/* This routine copies memory from the Guest. Here we can see how useful the
* kill_lguest() routine we met in the Launcher can be: we return a random
* value (all zeroes) instead of needing to return an error. */
-void __lgread(struct lguest *lg, void *b, unsigned long addr, unsigned bytes)
+void __lgread(struct lg_cpu *cpu, void *b, unsigned long addr, unsigned bytes)
{
- if (!lguest_address_ok(lg, addr, bytes)
- || copy_from_user(b, lg->mem_base + addr, bytes) != 0) {
+ if (!lguest_address_ok(cpu->lg, addr, bytes)
+ || copy_from_user(b, cpu->lg->mem_base + addr, bytes) != 0) {
/* copy_from_user should do this, but as we rely on it... */
memset(b, 0, bytes);
- kill_guest(lg, "bad read address %#lx len %u", addr, bytes);
+ kill_guest(cpu, "bad read address %#lx len %u", addr, bytes);
}
}
/* This is the write (copy into guest) version. */
-void __lgwrite(struct lguest *lg, unsigned long addr, const void *b,
+void __lgwrite(struct lg_cpu *cpu, unsigned long addr, const void *b,
unsigned bytes)
{
- if (!lguest_address_ok(lg, addr, bytes)
- || copy_to_user(lg->mem_base + addr, b, bytes) != 0)
- kill_guest(lg, "bad write address %#lx len %u", addr, bytes);
+ if (!lguest_address_ok(cpu->lg, addr, bytes)
+ || copy_to_user(cpu->lg->mem_base + addr, b, bytes) != 0)
+ kill_guest(cpu, "bad write address %#lx len %u", addr, bytes);
}
/*:*/
* going around and around until something interesting happens. */
int run_guest(struct lg_cpu *cpu, unsigned long __user *user)
{
- struct lguest *lg = cpu->lg;
-
/* We stop running once the Guest is dead. */
- while (!lg->dead) {
+ while (!cpu->lg->dead) {
/* First we run any hypercalls the Guest wants done. */
if (cpu->hcall)
do_hypercalls(cpu);
/* Just make absolutely sure the Guest is still alive. One of
* those hypercalls could have been fatal, for example. */
- if (lg->dead)
+ if (cpu->lg->dead)
break;
/* If the Guest asked to be stopped, we sleep. The Guest's
lguest_arch_handle_trap(cpu);
}
- if (lg->dead == ERR_PTR(-ERESTART))
+ if (cpu->lg->dead == ERR_PTR(-ERESTART))
return -ERESTART;
/* The Guest is dead => "No such file or directory" */
return -ENOENT;
* Or gets killed. Or, in the case of LHCALL_CRASH, both. */
static void do_hcall(struct lg_cpu *cpu, struct hcall_args *args)
{
- struct lguest *lg = cpu->lg;
-
switch (args->arg0) {
case LHCALL_FLUSH_ASYNC:
/* This call does nothing, except by breaking out of the Guest
case LHCALL_LGUEST_INIT:
/* You can't get here unless you're already initialized. Don't
* do that. */
- kill_guest(lg, "already have lguest_data");
+ kill_guest(cpu, "already have lguest_data");
break;
case LHCALL_SHUTDOWN: {
/* Shutdown is such a trivial hypercall that we do it in four
char msg[128];
/* If the lgread fails, it will call kill_guest() itself; the
* kill_guest() with the message will be ignored. */
- __lgread(lg, msg, args->arg1, sizeof(msg));
+ __lgread(cpu, msg, args->arg1, sizeof(msg));
msg[sizeof(msg)-1] = '\0';
- kill_guest(lg, "CRASH: %s", msg);
+ kill_guest(cpu, "CRASH: %s", msg);
if (args->arg2 == LGUEST_SHUTDOWN_RESTART)
- lg->dead = ERR_PTR(-ERESTART);
+ cpu->lg->dead = ERR_PTR(-ERESTART);
break;
}
case LHCALL_FLUSH_TLB:
guest_set_stack(cpu, args->arg1, args->arg2, args->arg3);
break;
case LHCALL_SET_PTE:
- guest_set_pte(lg, args->arg1, args->arg2, __pte(args->arg3));
+ guest_set_pte(cpu, args->arg1, args->arg2, __pte(args->arg3));
break;
case LHCALL_SET_PMD:
- guest_set_pmd(lg, args->arg1, args->arg2);
+ guest_set_pmd(cpu->lg, args->arg1, args->arg2);
break;
case LHCALL_SET_CLOCKEVENT:
guest_set_clockevent(cpu, args->arg1);
default:
/* It should be an architecture-specific hypercall. */
if (lguest_arch_do_hcall(cpu, args))
- kill_guest(lg, "Bad hypercall %li\n", args->arg0);
