Merge branch 'x86/for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip...
[pandora-kernel.git] / arch / x86 / kernel / vmi_32.c
index f02bad6..b153460 100644 (file)
@@ -62,7 +62,10 @@ static struct {
        void (*cpuid)(void /* non-c */);
        void (*_set_ldt)(u32 selector);
        void (*set_tr)(u32 selector);
-       void (*set_kernel_stack)(u32 selector, u32 esp0);
+       void (*write_idt_entry)(struct desc_struct *, int, u32, u32);
+       void (*write_gdt_entry)(struct desc_struct *, int, u32, u32);
+       void (*write_ldt_entry)(struct desc_struct *, int, u32, u32);
+       void (*set_kernel_stack)(u32 selector, u32 sp0);
        void (*allocate_page)(u32, u32, u32, u32, u32);
        void (*release_page)(u32, u32);
        void (*set_pte)(pte_t, pte_t *, unsigned);
@@ -88,13 +91,13 @@ struct vmi_timer_ops vmi_timer_ops;
 #define IRQ_PATCH_DISABLE  5
 
 static inline void patch_offset(void *insnbuf,
-                               unsigned long eip, unsigned long dest)
+                               unsigned long ip, unsigned long dest)
 {
-        *(unsigned long *)(insnbuf+1) = dest-eip-5;
+        *(unsigned long *)(insnbuf+1) = dest-ip-5;
 }
 
 static unsigned patch_internal(int call, unsigned len, void *insnbuf,
-                              unsigned long eip)
+                              unsigned long ip)
 {
        u64 reloc;
        struct vmi_relocation_info *const rel = (struct vmi_relocation_info *)&reloc;
@@ -103,13 +106,13 @@ static unsigned patch_internal(int call, unsigned len, void *insnbuf,
                case VMI_RELOCATION_CALL_REL:
                        BUG_ON(len < 5);
                        *(char *)insnbuf = MNEM_CALL;
-                       patch_offset(insnbuf, eip, (unsigned long)rel->eip);
+                       patch_offset(insnbuf, ip, (unsigned long)rel->eip);
                        return 5;
 
                case VMI_RELOCATION_JUMP_REL:
                        BUG_ON(len < 5);
                        *(char *)insnbuf = MNEM_JMP;
-                       patch_offset(insnbuf, eip, (unsigned long)rel->eip);
+                       patch_offset(insnbuf, ip, (unsigned long)rel->eip);
                        return 5;
 
                case VMI_RELOCATION_NOP:
@@ -131,25 +134,25 @@ static unsigned patch_internal(int call, unsigned len, void *insnbuf,
  * sequence.  The callee does nop padding for us.
  */
 static unsigned vmi_patch(u8 type, u16 clobbers, void *insns,
-                         unsigned long eip, unsigned len)
+                         unsigned long ip, unsigned len)
 {
        switch (type) {
                case PARAVIRT_PATCH(pv_irq_ops.irq_disable):
                        return patch_internal(VMI_CALL_DisableInterrupts, len,
-                                             insns, eip);
+                                             insns, ip);
                case PARAVIRT_PATCH(pv_irq_ops.irq_enable):
                        return patch_internal(VMI_CALL_EnableInterrupts, len,
-                                             insns, eip);
+                                             insns, ip);
                case PARAVIRT_PATCH(pv_irq_ops.restore_fl):
                        return patch_internal(VMI_CALL_SetInterruptMask, len,
-                                             insns, eip);
+                                             insns, ip);
                case PARAVIRT_PATCH(pv_irq_ops.save_fl):
                        return patch_internal(VMI_CALL_GetInterruptMask, len,
-                                             insns, eip);
+                                             insns, ip);
                case PARAVIRT_PATCH(pv_cpu_ops.iret):
-                       return patch_internal(VMI_CALL_IRET, len, insns, eip);
+                       return patch_internal(VMI_CALL_IRET, len, insns, ip);
                case PARAVIRT_PATCH(pv_cpu_ops.irq_enable_sysexit):
-                       return patch_internal(VMI_CALL_SYSEXIT, len, insns, eip);
+                       return patch_internal(VMI_CALL_SYSEXIT, len, insns, ip);
                default:
                        break;
        }
@@ -157,36 +160,36 @@ static unsigned vmi_patch(u8 type, u16 clobbers, void *insns,
 }
 
