lguest: make registers per-vcpu

This is the most obvious per-vcpu field: registers.

So this patch moves it from struct lguest to struct vcpu,
and patch the places in which they are used, accordingly
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

drivers/lguest/interrupts_and_traps.c

@@ -70,7 +70,7 @@ static void set_guest_interrupt(struct lg_cpu *cpu, u32 lo, u32 hi, int has_err)
 	/* 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
 	 * userspace.  We check the privilege level to find out. */
-	if ((lg->regs->ss&0x3) != GUEST_PL) {
+	if ((cpu->regs->ss&0x3) != GUEST_PL) {
 		/* The Guest told us their kernel stack with the SET_STACK
 		 * hypercall: both the virtual address and the segment */
 		virtstack = lg->esp1;
@@ -81,12 +81,12 @@ static void set_guest_interrupt(struct lg_cpu *cpu, u32 lo, u32 hi, int has_err)
 		 * 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, lg->regs->ss);
-		push_guest_stack(lg, &gstack, lg->regs->esp);
+		push_guest_stack(lg, &gstack, cpu->regs->ss);
+		push_guest_stack(lg, &gstack, cpu->regs->esp);
 	} else {
 		/* We're staying on the same Guest (kernel) stack. */
-		virtstack = lg->regs->esp;
-		ss = lg->regs->ss;
+		virtstack = cpu->regs->esp;
+		ss = cpu->regs->ss;
 
 		origstack = gstack = guest_pa(lg, virtstack);
 	}
@@ -95,7 +95,7 @@ static void set_guest_interrupt(struct lg_cpu *cpu, u32 lo, u32 hi, int has_err)
 	 * the "Interrupt Flag" bit is always set.  We copy that bit from the
 	 * Guest's "irq_enabled" field into the eflags word: we saw the Guest
 	 * copy it back in "lguest_iret". */
-	eflags = lg->regs->eflags;
+	eflags = cpu->regs->eflags;
 	if (get_user(irq_enable, &lg->lguest_data->irq_enabled) == 0
 	    && !(irq_enable & X86_EFLAGS_IF))
 		eflags &= ~X86_EFLAGS_IF;
@@ -104,19 +104,19 @@ static void set_guest_interrupt(struct lg_cpu *cpu, u32 lo, u32 hi, int has_err)
 	 * "eflags" word, the old code segment, and the old instruction
 	 * pointer. */
 	push_guest_stack(lg, &gstack, eflags);
-	push_guest_stack(lg, &gstack, lg->regs->cs);
-	push_guest_stack(lg, &gstack, lg->regs->eip);
+	push_guest_stack(lg, &gstack, cpu->regs->cs);
+	push_guest_stack(lg, &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, lg->regs->errcode);
+		push_guest_stack(lg, &gstack, cpu->regs->errcode);
 
 	/* Now we've pushed all the old state, we change the stack, the code
 	 * segment and the address to execute. */
-	lg->regs->ss = ss;
-	lg->regs->esp = virtstack + (gstack - origstack);
-	lg->regs->cs = (__KERNEL_CS|GUEST_PL);
-	lg->regs->eip = idt_address(lo, hi);
+	cpu->regs->ss = ss;
+	cpu->regs->esp = virtstack + (gstack - origstack);
+	cpu->regs->cs = (__KERNEL_CS|GUEST_PL);
+	cpu->regs->eip = idt_address(lo, hi);
 
 	/* There are two kinds of interrupt handlers: 0xE is an "interrupt
 	 * gate" which expects interrupts to be disabled on entry. */
@@ -157,7 +157,7 @@ void maybe_do_interrupt(struct lg_cpu *cpu)
 
 	/* They may be in the middle of an iret, where they asked us never to
 	 * deliver interrupts. */
-	if (lg->regs->eip >= lg->noirq_start && lg->regs->eip < lg->noirq_end)
+	if (cpu->regs->eip >= lg->noirq_start && cpu->regs->eip < lg->noirq_end)
 		return;
 
 	/* If they're halted, interrupts restart them. */

drivers/lguest/lg.h

@@ -44,6 +44,10 @@ struct lg_cpu {
 	unsigned int id;
 	struct lguest *lg;
 
+	/* At end of a page shared mapped over lguest_pages in guest.  */
+	unsigned long regs_page;
+	struct lguest_regs *regs;
+
 	/* If a hypercall was asked for, this points to the arguments. */
 	struct hcall_args *hcall;
 	u32 next_hcall;
@@ -58,9 +62,6 @@ struct lg_cpu {
 /* 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 task_struct *tsk;
 	struct mm_struct *mm; 	/* == tsk->mm, but that becomes NULL on exit */
@@ -181,7 +182,7 @@ 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 lguest *lg, unsigned long start);
+void lguest_arch_setup_regs(struct lg_cpu *cpu, unsigned long start);
 