+ kill_guest(cpu, "Bad hypercall %li\n", args->arg0);
}
}
/*:*/
{
unsigned int i;
u8 st[LHCALL_RING_SIZE];
- struct lguest *lg = cpu->lg;
/* For simplicity, we copy the entire call status array in at once. */
- if (copy_from_user(&st, &lg->lguest_data->hcall_status, sizeof(st)))
+ if (copy_from_user(&st, &cpu->lg->lguest_data->hcall_status, sizeof(st)))
return;
/* We process "struct lguest_data"s hcalls[] ring once. */
/* Copy the hypercall arguments into a local copy of
* the hcall_args struct. */
- if (copy_from_user(&args, &lg->lguest_data->hcalls[n],
+ if (copy_from_user(&args, &cpu->lg->lguest_data->hcalls[n],
sizeof(struct hcall_args))) {
- kill_guest(lg, "Fetching async hypercalls");
+ kill_guest(cpu, "Fetching async hypercalls");
break;
}
do_hcall(cpu, &args);
/* Mark the hypercall done. */
- if (put_user(0xFF, &lg->lguest_data->hcall_status[n])) {
- kill_guest(lg, "Writing result for async hypercall");
+ if (put_user(0xFF, &cpu->lg->lguest_data->hcall_status[n])) {
+ kill_guest(cpu, "Writing result for async hypercall");
break;
}
* Guest makes a hypercall, we end up here to set things up: */
static void initialize(struct lg_cpu *cpu)
{
- struct lguest *lg = cpu->lg;
/* You can't do anything until you're initialized. The Guest knows the
* rules, so we're unforgiving here. */
if (cpu->hcall->arg0 != LHCALL_LGUEST_INIT) {
- kill_guest(lg, "hypercall %li before INIT", cpu->hcall->arg0);
+ kill_guest(cpu, "hypercall %li before INIT", cpu->hcall->arg0);
return;
}
if (lguest_arch_init_hypercalls(cpu))
- kill_guest(lg, "bad guest page %p", lg->lguest_data);
+ kill_guest(cpu, "bad guest page %p", cpu->lg->lguest_data);
/* The Guest tells us where we're not to deliver interrupts by putting
* the range of addresses into "struct lguest_data". */
- if (get_user(lg->noirq_start, &lg->lguest_data->noirq_start)
- || get_user(lg->noirq_end, &lg->lguest_data->noirq_end))
- kill_guest(lg, "bad guest page %p", lg->lguest_data);
+ if (get_user(cpu->lg->noirq_start, &cpu->lg->lguest_data->noirq_start)
+ || get_user(cpu->lg->noirq_end, &cpu->lg->lguest_data->noirq_end))
+ kill_guest(cpu, "bad guest page %p", cpu->lg->lguest_data);
/* We write the current time into the Guest's data page once so it can
* set its clock. */
- write_timestamp(lg);
+ write_timestamp(cpu);
/* page_tables.c will also do some setup. */
- page_table_guest_data_init(lg);
+ page_table_guest_data_init(cpu);
/* This is the one case where the above accesses might have been the
* first write to a Guest page. This may have caused a copy-on-write
/* This routine supplies the Guest with time: it's used for wallclock time at
* initial boot and as a rough time source if the TSC isn't available. */
-void write_timestamp(struct lguest *lg)
+void write_timestamp(struct lg_cpu *cpu)
{
struct timespec now;
ktime_get_real_ts(&now);
- if (copy_to_user(&lg->lguest_data->time, &now, sizeof(struct timespec)))
- kill_guest(lg, "Writing timestamp");
+ if (copy_to_user(&cpu->lg->lguest_data->time,
+ &now, sizeof(struct timespec)))
+ kill_guest(cpu, "Writing timestamp");
}
/* We need a helper to "push" a value onto the Guest's stack, since that's a
* big part of what delivering an interrupt does. */
-static void push_guest_stack(struct lguest *lg, unsigned long *gstack, u32 val)
+static void push_guest_stack(struct lg_cpu *cpu, unsigned long *gstack, u32 val)
{
/* Stack grows upwards: move stack then write value. */
*gstack -= 4;
- lgwrite(lg, *gstack, u32, val);
+ lgwrite(cpu, *gstack, u32, val);
}
/*H:210 The set_guest_interrupt() routine actually delivers the interrupt or
unsigned long gstack, origstack;
u32 eflags, ss, irq_enable;
unsigned long virtstack;
- struct lguest *lg = cpu->lg;