 /* CPUID has non-C semantics, and paravirt-ops API doesn't match hardware ISA */
-static void vmi_cpuid(unsigned int *eax, unsigned int *ebx,
-                               unsigned int *ecx, unsigned int *edx)
+static void vmi_cpuid(unsigned int *ax, unsigned int *bx,
+                               unsigned int *cx, unsigned int *dx)
 {
        int override = 0;
-       if (*eax == 1)
+       if (*ax == 1)
                override = 1;
         asm volatile ("call *%6"
-                      : "=a" (*eax),
-                        "=b" (*ebx),
-                        "=c" (*ecx),
-                        "=d" (*edx)
-                      : "0" (*eax), "2" (*ecx), "r" (vmi_ops.cpuid));
+                      : "=a" (*ax),
+                        "=b" (*bx),
+                        "=c" (*cx),
+                        "=d" (*dx)
+                      : "0" (*ax), "2" (*cx), "r" (vmi_ops.cpuid));
        if (override) {
                if (disable_pse)
-                       *edx &= ~X86_FEATURE_PSE;
+                       *dx &= ~X86_FEATURE_PSE;
                if (disable_pge)
-                       *edx &= ~X86_FEATURE_PGE;
+                       *dx &= ~X86_FEATURE_PGE;
                if (disable_sep)
-                       *edx &= ~X86_FEATURE_SEP;
+                       *dx &= ~X86_FEATURE_SEP;
                if (disable_tsc)
-                       *edx &= ~X86_FEATURE_TSC;
+                       *dx &= ~X86_FEATURE_TSC;
                if (disable_mtrr)
-                       *edx &= ~X86_FEATURE_MTRR;
+                       *dx &= ~X86_FEATURE_MTRR;
        }
 }
 
 static inline void vmi_maybe_load_tls(struct desc_struct *gdt, int nr, struct desc_struct *new)
 {
        if (gdt[nr].a != new->a || gdt[nr].b != new->b)
-               write_gdt_entry(gdt, nr, new->a, new->b);
+               write_gdt_entry(gdt, nr, new, 0);
 }
 
 static void vmi_load_tls(struct thread_struct *t, unsigned int cpu)
@@ -200,12 +203,12 @@ static void vmi_load_tls(struct thread_struct *t, unsigned int cpu)
 static void vmi_set_ldt(const void *addr, unsigned entries)
 {
        unsigned cpu = smp_processor_id();
-       u32 low, high;
+       struct desc_struct desc;
 
-       pack_descriptor(&low, &high, (unsigned long)addr,
+       pack_descriptor(&desc, (unsigned long)addr,
                        entries * sizeof(struct desc_struct) - 1,
-                       DESCTYPE_LDT, 0);
-       write_gdt_entry(get_cpu_gdt_table(cpu), GDT_ENTRY_LDT, low, high);
+                       DESC_LDT, 0);
+       write_gdt_entry(get_cpu_gdt_table(cpu), GDT_ENTRY_LDT, &desc, DESC_LDT);
        vmi_ops._set_ldt(entries ? GDT_ENTRY_LDT*sizeof(struct desc_struct) : 0);
 }
 
@@ -214,17 +217,37 @@ static void vmi_set_tr(void)
        vmi_ops.set_tr(GDT_ENTRY_TSS*sizeof(struct desc_struct));
 }
 
-static void vmi_load_esp0(struct tss_struct *tss,
+static void vmi_write_idt_entry(gate_desc *dt, int entry, const gate_desc *g)
+{
+       u32 *idt_entry = (u32 *)g;
+       vmi_ops.write_idt_entry(dt, entry, idt_entry[0], idt_entry[1]);
+}
+
+static void vmi_write_gdt_entry(struct desc_struct *dt, int entry,
+                               const void *desc, int type)
+{
+       u32 *gdt_entry = (u32 *)desc;
+       vmi_ops.write_gdt_entry(dt, entry, gdt_entry[0], gdt_entry[1]);
+}
+
+static void vmi_write_ldt_entry(struct desc_struct *dt, int entry,
+                               const void *desc)
+{
+       u32 *ldt_entry = (u32 *)desc;
+       vmi_ops.write_idt_entry(dt, entry, ldt_entry[0], ldt_entry[1]);
+}
+
+static void vmi_load_sp0(struct tss_struct *tss,
                                   struct thread_struct *thread)
 {
-       tss->x86_tss.esp0 = thread->esp0;
+       tss->x86_tss.sp0 = thread->sp0;
 
