X-Git-Url: https://err.no/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=drivers%2Flguest%2Flg.h;h=2337e1a06f023ffd226c3ad9748ca3104c41d9e4;hb=2b7e5bcbd9e03f7236d2869f4261059074ea50a2;hp=64f0abed317c2fbb6afe4c7d3ca881c163353ef7;hpb=a6ce22a5f61ba838b43763bf1e5be399f9dee4d0;p=linux-2.6 diff --git a/drivers/lguest/lg.h b/drivers/lguest/lg.h index 64f0abed31..2337e1a06f 100644 --- a/drivers/lguest/lg.h +++ b/drivers/lguest/lg.h @@ -1,119 +1,26 @@ #ifndef _LGUEST_H #define _LGUEST_H -#include - -#define GDT_ENTRY_LGUEST_CS 10 -#define GDT_ENTRY_LGUEST_DS 11 -#define LGUEST_CS (GDT_ENTRY_LGUEST_CS * 8) -#define LGUEST_DS (GDT_ENTRY_LGUEST_DS * 8) - #ifndef __ASSEMBLY__ #include #include #include -#include -#include #include #include #include +#include #include #include -#include "irq_vectors.h" - -#define GUEST_PL 1 -struct lguest_regs -{ - /* Manually saved part. */ - unsigned long ebx, ecx, edx; - unsigned long esi, edi, ebp; - unsigned long gs; - unsigned long eax; - unsigned long fs, ds, es; - unsigned long trapnum, errcode; - /* Trap pushed part */ - unsigned long eip; - unsigned long cs; - unsigned long eflags; - unsigned long esp; - unsigned long ss; -}; +#include void free_pagetables(void); int init_pagetables(struct page **switcher_page, unsigned int pages); -/* Full 4G segment descriptors, suitable for CS and DS. */ -#define FULL_EXEC_SEGMENT ((struct desc_struct){0x0000ffff, 0x00cf9b00}) -#define FULL_SEGMENT ((struct desc_struct){0x0000ffff, 0x00cf9300}) - -struct lguest_dma_info -{ - struct list_head list; - union futex_key key; - unsigned long dmas; - u16 next_dma; - u16 num_dmas; - u16 guestid; - u8 interrupt; /* 0 when not registered */ -}; - -/*H:310 The page-table code owes a great debt of gratitude to Andi Kleen. He - * reviewed the original code which used "u32" for all page table entries, and - * insisted that it would be far clearer with explicit typing. I thought it - * was overkill, but he was right: it is much clearer than it was before. - * - * We have separate types for the Guest's ptes & pgds and the shadow ptes & - * pgds. There's already a Linux type for these (pte_t and pgd_t) but they - * change depending on kernel config options (PAE). */ - -/* Each entry is identical: lower 12 bits of flags and upper 20 bits for the - * "page frame number" (0 == first physical page, etc). They are different - * types so the compiler will warn us if we mix them improperly. */ -typedef union { - struct { unsigned flags:12, pfn:20; }; - struct { unsigned long val; } raw; -} spgd_t; -typedef union { - struct { unsigned flags:12, pfn:20; }; - struct { unsigned long val; } raw; -} spte_t; -typedef union { - struct { unsigned flags:12, pfn:20; }; - struct { unsigned long val; } raw; -} gpgd_t; -typedef union { - struct { unsigned flags:12, pfn:20; }; - struct { unsigned long val; } raw; -} gpte_t; - -/* We have two convenient macros to convert a "raw" value as handed to us by - * the Guest into the correct Guest PGD or PTE type. */ -#define mkgpte(_val) ((gpte_t){.raw.val = _val}) -#define mkgpgd(_val) ((gpgd_t){.raw.val = _val}) -/*:*/ - struct pgdir { - unsigned long cr3; - spgd_t *pgdir; -}; - -/* This is a guest-specific page (mapped ro) into the guest. */ -struct lguest_ro_state -{ - /* Host information we need to restore when we switch back. */ - u32 host_cr3; - struct Xgt_desc_struct host_idt_desc; - struct Xgt_desc_struct host_gdt_desc; - u32 host_sp; - - /* Fields which are used