/* There are two cases for interrupts: one where the Guest is already
* in the kernel, and a more complex one where the Guest is in
* stack: when the Guest does an "iret" back from the interrupt
* handler the CPU will notice they're dropping privilege
* levels and expect these here. */
- push_guest_stack(lg, &gstack, cpu->regs->ss);
- push_guest_stack(lg, &gstack, cpu->regs->esp);
+ push_guest_stack(cpu, &gstack, cpu->regs->ss);
+ push_guest_stack(cpu, &gstack, cpu->regs->esp);
} else {
/* We're staying on the same Guest (kernel) stack. */
virtstack = cpu->regs->esp;
* Guest's "irq_enabled" field into the eflags word: we saw the Guest
* copy it back in "lguest_iret". */
eflags = cpu->regs->eflags;
- if (get_user(irq_enable, &lg->lguest_data->irq_enabled) == 0
+ if (get_user(irq_enable, &cpu->lg->lguest_data->irq_enabled) == 0
&& !(irq_enable & X86_EFLAGS_IF))
eflags &= ~X86_EFLAGS_IF;
/* An interrupt is expected to push three things on the stack: the old
* "eflags" word, the old code segment, and the old instruction
* pointer. */
- push_guest_stack(lg, &gstack, eflags);
- push_guest_stack(lg, &gstack, cpu->regs->cs);
- push_guest_stack(lg, &gstack, cpu->regs->eip);
+ push_guest_stack(cpu, &gstack, eflags);
+ push_guest_stack(cpu, &gstack, cpu->regs->cs);
+ push_guest_stack(cpu, &gstack, cpu->regs->eip);
/* For the six traps which supply an error code, we push that, too. */
if (has_err)
- push_guest_stack(lg, &gstack, cpu->regs->errcode);
+ push_guest_stack(cpu, &gstack, cpu->regs->errcode);
/* Now we've pushed all the old state, we change the stack, the code
* segment and the address to execute. */
/* There are two kinds of interrupt handlers: 0xE is an "interrupt
* gate" which expects interrupts to be disabled on entry. */
if (idt_type(lo, hi) == 0xE)
- if (put_user(0, &lg->lguest_data->irq_enabled))
- kill_guest(lg, "Disabling interrupts");
+ if (put_user(0, &cpu->lg->lguest_data->irq_enabled))
+ kill_guest(cpu, "Disabling interrupts");
}
/*H:205
void maybe_do_interrupt(struct lg_cpu *cpu)
{
unsigned int irq;
- struct lguest *lg = cpu->lg;
DECLARE_BITMAP(blk, LGUEST_IRQS);
struct desc_struct *idt;
/* If the Guest hasn't even initialized yet, we can do nothing. */
- if (!lg->lguest_data)
+ if (!cpu->lg->lguest_data)
return;
/* Take our "irqs_pending" array and remove any interrupts the Guest
* wants blocked: the result ends up in "blk". */
- if (copy_from_user(&blk, lg->lguest_data->blocked_interrupts,
+ if (copy_from_user(&blk, cpu->lg->lguest_data->blocked_interrupts,
sizeof(blk)))
return;
/* They may be in the middle of an iret, where they asked us never to
* deliver interrupts. */
- if (cpu->regs->eip >= lg->noirq_start && cpu->regs->eip < lg->noirq_end)
+ if (cpu->regs->eip >= cpu->lg->noirq_start &&
+ (cpu->regs->eip < cpu->lg->noirq_end))
return;
/* If they're halted, interrupts restart them. */
if (cpu->halted) {
/* Re-enable interrupts. */
- if (put_user(X86_EFLAGS_IF, &lg->lguest_data->irq_enabled))
- kill_guest(lg, "Re-enabling interrupts");
+ if (put_user(X86_EFLAGS_IF, &cpu->lg->lguest_data->irq_enabled))
+ kill_guest(cpu, "Re-enabling interrupts");
cpu->halted = 0;
} else {
/* Otherwise we check if they have interrupts disabled. */
u32 irq_enabled;
- if (get_user(irq_enabled, &lg->lguest_data->irq_enabled))
+ if (get_user(irq_enabled, &cpu->lg->lguest_data->irq_enabled))
irq_enabled = 0;
if (!irq_enabled)
return;
* did this more often, but it can actually be quite slow: doing it
* here is a compromise which means at least it gets updated every
* timer interrupt. */
- write_timestamp(lg);
+ write_timestamp(cpu);
}
/*:*/
{
unsigned int i;
- struct lguest *lg = cpu->lg;
/* Depending on the CONFIG_4KSTACKS option, the Guest can have one or