        /* This can only happen when SEP is enabled, no need to test "SEP"arately */
        if (unlikely(tss->x86_tss.ss1 != thread->sysenter_cs)) {
                tss->x86_tss.ss1 = thread->sysenter_cs;
                wrmsr(MSR_IA32_SYSENTER_CS, thread->sysenter_cs, 0);
        }
-       vmi_ops.set_kernel_stack(__KERNEL_DS, tss->x86_tss.esp0);
+       vmi_ops.set_kernel_stack(__KERNEL_DS, tss->x86_tss.sp0);
 }
 
 static void vmi_flush_tlb_user(void)
@@ -297,7 +320,7 @@ static void check_zeroed_page(u32 pfn, int type, struct page *page)
         * pdes need to be zeroed.
         */
        if (type & VMI_PAGE_CLONE)
-               limit = USER_PTRS_PER_PGD;
+               limit = KERNEL_PGD_BOUNDARY;
        for (i = 0; i < limit; i++)
                BUG_ON(ptr[i]);
 }
@@ -369,13 +392,13 @@ static void *vmi_kmap_atomic_pte(struct page *page, enum km_type type)
 }
 #endif
 
-static void vmi_allocate_pt(struct mm_struct *mm, u32 pfn)
+static void vmi_allocate_pte(struct mm_struct *mm, u32 pfn)
 {
        vmi_set_page_type(pfn, VMI_PAGE_L1);
        vmi_ops.allocate_page(pfn, VMI_PAGE_L1, 0, 0, 0);
 }
 
-static void vmi_allocate_pd(u32 pfn)
+static void vmi_allocate_pmd(struct mm_struct *mm, u32 pfn)
 {
        /*
         * This call comes in very early, before mem_map is setup.
@@ -386,20 +409,20 @@ static void vmi_allocate_pd(u32 pfn)
        vmi_ops.allocate_page(pfn, VMI_PAGE_L2, 0, 0, 0);
 }
 
-static void vmi_allocate_pd_clone(u32 pfn, u32 clonepfn, u32 start, u32 count)
+static void vmi_allocate_pmd_clone(u32 pfn, u32 clonepfn, u32 start, u32 count)
 {
        vmi_set_page_type(pfn, VMI_PAGE_L2 | VMI_PAGE_CLONE);
        vmi_check_page_type(clonepfn, VMI_PAGE_L2);
        vmi_ops.allocate_page(pfn, VMI_PAGE_L2 | VMI_PAGE_CLONE, clonepfn, start, count);
 }
 
-static void vmi_release_pt(u32 pfn)
+static void vmi_release_pte(u32 pfn)
 {
        vmi_ops.release_page(pfn, VMI_PAGE_L1);
        vmi_set_page_type(pfn, VMI_PAGE_NORMAL);
 }
 
-static void vmi_release_pd(u32 pfn)
+static void vmi_release_pmd(u32 pfn)
 {
        vmi_ops.release_page(pfn, VMI_PAGE_L2);
        vmi_set_page_type(pfn, VMI_PAGE_NORMAL);
@@ -452,7 +475,7 @@ static void vmi_set_pte_at(struct mm_struct *mm, unsigned long addr, pte_t *ptep
 static void vmi_set_pmd(pmd_t *pmdp, pmd_t pmdval)
 {
 #ifdef CONFIG_X86_PAE
-       const pte_t pte = { pmdval.pmd, pmdval.pmd >> 32 };
+       const pte_t pte = { .pte = pmdval.pmd };
        vmi_check_page_type(__pa(pmdp) >> PAGE_SHIFT, VMI_PAGE_PMD);
 #else
        const pte_t pte = { pmdval.pud.pgd.pgd };
@@ -485,21 +508,21 @@ static void vmi_set_pte_present(struct mm_struct *mm, unsigned long addr, pte_t
 static void vmi_set_pud(pud_t *pudp, pud_t pudval)
 {
        /* Um, eww */
-       const pte_t pte = { pudval.pgd.pgd, pudval.pgd.pgd >> 32 };
+       const pte_t pte = { .pte = pudval.pgd.pgd };
        vmi_check_page_type(__pa(pudp) >> PAGE_SHIFT, VMI_PAGE_PGD);
        vmi_ops.set_pte(pte, (pte_t *)pudp, VMI_PAGE_PDP);
 }
 
 static void vmi_pte_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
 {
-       const pte_t pte = { 0 };
+       const pte_t pte = { .pte = 0 };
        vmi_check_page_type(__pa(ptep) >> PAGE_SHIFT, VMI_PAGE_PTE);
        vmi_ops.set_pte(pte, ptep, vmi_flags_addr(mm, addr, VMI_PAGE_PT, 0));
 }
 