 /* <arch>/switcher.S: */
 extern char start_switcher_text[], end_switcher_text[], switch_to_guest[];

drivers/lguest/lguest_user.c

@@ -106,6 +106,19 @@ static int lg_cpu_start(struct lg_cpu *cpu, unsigned id, unsigned long start_ip)
 	cpu->lg->nr_cpus++;
 	init_clockdev(cpu);
 
+	/* We need a complete page for the Guest registers: they are accessible
+	 * to the Guest and we can only grant it access to whole pages. */
+	cpu->regs_page = get_zeroed_page(GFP_KERNEL);
+	if (!cpu->regs_page)
+		return -ENOMEM;
+
+	/* We actually put the registers at the bottom of the page. */
+	cpu->regs = (void *)cpu->regs_page + PAGE_SIZE - sizeof(*cpu->regs);
+
+	/* Now we initialize the Guest's registers, handing it the start
+	 * address. */
+	lguest_arch_setup_regs(cpu, start_ip);
+
 	return 0;
 }
 
@@ -160,16 +173,6 @@ static int initialize(struct file *file, const unsigned long __user *input)
 	if (err)
 		goto release_guest;
 
-	/* We need a complete page for the Guest registers: they are accessible
-	 * to the Guest and we can only grant it access to whole pages. */
-	lg->regs_page = get_zeroed_page(GFP_KERNEL);
-	if (!lg->regs_page) {
-		err = -ENOMEM;
-		goto release_guest;
-	}
-	/* We actually put the registers at the bottom of the page. */
-	lg->regs = (void *)lg->regs_page + PAGE_SIZE - sizeof(*lg->regs);
-
 	/* Initialize the Guest's shadow page tables, using the toplevel
 	 * address the Launcher gave us.  This allocates memory, so can
 	 * fail. */
@@ -177,10 +180,6 @@ static int initialize(struct file *file, const unsigned long __user *input)
 	if (err)
 		goto free_regs;
 
-	/* Now we initialize the Guest's registers, handing it the start
-	 * address. */
-	lguest_arch_setup_regs(lg, args[3]);
-
 	/* We keep a pointer to the Launcher task (ie. current task) for when
 	 * other Guests want to wake this one (inter-Guest I/O). */
 	lg->tsk = current;
@@ -205,7 +204,8 @@ static int initialize(struct file *file, const unsigned long __user *input)
 	return sizeof(args);
 
 free_regs:
-	free_page(lg->regs_page);
+	/* FIXME: This should be in free_vcpu */
+	free_page(lg->cpus[0].regs_page);
 release_guest:
 	kfree(lg);
 unlock:
@@ -280,9 +280,12 @@ static int close(struct inode *inode, struct file *file)
 	/* We need the big lock, to protect from inter-guest I/O and other
 	 * Launchers initializing guests. */
 	mutex_lock(&lguest_lock);
-	for (i = 0; i < lg->nr_cpus; i++)
+	for (i = 0; i < lg->nr_cpus; i++) {
 		/* Cancels the hrtimer set via LHCALL_SET_CLOCKEVENT. */
 		hrtimer_cancel(&lg->cpus[i].hrt);
+		/* We can free up the register page we allocated. */
+		free_page(lg->cpus[i].regs_page);
+	}
 	/* Free up the shadow page tables for the Guest. */
 	free_guest_pagetable(lg);
 	/* Now all the memory cleanups are done, it's safe to release the
@@ -292,8 +295,6 @@ static int close(struct inode *inode, struct file *file)
 	 * kmalloc()ed string, either of which is ok to hand to kfree(). */
 	if (!IS_ERR(lg->dead))
 		kfree(lg->dead);
-	/* We can free up the register page we allocated. */
-	free_page(lg->regs_page);
 	/* We clear the entire structure, which also marks it as free for the
 	 * next user. */
 	memset(lg, 0, sizeof(*lg));

drivers/lguest/page_tables.c

@@ -640,6 +640,7 @@ void map_switcher_in_guest(struct lg_cpu *cpu, struct lguest_pages *pages)
 	pte_t *switcher_pte_page = __get_cpu_var(switcher_pte_pages);
 	pgd_t switcher_pgd;
 	pte_t regs_pte;
+	unsigned long pfn;
 
 	/* Make the last PGD entry for this Guest point to the Switcher's PTE
 	 * page for this CPU (with appropriate flags). */
@@ -654,7 +655,8 @@ void map_switcher_in_guest(struct lg_cpu *cpu, struct lguest_pages *pages)
 	 * CPU's "struct lguest_pages": if we make sure the Guest's register
 	 * page is already mapped there, we don't have to copy them out
 	 * again. */
-	regs_pte = pfn_pte (__pa(lg->regs_page) >> PAGE_SHIFT, __pgprot(_PAGE_KERNEL));
+	pfn = __pa(cpu->regs_page) >> PAGE_SHIFT;
+	regs_pte = pfn_pte(pfn, __pgprot(_PAGE_KERNEL));
 	switcher_pte_page[(unsigned long)pages/PAGE_SIZE%PTRS_PER_PTE] = regs_pte;
 }
 /*:*/

drivers/lguest/x86/core.c

@@ -127,7 +127,7 @@ static void run_guest_once(struct lg_cpu *cpu, struct lguest_pages *pages)
 	/* Set the trap number to 256 (impossible value).  If we fault while
 	 * switching to the Guest (bad segment registers or bug), this will
 	 * cause us to abort the Guest. */
-	lg->regs->trapnum = 256;
+	cpu->regs->trapnum = 256;
 