* two pages of stack space. */
- for (i = 0; i < lg->stack_pages; i++)
+ for (i = 0; i < cpu->lg->stack_pages; i++)
/* The stack grows *upwards*, so the address we're given is the
* start of the page after the kernel stack. Subtract one to
* get back onto the first stack page, and keep subtracting to
/* You are not allowed have a stack segment with privilege level 0: bad
* Guest! */
if ((seg & 0x3) != GUEST_PL)
- kill_guest(cpu->lg, "bad stack segment %i", seg);
+ kill_guest(cpu, "bad stack segment %i", seg);
/* We only expect one or two stack pages. */
if (pages > 2)
- kill_guest(cpu->lg, "bad stack pages %u", pages);
+ kill_guest(cpu, "bad stack pages %u", pages);
/* Save where the stack is, and how many pages */
cpu->ss1 = seg;
cpu->esp1 = esp;
/*H:235 This is the routine which actually checks the Guest's IDT entry and
* transfers it into the entry in "struct lguest": */
-static void set_trap(struct lguest *lg, struct desc_struct *trap,
+static void set_trap(struct lg_cpu *cpu, struct desc_struct *trap,
unsigned int num, u32 lo, u32 hi)
{
u8 type = idt_type(lo, hi);
/* We only support interrupt and trap gates. */
if (type != 0xE && type != 0xF)
- kill_guest(lg, "bad IDT type %i", type);
+ kill_guest(cpu, "bad IDT type %i", type);
/* We only copy the handler address, present bit, privilege level and
* type. The privilege level controls where the trap can be triggered
/* Check that the Guest doesn't try to step outside the bounds. */
if (num >= ARRAY_SIZE(cpu->arch.idt))
- kill_guest(cpu->lg, "Setting idt entry %u", num);
+ kill_guest(cpu, "Setting idt entry %u", num);
else
- set_trap(cpu->lg, &cpu->arch.idt[num], num, lo, hi);
+ set_trap(cpu, &cpu->arch.idt[num], num, lo, hi);
}
/* The default entry for each interrupt points into the Switcher routines which
/* core.c: */
int lguest_address_ok(const struct lguest *lg,
unsigned long addr, unsigned long len);
-void __lgread(struct lguest *, void *, unsigned long, unsigned);
-void __lgwrite(struct lguest *, unsigned long, const void *, unsigned);
+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(lg, addr, type) \
- ({ type _v; __lgread((lg), &_v, (addr), sizeof(_v)); _v; })
+#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(lg, addr, type, val) \
+#define lgwrite(cpu, addr, type, val) \
do { \
typecheck(type, val); \
- __lgwrite((lg), (addr), &(val), sizeof(val)); \
+ __lgwrite((cpu), (addr), &(val), sizeof(val)); \
} while(0)
/* (end of memory access helper routines) :*/
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 lguest *lg, unsigned long gpgdir,
+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 lguest *lg);
+void page_table_guest_data_init(struct lg_cpu *cpu);
/* <arch>/core.c: */
void lguest_arch_host_init(void);
/* hypercalls.c: */
void do_hypercalls(struct lg_cpu *cpu);
-void write_timestamp(struct lguest *lg);
+void write_timestamp(struct lg_cpu *cpu);
/*L:035
* Let's step aside for the moment, to study one important routine that's used
* 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) :*/
* page directory entry (PGD) for that address. Since we keep track of several
* page tables, the "i" argument tells us which one we're interested in (it's
* usually the current one). */
-static pgd_t *spgd_addr(struct lguest *lg, u32 i, unsigned long vaddr)
+static pgd_t *spgd_addr(struct lg_cpu *cpu, u32 i, unsigned long vaddr)
{
unsigned int index = pgd_index(vaddr);
/* We kill any Guest trying to touch the Switcher addresses. */
if (index >= SWITCHER_PGD_INDEX) {
- kill_guest(lg, "attempt to access switcher pages");
+ kill_guest(cpu, "attempt to access switcher pages");
index = 0;
}
/* Return a pointer index'th pgd entry for the i'th page table. */