 static void vmi_pmd_clear(pmd_t *pmd)
 {
-       const pte_t pte = { 0 };
+       const pte_t pte = { .pte = 0 };
        vmi_check_page_type(__pa(pmd) >> PAGE_SHIFT, VMI_PAGE_PMD);
        vmi_ops.set_pte(pte, (pte_t *)pmd, VMI_PAGE_PD);
 }
@@ -790,10 +813,13 @@ static inline int __init activate_vmi(void)
        para_fill(pv_cpu_ops.store_idt, GetIDT);
        para_fill(pv_cpu_ops.store_tr, GetTR);
        pv_cpu_ops.load_tls = vmi_load_tls;
-       para_fill(pv_cpu_ops.write_ldt_entry, WriteLDTEntry);
-       para_fill(pv_cpu_ops.write_gdt_entry, WriteGDTEntry);
-       para_fill(pv_cpu_ops.write_idt_entry, WriteIDTEntry);
-       para_wrap(pv_cpu_ops.load_esp0, vmi_load_esp0, set_kernel_stack, UpdateKernelStack);
+       para_wrap(pv_cpu_ops.write_ldt_entry, vmi_write_ldt_entry,
+                 write_ldt_entry, WriteLDTEntry);
+       para_wrap(pv_cpu_ops.write_gdt_entry, vmi_write_gdt_entry,
+                 write_gdt_entry, WriteGDTEntry);
+       para_wrap(pv_cpu_ops.write_idt_entry, vmi_write_idt_entry,
+                 write_idt_entry, WriteIDTEntry);
+       para_wrap(pv_cpu_ops.load_sp0, vmi_load_sp0, set_kernel_stack, UpdateKernelStack);
        para_fill(pv_cpu_ops.set_iopl_mask, SetIOPLMask);
        para_fill(pv_cpu_ops.io_delay, IODelay);
 
@@ -845,15 +871,15 @@ static inline int __init activate_vmi(void)
 
        vmi_ops.allocate_page = vmi_get_function(VMI_CALL_AllocatePage);
        if (vmi_ops.allocate_page) {
-               pv_mmu_ops.alloc_pt = vmi_allocate_pt;
-               pv_mmu_ops.alloc_pd = vmi_allocate_pd;
-               pv_mmu_ops.alloc_pd_clone = vmi_allocate_pd_clone;
+               pv_mmu_ops.alloc_pte = vmi_allocate_pte;
+               pv_mmu_ops.alloc_pmd = vmi_allocate_pmd;
+               pv_mmu_ops.alloc_pmd_clone = vmi_allocate_pmd_clone;
        }
 
        vmi_ops.release_page = vmi_get_function(VMI_CALL_ReleasePage);
        if (vmi_ops.release_page) {
-               pv_mmu_ops.release_pt = vmi_release_pt;
-               pv_mmu_ops.release_pd = vmi_release_pd;
+               pv_mmu_ops.release_pte = vmi_release_pte;
+               pv_mmu_ops.release_pmd = vmi_release_pmd;
        }
 
        /* Set linear is needed in all cases */
@@ -906,7 +932,7 @@ static inline int __init activate_vmi(void)
                pv_apic_ops.setup_secondary_clock = vmi_time_ap_init;
 #endif
                pv_time_ops.sched_clock = vmi_sched_clock;
-               pv_time_ops.get_cpu_khz = vmi_cpu_khz;
+               pv_time_ops.get_tsc_khz = vmi_tsc_khz;
 
                /* We have true wallclock functions; disable CMOS clock sync */
                no_sync_cmos_clock = 1;
@@ -963,19 +989,19 @@ static int __init parse_vmi(char *arg)
                return -EINVAL;
 
        if (!strcmp(arg, "disable_pge")) {
-               clear_bit(X86_FEATURE_PGE, boot_cpu_data.x86_capability);
+               clear_cpu_cap(&boot_cpu_data, X86_FEATURE_PGE);
                disable_pge = 1;
        } else if (!strcmp(arg, "disable_pse")) {
-               clear_bit(X86_FEATURE_PSE, boot_cpu_data.x86_capability);
+               clear_cpu_cap(&boot_cpu_data, X86_FEATURE_PSE);
                disable_pse = 1;
        } else if (!strcmp(arg, "disable_sep")) {
-               clear_bit(X86_FEATURE_SEP, boot_cpu_data.x86_capability);
+               clear_cpu_cap(&boot_cpu_data, X86_FEATURE_SEP);
                disable_sep = 1;
        } else if (!strcmp(arg, "disable_tsc")) {
-               clear_bit(X86_FEATURE_TSC, boot_cpu_data.x86_capability);
+               clear_cpu_cap(&boot_cpu_data, X86_FEATURE_TSC);
                disable_tsc = 1;
        } else if (!strcmp(arg, "disable_mtrr")) {
-               clear_bit(X86_FEATURE_MTRR, boot_cpu_data.x86_capability);
+               clear_cpu_cap(&boot_cpu_data, X86_FEATURE_MTRR);
                disable_mtrr = 1;
        } else if (!strcmp(arg, "disable_timer")) {
                disable_vmi_timer = 1;