when guest is running. */ - struct Xgt_desc_struct guest_idt_desc; - struct Xgt_desc_struct guest_gdt_desc; - struct i386_hw_tss guest_tss; - struct desc_struct guest_idt[IDT_ENTRIES]; - struct desc_struct guest_gdt[GDT_ENTRIES]; + unsigned long gpgdir; + pgd_t *pgdir; }; /* We have two pages shared with guests, per cpu. */ @@ -132,137 +39,172 @@ struct lguest_pages #define CHANGED_GDT_TLS 4 /* Actually a subset of CHANGED_GDT */ #define CHANGED_ALL 3 -/* The private info the thread maintains about the guest. */ -struct lguest -{ - /* At end of a page shared mapped over lguest_pages in guest. */ - unsigned long regs_page; - struct lguest_regs *regs; - struct lguest_data __user *lguest_data; +struct lguest; + +struct lg_cpu { + unsigned int id; + struct lguest *lg; struct task_struct *tsk; struct mm_struct *mm; /* == tsk->mm, but that becomes NULL on exit */ - u16 guestid; - u32 pfn_limit; - u32 page_offset; + u32 cr2; - int halted; int ts; - u32 next_hcall; u32 esp1; u8 ss1; + /* Bitmap of what has changed: see CHANGED_* above. */ + int changed; + + unsigned long pending_notify; /* pfn from LHCALL_NOTIFY */ + + /* At end of a page shared mapped over lguest_pages in guest. */ + unsigned long regs_page; + struct lguest_regs *regs; + + struct lguest_pages *last_pages; + + int cpu_pgd; /* which pgd this cpu is currently using */ + + /* If a hypercall was asked for, this points to the arguments. */ + struct hcall_args *hcall; + u32 next_hcall; + + /* Virtual clock device */ + struct hrtimer hrt; + /* Do we need to stop what we're doing and return to userspace? */ int break_out; wait_queue_head_t break_wq; + int halted; - /* Bitmap of what has changed: see CHANGED_* above. */ - int changed; - struct lguest_pages *last_pages; + /* Pending virtual interrupts */ + DECLARE_BITMAP(irqs_pending, LGUEST_IRQS); - /* We keep a small number of these. */ - u32 pgdidx; - struct pgdir pgdirs[4]; + struct lg_cpu_arch arch; +}; - /* Cached wakeup: we hold a reference to this task. */ - struct task_struct *wake; +/* The private info the thread maintains about the guest. */ +struct lguest +{ + struct lguest_data __user *lguest_data; + struct lg_cpu cpus[NR_CPUS]; + unsigned int nr_cpus; + + u32 pfn_limit; + /* This provides the offset to the base of guest-physical + * memory in the Launcher. */ + void __user *mem_base; + unsigned long kernel_address; + + struct pgdir pgdirs[4]; unsigned long noirq_start, noirq_end; - int dma_is_pending; - unsigned long pending_dma; /* struct lguest_dma */ - unsigned long pending_key; /* address they're sending to */ unsigned int stack_pages; u32 tsc_khz; - struct lguest_dma_info dma[LGUEST_MAX_DMA]; - /* Dead? */ const char *dead; - - /* The GDT entries copied into lguest_ro_state when running. */ - struct desc_struct gdt[GDT_ENTRIES]; - - /* The IDT entries: some copied into lguest_ro_state when running. */ - struct desc_struct idt[FIRST_EXTERNAL_VECTOR+LGUEST_IRQS]; - struct desc_struct syscall_idt; - - /* Virtual clock device */ - struct hrtimer hrt; - - /* Pending virtual interrupts */ - DECLARE_BITMAP(irqs_pending, LGUEST_IRQS); }; -extern struct lguest lguests[]; extern struct mutex lguest_lock; /* core.c: */ -u32 lgread_u32(struct lguest *lg, unsigned long addr); -void lgwrite_u32(struct lguest *lg, unsigned long addr, u32 val); -void lgread(struct lguest *lg, void *buf, unsigned long addr, unsigned len); -void lgwrite(struct lguest *lg, unsigned long, const void *buf, unsigned