- return &lg->pgdirs[i].pgdir[index];
+ return &cpu->lg->pgdirs[i].pgdir[index];
}
/* This routine then takes the page directory entry returned above, which
* entry can be a little tricky. The flags are (almost) the same, but the
* Guest PTE contains a virtual page number: the CPU needs the real page
* number. */
-static pte_t gpte_to_spte(struct lguest *lg, pte_t gpte, int write)
+static pte_t gpte_to_spte(struct lg_cpu *cpu, pte_t gpte, int write)
{
unsigned long pfn, base, flags;
flags = (pte_flags(gpte) & ~_PAGE_GLOBAL);
/* The Guest's pages are offset inside the Launcher. */
- base = (unsigned long)lg->mem_base / PAGE_SIZE;
+ base = (unsigned long)cpu->lg->mem_base / PAGE_SIZE;
/* We need a temporary "unsigned long" variable to hold the answer from
* get_pfn(), because it returns 0xFFFFFFFF on failure, which wouldn't
* page, given the virtual number. */
pfn = get_pfn(base + pte_pfn(gpte), write);
if (pfn == -1UL) {
- kill_guest(lg, "failed to get page %lu", pte_pfn(gpte));
+ kill_guest(cpu, "failed to get page %lu", pte_pfn(gpte));
/* When we destroy the Guest, we'll go through the shadow page
* tables and release_pte() them. Make sure we don't think
* this one is valid! */
}
/*:*/
-static void check_gpte(struct lguest *lg, pte_t gpte)
+static void check_gpte(struct lg_cpu *cpu, pte_t gpte)
{
if ((pte_flags(gpte) & (_PAGE_PWT|_PAGE_PSE))
- || pte_pfn(gpte) >= lg->pfn_limit)
- kill_guest(lg, "bad page table entry");
+ || pte_pfn(gpte) >= cpu->lg->pfn_limit)
+ kill_guest(cpu, "bad page table entry");
}
-static void check_gpgd(struct lguest *lg, pgd_t gpgd)
+static void check_gpgd(struct lg_cpu *cpu, pgd_t gpgd)
{
- if ((pgd_flags(gpgd) & ~_PAGE_TABLE) || pgd_pfn(gpgd) >= lg->pfn_limit)
- kill_guest(lg, "bad page directory entry");
+ if ((pgd_flags(gpgd) & ~_PAGE_TABLE) ||
+ (pgd_pfn(gpgd) >= cpu->lg->pfn_limit))
+ kill_guest(cpu, "bad page directory entry");
}
/*H:330
unsigned long gpte_ptr;
pte_t gpte;
pte_t *spte;
- struct lguest *lg = cpu->lg;
/* First step: get the top-level Guest page table entry. */
- gpgd = lgread(lg, gpgd_addr(cpu, vaddr), pgd_t);
+ gpgd = lgread(cpu, gpgd_addr(cpu, vaddr), pgd_t);
/* Toplevel not present? We can't map it in. */
if (!(pgd_flags(gpgd) & _PAGE_PRESENT))
return 0;
/* Now look at the matching shadow entry. */
- spgd = spgd_addr(lg, cpu->cpu_pgd, vaddr);
+ spgd = spgd_addr(cpu, cpu->cpu_pgd, vaddr);
if (!(pgd_flags(*spgd) & _PAGE_PRESENT)) {
/* No shadow entry: allocate a new shadow PTE page. */
unsigned long ptepage = get_zeroed_page(GFP_KERNEL);
/* This is not really the Guest's fault, but killing it is
* simple for this corner case. */
if (!ptepage) {
- kill_guest(lg, "out of memory allocating pte page");
+ kill_guest(cpu, "out of memory allocating pte page");
return 0;
}
/* We check that the Guest pgd is OK. */
- check_gpgd(lg, gpgd);
+ check_gpgd(cpu, gpgd);
/* And we copy the flags to the shadow PGD entry. The page
* number in the shadow PGD is the page we just allocated. */
*spgd = __pgd(__pa(ptepage) | pgd_flags(gpgd));
/* OK, now we look at the lower level in the Guest page table: keep its
* address, because we might update it later. */
gpte_ptr = gpte_addr(gpgd, vaddr);
- gpte = lgread(lg, gpte_ptr, pte_t);
+ gpte = lgread(cpu, gpte_ptr, pte_t);
/* If this page isn't in the Guest page tables, we can't page it in. */
if (!(pte_flags(gpte) & _PAGE_PRESENT))
/* Check that the Guest PTE flags are OK, and the page number is below
* the pfn_limit (ie. not mapping the Launcher binary). */
- check_gpte(lg, gpte);
+ check_gpte(cpu, gpte);
/* Add the _PAGE_ACCESSED and (for a write) _PAGE_DIRTY flag */
gpte = pte_mkyoung(gpte);
/* If this is a write, we insist that the Guest page is writable (the
* final arg to gpte_to_spte()). */
