (pg_mode & PG_MODE_WP))) {
            pkr_prot &= ~PAGE_WRITE;
        }

        *prot &= pkr_prot;
        if ((pkr_prot & (1 << is_write1)) == 0) {
            assert(is_write1 != 2);
            error_code |= PG_ERROR_PK_MASK;
            goto do_fault_protect;
        }
    }

    if ((*prot & (1 << is_write1)) == 0) {
        goto do_fault_protect;
    }

    /* yes, it can! */
    is_dirty = is_write && !(pte & PG_DIRTY_MASK);
    if (!(pte & PG_ACCESSED_MASK) || is_dirty) {
        pte |= PG_ACCESSED_MASK;
        if (is_dirty) {
            pte |= PG_DIRTY_MASK;
        }
        x86_stl_phys_notdirty(cs, pte_addr, pte);
    }

    if (!(pte & PG_DIRTY_MASK)) {
        /* only set write access if already dirty... otherwise wait
           for dirty access */
        assert(!is_write);
        *prot &= ~PAGE_WRITE;
    }

    pte = pte & a20_mask;

    /* align to page_size */
    pte &= PG_ADDRESS_MASK & ~(*page_size - 1);
    page_offset = addr & (*page_size - 1);
    *xlat = GET_HPHYS(cs, pte + page_offset, is_write1, prot);
    return PG_ERROR_OK;

 do_fault_rsvd:
    error_code |= PG_ERROR_RSVD_MASK;
 do_fault_protect:
    error_code |= PG_ERROR_P_MASK;
 do_fault:
    error_code |= (is_write << PG_ERROR_W_BIT);
    if (is_user)
        error_code |= PG_ERROR_U_MASK;
    if (is_write1 == 2 &&
        (((pg_mode & PG_MODE_NXE) && (pg_mode & PG_MODE_PAE)) ||
         (pg_mode & PG_MODE_SMEP)))
        error_code |= PG_ERROR_I_D_MASK;
    return error_code;
}

hwaddr get_hphys(CPUState *cs, hwaddr gphys, MMUAccessType access_type,
                        int *prot)
{
    CPUX86State *env = &X86_CPU(cs)->env;
    uint64_t exit_info_1;
    int page_size;
    int next_prot;
    hwaddr hphys;

    if (likely(!(env->hflags2 & HF2_NPT_MASK))) {
        return gphys;
    }

    exit_info_1 = mmu_translate(cs, gphys, NULL, env->nested_cr3,
                               access_type, MMU_USER_IDX, env->nested_pg_mode,
                               &hphys, &page_size, &next_prot);
    if (exit_info_1 == PG_ERROR_OK) {
        if (prot) {
            *prot &= next_prot;
        }
        return hphys;
    }

    x86_stq_phys(cs, env->vm_vmcb + offsetof(struct vmcb, control.exit_info_2),
                 gphys);
    if (prot) {
        exit_info_1 |= SVM_NPTEXIT_GPA;
    } else { /* page table access */
        exit_info_1 |= SVM_NPTEXIT_GPT;
    }
    cpu_vmexit(env, SVM_EXIT_NPF, exit_info_1, env->retaddr);
}

/* return value:
 * -1 = cannot handle fault
 * 0  = nothing more to do
 * 1  = generate PF fault
 */
static int handle_mmu_fault(CPUState *cs, vaddr addr, int size,
                            int is_write1, int mmu_idx)
{
    X86CPU *cpu = X86_CPU(cs);
    CPUX86State *env = &cpu->env;
    int error_code = PG_ERROR_OK;
    int pg_mode, prot, page_size;
    hwaddr paddr;
    hwaddr vaddr;

#if defined(DEBUG_MMU)
    printf("MMU fault: addr=%" VADDR_PRIx " w=%d mmu=%d eip=" TARGET_FMT_lx "\n",
           addr, is_write1, mmu_idx, env->eip);
#endif

    if (!(env->cr[0] & CR0_PG_MASK)) {
        paddr = addr;
#ifdef TARGET_X86_64
        if (!(env->hflags & HF_LMA_MASK)) {
            /* Without long mode we can only address 32bits in real mode */
            paddr = (uint32_t)paddr;
        }
#endif
        prot = PAGE_READ | PAGE_WRITE | PAGE_EXEC;
        page_size = 4096;
    } else {
        pg_mode = get_pg_mode(env);
        error_code = mmu_translate(cs, addr, get_hphys, env->cr[3], is_write1,
                                   mmu_idx, pg_mode,
                                   &paddr, &page_size, &prot);
    }

    if (error_code == PG_ERROR_OK) {
        /* Even if 4MB pages, we map only one 4KB page in the cache to
           avoid filling it too fast */
        vaddr = addr & TARGET_PAGE_MASK;
        paddr &= TARGET_PAGE_MASK;

        assert(prot & (1 << is_write1));
        tlb_set_page_with_attrs(cs, vaddr, paddr, cpu_get_mem_attrs(env),
                                prot, mmu_idx, page_size);
        return 0;
    } else {
        if (env->intercept_exceptions & (1 << EXCP0E_PAGE)) {
            /* cr2 is not modified in case of exceptions */
            x86_stq_phys(cs,
                     env->vm_vmcb + offsetof(struct vmcb, control.exit_info_2),
                     addr);
        } else {
            env->cr[2] = addr;
        }
        env->error_code = error_code;
        cs->exception_index = EXCP0E_PAGE;
        return 1;
    }
}

bool x86_cpu_tlb_fill(CPUState *cs, vaddr addr, int size,
                      MMUAccessType access_type, int mmu_idx,
                      bool probe, uintptr_t retaddr)
{
    X86CPU *cpu = X86_CPU(cs);
    CPUX86State *env = &cpu->env;

    env->retaddr = retaddr;
    if (handle_mmu_fault(cs, addr, size, access_type, mmu_idx)) {
        /* FIXME: On error in get_hphys we have already jumped out.  */
        g_assert(!probe);
        raise_exception_err_ra(env, cs->exception_index,
                               env->error_code, retaddr);
    }
    return true;
}