len); -int find_free_guest(void); int lguest_address_ok(const struct lguest *lg, unsigned long addr, unsigned long len); -int run_guest(struct lguest *lg, unsigned long __user *user); +void __lgread(struct lg_cpu *, void *, unsigned long, unsigned); +void __lgwrite(struct lg_cpu *, unsigned long, const void *, unsigned); +/*H:035 Using memory-copy operations like that is usually inconvient, so we + * have the following helper macros which read and write a specific type (often + * an unsigned long). + * + * This reads into a variable of the given type then returns that. */ +#define lgread(cpu, addr, type) \ + ({ type _v; __lgread((cpu), &_v, (addr), sizeof(_v)); _v; }) + +/* This checks that the variable is of the given type, then writes it out. */ +#define lgwrite(cpu, addr, type, val) \ + do { \ + typecheck(type, val); \ + __lgwrite((cpu), (addr), &(val), sizeof(val)); \ + } while(0) +/* (end of memory access helper routines) :*/ + +int run_guest(struct lg_cpu *cpu, unsigned long __user *user); + +/* Helper macros to obtain the first 12 or the last 20 bits, this is only the + * first step in the migration to the kernel types. pte_pfn is already defined + * in the kernel. */ +#define pgd_flags(x) (pgd_val(x) & ~PAGE_MASK) +#define pte_flags(x) (pte_val(x) & ~PAGE_MASK) +#define pgd_pfn(x) (pgd_val(x) >> PAGE_SHIFT) /* interrupts_and_traps.c: */ -void maybe_do_interrupt(struct lguest *lg); -int deliver_trap(struct lguest *lg, unsigned int num); -void load_guest_idt_entry(struct lguest *lg, unsigned int i, u32 low, u32 hi); -void guest_set_stack(struct lguest *lg, u32 seg, u32 esp, unsigned int pages); -void pin_stack_pages(struct lguest *lg); +void maybe_do_interrupt(struct lg_cpu *cpu); +int deliver_trap(struct lg_cpu *cpu, unsigned int num); +void load_guest_idt_entry(struct lg_cpu *cpu, unsigned int i, + u32 low, u32 hi); +void guest_set_stack(struct lg_cpu *cpu, u32 seg, u32 esp, unsigned int pages); +void pin_stack_pages(struct lg_cpu *cpu); void setup_default_idt_entries(struct lguest_ro_state *state, const unsigned long *def); -void copy_traps(const struct lguest *lg, struct desc_struct *idt, +void copy_traps(const struct lg_cpu *cpu, struct desc_struct *idt, const unsigned long *def); -void guest_set_clockevent(struct lguest *lg, unsigned long delta); -void init_clockdev(struct lguest *lg); +void guest_set_clockevent(struct lg_cpu *cpu, unsigned long delta); +void init_clockdev(struct lg_cpu *cpu); +bool check_syscall_vector(struct lguest *lg); +int init_interrupts(void); +void free_interrupts(void); /* segments.c: */ void setup_default_gdt_entries(struct lguest_ro_state *state); -void setup_guest_gdt(struct lguest *lg); -void load_guest_gdt(struct lguest *lg, unsigned long table, u32 num); -void guest_load_tls(struct lguest *lg, unsigned long tls_array); -void copy_gdt(const struct lguest *lg, struct desc_struct *gdt); -void copy_gdt_tls(const struct lguest *lg, struct desc_struct *gdt); +void setup_guest_gdt(struct lg_cpu *cpu); +void load_guest_gdt(struct lg_cpu *cpu, unsigned long table, u32 num); +void guest_load_tls(struct lg_cpu *cpu, unsigned long tls_array); +void copy_gdt(const struct lg_cpu *cpu, struct desc_struct *gdt); +void copy_gdt_tls(const struct lg_cpu *cpu, struct desc_struct *gdt); /* page_tables.c: */ int init_guest_pagetable(struct lguest *lg, unsigned long pgtable); void free_guest_pagetable(struct lguest *lg); -void guest_new_pagetable(struct lguest *lg, unsigned long