if (pte_dirty(gpte))
- *spte = gpte_to_spte(lg, gpte, 1);
+ *spte = gpte_to_spte(cpu, gpte, 1);
else
/* If this is a read, don't set the "writable" bit in the page
* table entry, even if the Guest says it's writable. That way
* we will come back here when a write does actually occur, so
* we can update the Guest's _PAGE_DIRTY flag. */
- *spte = gpte_to_spte(lg, pte_wrprotect(gpte), 0);
+ *spte = gpte_to_spte(cpu, pte_wrprotect(gpte), 0);
/* Finally, we write the Guest PTE entry back: we've set the
* _PAGE_ACCESSED and maybe the _PAGE_DIRTY flags. */
- lgwrite(lg, gpte_ptr, pte_t, gpte);
+ lgwrite(cpu, gpte_ptr, pte_t, gpte);
/* The fault is fixed, the page table is populated, the mapping
* manipulated, the result returned and the code complete. A small
unsigned long flags;
/* Look at the current top level entry: is it present? */
- spgd = spgd_addr(cpu->lg, cpu->cpu_pgd, vaddr);
+ spgd = spgd_addr(cpu, cpu->cpu_pgd, vaddr);
if (!(pgd_flags(*spgd) & _PAGE_PRESENT))
return 0;
void pin_page(struct lg_cpu *cpu, unsigned long vaddr)
{
if (!page_writable(cpu, vaddr) && !demand_page(cpu, vaddr, 2))
- kill_guest(cpu->lg, "bad stack page %#lx", vaddr);
+ kill_guest(cpu, "bad stack page %#lx", vaddr);
}
/*H:450 If we chase down the release_pgd() code, it looks like this: */
pte_t gpte;
/* First step: get the top-level Guest page table entry. */
- gpgd = lgread(cpu->lg, gpgd_addr(cpu, vaddr), pgd_t);
+ gpgd = lgread(cpu, gpgd_addr(cpu, vaddr), pgd_t);
/* Toplevel not present? We can't map it in. */
if (!(pgd_flags(gpgd) & _PAGE_PRESENT))
- kill_guest(cpu->lg, "Bad address %#lx", vaddr);
+ kill_guest(cpu, "Bad address %#lx", vaddr);
- gpte = lgread(cpu->lg, gpte_addr(gpgd, vaddr), pte_t);
+ gpte = lgread(cpu, gpte_addr(gpgd, vaddr), pte_t);
if (!(pte_flags(gpte) & _PAGE_PRESENT))
- kill_guest(cpu->lg, "Bad address %#lx", vaddr);
+ kill_guest(cpu, "Bad address %#lx", vaddr);
return pte_pfn(gpte) * PAGE_SIZE | (vaddr & ~PAGE_MASK);
}
int *blank_pgdir)
{
unsigned int next;
- struct lguest *lg = cpu->lg;
/* We pick one entry at random to throw out. Choosing the Least
* Recently Used might be better, but this is easy. */
- next = random32() % ARRAY_SIZE(lg->pgdirs);
+ next = random32() % ARRAY_SIZE(cpu->lg->pgdirs);
/* If it's never been allocated at all before, try now. */
- if (!lg->pgdirs[next].pgdir) {
- lg->pgdirs[next].pgdir = (pgd_t *)get_zeroed_page(GFP_KERNEL);
+ if (!cpu->lg->pgdirs[next].pgdir) {
+ cpu->lg->pgdirs[next].pgdir =
+ (pgd_t *)get_zeroed_page(GFP_KERNEL);
/* If the allocation fails, just keep using the one we have */
- if (!lg->pgdirs[next].pgdir)
+ if (!cpu->lg->pgdirs[next].pgdir)
next = cpu->cpu_pgd;
else
/* This is a blank page, so there are no kernel
*blank_pgdir = 1;
}
/* Record which Guest toplevel this shadows. */
- lg->pgdirs[next].gpgdir = gpgdir;
+ cpu->lg->pgdirs[next].gpgdir = gpgdir;
/* Release all the non-kernel mappings. */
- flush_user_mappings(lg, next);
+ flush_user_mappings(cpu->lg, next);
return next;
}
void guest_new_pagetable(struct lg_cpu *cpu, unsigned long pgtable)
{
int newpgdir, repin = 0;
- struct lguest *lg = cpu->lg;
/* Look to see if we have this one already. */
- newpgdir = find_pgdir(lg, pgtable);
+ newpgdir = find_pgdir(cpu->lg, pgtable);
/* If not, we allocate or mug an existing one: if it's a fresh one,
* repin gets set to 1. */
- if (newpgdir == ARRAY_SIZE(lg->pgdirs))
+ if (newpgdir == ARRAY_SIZE(cpu->lg->pgdirs))
newpgdir = new_pgdir(cpu, pgtable, &repin);
/* Change the current pgd index to the new one. */
cpu->cpu_pgd = newpgdir;
* _PAGE_ACCESSED then we can put a read-only PTE entry in immediately, and if
* they set _PAGE_DIRTY then we can put a writable PTE entry in immediately.
*/
-static void do_set_pte(struct lguest *lg, int idx,
+static void do_set_pte(struct lg_cpu *cpu, int idx,
unsigned long vaddr, pte_t gpte)
{
/* Look up the matching shadow page directory entry. */
- pgd_t *spgd = spgd_addr(lg, idx, vaddr);
+ pgd_t *spgd = spgd_addr(cpu, idx, vaddr);
/* If the top level isn't present, there's no entry to update. */
if (pgd_flags(*spgd) & _PAGE_PRESENT) {
* as well put that entry they've given us in now. This shaves
* 10% off a copy-on-write micro-benchmark. */
if (pte_flags(gpte) & (_PAGE_DIRTY | _PAGE_ACCESSED)) {
- check_gpte(lg, gpte);
- *spte = gpte_to_spte(lg, gpte,
+ check_gpte(cpu, gpte);
+ *spte = gpte_to_spte(cpu, gpte,
pte_flags(gpte) & _PAGE_DIRTY);
} else
/* Otherwise kill it and we can demand_page() it in
*
* The benefit is that when we have to track a new page table, we can copy keep
* all the kernel mappings. This speeds up context switch immensely. */
-void guest_set_pte(struct lguest *lg,
+void guest_set_pte(struct lg_cpu *cpu,
unsigned long gpgdir, unsigned long vaddr, pte_t gpte)
{
/* Kernel mappings must be changed on all top levels. Slow, but
* doesn't happen often. */
- if (vaddr >= lg->kernel_address) {
+ if (vaddr >= cpu->lg->kernel_address) {
unsigned int i;
- for (i = 0; i < ARRAY_SIZE(lg->pgdirs); i++)
- if (lg->pgdirs[i].pgdir)
- do_set_pte(lg, i, vaddr, gpte);
+ for (i = 0; i < ARRAY_SIZE(cpu->lg->pgdirs); i++)
+ if (cpu->lg->pgdirs[i].pgdir)
+ do_set_pte(cpu, i, vaddr, gpte);
} else {
/* Is this page table one we have a shadow for? */
- int pgdir = find_pgdir(lg, gpgdir);
- if (pgdir != ARRAY_SIZE(lg->pgdirs))
+ int pgdir = find_pgdir(cpu->lg, gpgdir);
+ if (pgdir != ARRAY_SIZE(cpu->lg->pgdirs))
/* If so, do the update. */
- do_set_pte(lg, pgdir, vaddr, gpte);
+ do_set_pte(cpu, pgdir, vaddr, gpte);
}
}
}
/* When the Guest calls LHCALL_LGUEST_INIT we do more setup. */
-void page_table_guest_data_init(struct lguest *lg)
+void page_table_guest_data_init(struct lg_cpu *cpu)
{
/* We get the kernel address: above this is all kernel memory. */
- if (get_user(lg->kernel_address, &lg->lguest_data->kernel_address)
+ if (get_user(cpu->lg->kernel_address,
+ &cpu->lg->lguest_data->kernel_address)
/* We tell the Guest that it can't use the top 4MB of virtual
* addresses used by the Switcher. */
- || put_user(4U*1024*1024, &lg->lguest_data->reserve_mem)
- || put_user(lg->pgdirs[0].gpgdir, &lg->lguest_data->pgdir))
- kill_guest(lg, "bad guest page %p", lg->lguest_data);
+ || put_user(4U*1024*1024, &cpu->lg->lguest_data->reserve_mem)
+ || put_user(cpu->lg->pgdirs[0].gpgdir, &cpu->lg->lguest_data->pgdir))
+ kill_guest(cpu, "bad guest page %p", cpu->lg->lguest_data);
/* In flush_user_mappings() we loop from 0 to
* "pgd_index(lg->kernel_address)". This assumes it won't hit the
* Switcher mappings, so check that now. */
- if (pgd_index(lg->kernel_address) >= SWITCHER_PGD_INDEX)
- kill_guest(lg, "bad kernel address %#lx", lg->kernel_address);
+ if (pgd_index(cpu->lg->kernel_address) >= SWITCHER_PGD_INDEX)
+ kill_guest(cpu, "bad kernel address %#lx",
+ cpu->lg->kernel_address);
}
/* When a Guest dies, our cleanup is fairly simple. */
* We copy it from the Guest and tweak the entries. */
void load_guest_gdt(struct lg_cpu *cpu, unsigned long table, u32 num)
{
- struct lguest *lg = cpu->lg;
/* We assume the Guest has the same number of GDT entries as the
* Host, otherwise we'd have to dynamically allocate the Guest GDT. */
if (num > ARRAY_SIZE(cpu->arch.gdt))
- kill_guest(lg, "too many gdt entries %i", num);
+ kill_guest(cpu, "too many gdt entries %i", num);
/* We read the whole thing in, then fix it up. */
- __lgread(lg, cpu->arch.gdt, table, num * sizeof(cpu->arch.gdt[0]));
+ __lgread(cpu, cpu->arch.gdt, table, num * sizeof(cpu->arch.gdt[0]));
fixup_gdt_table(cpu, 0, ARRAY_SIZE(cpu->arch.gdt));
/* Mark that the GDT changed so the core knows it has to copy it again,
* even if the Guest is run on the same CPU. */
void guest_load_tls(struct lg_cpu *cpu, unsigned long gtls)
{
struct desc_struct *tls = &cpu->arch.gdt[GDT_ENTRY_TLS_MIN];
- struct lguest *lg = cpu->lg;