pgtable); -void guest_set_pmd(struct lguest *lg, unsigned long cr3, u32 i); -void guest_pagetable_clear_all(struct lguest *lg); -void guest_pagetable_flush_user(struct lguest *lg); -void guest_set_pte(struct lguest *lg, unsigned long cr3, - unsigned long vaddr, gpte_t val); -void map_switcher_in_guest(struct lguest *lg, struct lguest_pages *pages); -int demand_page(struct lguest *info, unsigned long cr2, int errcode); -void pin_page(struct lguest *lg, unsigned long vaddr); +void guest_new_pagetable(struct lg_cpu *cpu, unsigned long pgtable); +void guest_set_pmd(struct lguest *lg, unsigned long gpgdir, u32 i); +void guest_pagetable_clear_all(struct lg_cpu *cpu); +void guest_pagetable_flush_user(struct lg_cpu *cpu); +void guest_set_pte(struct lg_cpu *cpu, unsigned long gpgdir, + unsigned long vaddr, pte_t val); +void map_switcher_in_guest(struct lg_cpu *cpu, struct lguest_pages *pages); +int demand_page(struct lg_cpu *cpu, unsigned long cr2, int errcode); +void pin_page(struct lg_cpu *cpu, unsigned long vaddr); +unsigned long guest_pa(struct lg_cpu *cpu, unsigned long vaddr); +void page_table_guest_data_init(struct lg_cpu *cpu); + +/* /core.c: */ +void lguest_arch_host_init(void); +void lguest_arch_host_fini(void); +void lguest_arch_run_guest(struct lg_cpu *cpu); +void lguest_arch_handle_trap(struct lg_cpu *cpu); +int lguest_arch_init_hypercalls(struct lg_cpu *cpu); +int lguest_arch_do_hcall(struct lg_cpu *cpu, struct hcall_args *args); +void lguest_arch_setup_regs(struct lg_cpu *cpu, unsigned long start); + +/* /switcher.S: */ +extern char start_switcher_text[], end_switcher_text[], switch_to_guest[]; /* lguest_user.c: */ int lguest_device_init(void); void lguest_device_remove(void); -/* io.c: */ -void lguest_io_init(void); -int bind_dma(struct lguest *lg, - unsigned long key, unsigned long udma, u16 numdmas, u8 interrupt); -void send_dma(struct lguest *info, unsigned long key, unsigned long udma); -void release_all_dma(struct lguest *lg); -unsigned long get_dma_buffer(struct lguest *lg, unsigned long key, - unsigned long *interrupt); - /* hypercalls.c: */ -void do_hypercalls(struct lguest *lg); -void write_timestamp(struct lguest *lg); +void do_hypercalls(struct lg_cpu *cpu); +void write_timestamp(struct lg_cpu *cpu); /*L:035 * Let's step aside for the moment, to study one important routine that's used * widely in the Host code. * - * There are many cases where the Guest does something invalid, like pass crap + * There are many cases where the Guest can do something invalid, like pass crap * to a hypercall. Since only the Guest kernel can make hypercalls, it's quite * acceptable to simply terminate the Guest and give the Launcher a nicely * formatted reason. It's also simpler for the Guest itself, which doesn't @@ -282,19 +224,15 @@ void write_timestamp(struct lguest *lg); * Like any macro which uses an "if", it is safely wrapped in a run-once "do { * } while(0)". */ -#define kill_guest(lg, fmt...) \ +#define kill_guest(cpu, fmt...) \ do { \ - if (!(lg)->dead) { \ - (lg)->dead = kasprintf(GFP_ATOMIC, fmt); \ - if (!(lg)->dead) \ - (lg)->dead = ERR_PTR(-ENOMEM); \ + if (!(cpu)->lg->dead) { \ + (cpu)->lg->dead = kasprintf(GFP_ATOMIC, fmt); \ + if (!(cpu)->lg->dead) \ + (cpu)->lg->dead = ERR_PTR(-ENOMEM); \ } \ } while(0) /* (End of aside) :*/ -static inline unsigned long guest_pa(struct lguest *lg, unsigned long vaddr) -{ - return vaddr - lg->page_offset; -} #endif /* __ASSEMBLY__ */ #endif /* _LGUEST_H */