- __lgread(lg, tls, gtls, sizeof(*tls)*GDT_ENTRY_TLS_ENTRIES);
+ __lgread(cpu, tls, gtls, sizeof(*tls)*GDT_ENTRY_TLS_ENTRIES);
fixup_gdt_table(cpu, GDT_ENTRY_TLS_MIN, GDT_ENTRY_TLS_MAX+1);
/* Note that just the TLS entries have changed. */
cpu->changed |= CHANGED_GDT_TLS;
{
/* This is a dummy value we need for GCC's sake. */
unsigned int clobber;
- struct lguest *lg = cpu->lg;
/* Copy the guest-specific information into this CPU's "struct
* lguest_pages". */
* 0-th argument above, ie "a"). %ebx contains the
* physical address of the Guest's top-level page
* directory. */
- : "0"(pages), "1"(__pa(lg->pgdirs[cpu->cpu_pgd].pgdir))
+ : "0"(pages), "1"(__pa(cpu->lg->pgdirs[cpu->cpu_pgd].pgdir))
/* We tell gcc that all these registers could change,
* which means we don't have to save and restore them in
* the Switcher. */
* instructions and skip over it. We return true if we did. */
static int emulate_insn(struct lg_cpu *cpu)
{
- struct lguest *lg = cpu->lg;
u8 insn;
unsigned int insnlen = 0, in = 0, shift = 0;
/* The eip contains the *virtual* address of the Guest's instruction:
return 0;
/* Decoding x86 instructions is icky. */
- insn = lgread(lg, physaddr, u8);
+ insn = lgread(cpu, physaddr, u8);
/* 0x66 is an "operand prefix". It means it's using the upper 16 bits
of the eax register. */
shift = 16;
/* The instruction is 1 byte so far, read the next byte. */
insnlen = 1;
- insn = lgread(lg, physaddr + insnlen, u8);
+ insn = lgread(cpu, physaddr + insnlen, u8);
}
/* We can ignore the lower bit for the moment and decode the 4 opcodes
/*H:050 Once we've re-enabled interrupts, we look at why the Guest exited. */
void lguest_arch_handle_trap(struct lg_cpu *cpu)
{
- struct lguest *lg = cpu->lg;
switch (cpu->regs->trapnum) {
case 13: /* We've intercepted a General Protection Fault. */
/* Check if this was one of those annoying IN or OUT
* Note that if the Guest were really messed up, this could
* happen before it's done the LHCALL_LGUEST_INIT hypercall, so
* lg->lguest_data could be NULL */
- if (lg->lguest_data &&
- put_user(cpu->arch.last_pagefault, &lg->lguest_data->cr2))
- kill_guest(lg, "Writing cr2");
+ if (cpu->lg->lguest_data &&
+ put_user(cpu->arch.last_pagefault,
+ &cpu->lg->lguest_data->cr2))
+ kill_guest(cpu, "Writing cr2");
break;
case 7: /* We've intercepted a Device Not Available fault. */
/* If the Guest doesn't want to know, we already restored the
/* If the Guest doesn't have a handler (either it hasn't
* registered any yet, or it's one of the faults we don't let
* it handle), it dies with a cryptic error message. */
- kill_guest(lg, "unhandled trap %li at %#lx (%#lx)",
+ kill_guest(cpu, "unhandled trap %li at %#lx (%#lx)",
cpu->regs->trapnum, cpu->regs->eip,
cpu->regs->trapnum == 14 ? cpu->arch.last_pagefault
: cpu->regs->errcode);
int lguest_arch_init_hypercalls(struct lg_cpu *cpu)
{
u32 tsc_speed;
- struct lguest *lg = cpu->lg;
/* The pointer to the Guest's "struct lguest_data" is the only
* argument. We check that address now. */
- if (!lguest_address_ok(lg, cpu->hcall->arg1, sizeof(*lg->lguest_data)))
+ if (!lguest_address_ok(cpu->lg, cpu->hcall->arg1,
+ sizeof(*cpu->lg->lguest_data)))
return -EFAULT;
/* Having checked it, we simply set lg->lguest_data to point straight
* copy_to_user/from_user from now on, instead of lgread/write. I put
* this in to show that I'm not immune to writing stupid
* optimizations. */
- lg->lguest_data = lg->mem_base + cpu->hcall->arg1;
+ cpu->lg->lguest_data = cpu->lg->mem_base + cpu->hcall->arg1;
/* We insist that the Time Stamp Counter exist and doesn't change with
* cpu frequency. Some devious chip manufacturers decided that TSC
tsc_speed = tsc_khz;
else
tsc_speed = 0;
- if (put_user(tsc_speed, &lg->lguest_data->tsc_khz))
+ if (put_user(tsc_speed, &cpu->lg->lguest_data->tsc_khz))
return -EFAULT;
/* The interrupt code might not like the system call vector. */
- if (!check_syscall_vector(lg))
- kill_guest(lg, "bad syscall vector");
+ if (!check_syscall_vector(cpu->lg))
+ kill_guest(cpu, "bad syscall vector");
return 0;
}
projects / linux-2.6 / commitdiff