 	/* Now: we push the "eflags" register on the stack, then do an "lcall".
 	 * This is how we change from using the kernel code segment to using
@@ -195,11 +195,11 @@ void lguest_arch_run_guest(struct lg_cpu *cpu)
 	 * bad virtual address.  We have to grab this now, because once we
 	 * re-enable interrupts an interrupt could fault and thus overwrite
 	 * cr2, or we could even move off to a different CPU. */
-	if (lg->regs->trapnum == 14)
+	if (cpu->regs->trapnum == 14)
 		lg->arch.last_pagefault = read_cr2();
 	/* Similarly, if we took a trap because the Guest used the FPU,
 	 * we have to restore the FPU it expects to see. */
-	else if (lg->regs->trapnum == 7)
+	else if (cpu->regs->trapnum == 7)
 		math_state_restore();
 
 	/* Restore SYSENTER if it's supposed to be on. */
@@ -225,12 +225,12 @@ static int emulate_insn(struct lg_cpu *cpu)
 	unsigned int insnlen = 0, in = 0, shift = 0;
 	/* The eip contains the *virtual* address of the Guest's instruction:
 	 * guest_pa just subtracts the Guest's page_offset. */
-	unsigned long physaddr = guest_pa(lg, lg->regs->eip);
+	unsigned long physaddr = guest_pa(lg, cpu->regs->eip);
 
 	/* This must be the Guest kernel trying to do something, not userspace!
 	 * The bottom two bits of the CS segment register are the privilege
 	 * level. */
-	if ((lg->regs->cs & 3) != GUEST_PL)
+	if ((cpu->regs->cs & 3) != GUEST_PL)
 		return 0;
 
 	/* Decoding x86 instructions is icky. */
@@ -273,12 +273,12 @@ static int emulate_insn(struct lg_cpu *cpu)
 	if (in) {
 		/* Lower bit tells is whether it's a 16 or 32 bit access */
 		if (insn & 0x1)
-			lg->regs->eax = 0xFFFFFFFF;
+			cpu->regs->eax = 0xFFFFFFFF;
 		else
-			lg->regs->eax |= (0xFFFF << shift);
+			cpu->regs->eax |= (0xFFFF << shift);
 	}
 	/* Finally, we've "done" the instruction, so move past it. */
-	lg->regs->eip += insnlen;
+	cpu->regs->eip += insnlen;
 	/* Success! */
 	return 1;
 }
@@ -287,12 +287,12 @@ static int emulate_insn(struct lg_cpu *cpu)
 void lguest_arch_handle_trap(struct lg_cpu *cpu)
 {
 	struct lguest *lg = cpu->lg;
-	switch (lg->regs->trapnum) {
+	switch (cpu->regs->trapnum) {
 	case 13: /* We've intercepted a General Protection Fault. */
 		/* Check if this was one of those annoying IN or OUT
 		 * instructions which we need to emulate.  If so, we just go
 		 * back into the Guest after we've done it. */
-		if (lg->regs->errcode == 0) {
+		if (cpu->regs->errcode == 0) {
 			if (emulate_insn(cpu))
 				return;
 		}
@@ -307,7 +307,7 @@ void lguest_arch_handle_trap(struct lg_cpu *cpu)
 		 *
 		 * The errcode tells whether this was a read or a write, and
 		 * whether kernel or userspace code. */
-		if (demand_page(lg, lg->arch.last_pagefault, lg->regs->errcode))
+		if (demand_page(lg, lg->arch.last_pagefault, cpu->regs->errcode))
 			return;
 
 		/* OK, it's really not there (or not OK): the Guest needs to
@@ -338,19 +338,19 @@ void lguest_arch_handle_trap(struct lg_cpu *cpu)
 	case LGUEST_TRAP_ENTRY:
 		/* Our 'struct hcall_args' maps directly over our regs: we set
 		 * up the pointer now to indicate a hypercall is pending. */
-		cpu->hcall = (struct hcall_args *)lg->regs;
+		cpu->hcall = (struct hcall_args *)cpu->regs;
 		return;
 	}
 
 	/* We didn't handle the trap, so it needs to go to the Guest. */
-	if (!deliver_trap(cpu, lg->regs->trapnum))
+	if (!deliver_trap(cpu, cpu->regs->trapnum))
 		/* 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)",
-			   lg->regs->trapnum, lg->regs->eip,
-			   lg->regs->trapnum == 14 ? lg->arch.last_pagefault
-			   : lg->regs->errcode);
+			   cpu->regs->trapnum, cpu->regs->eip,
+			   cpu->regs->trapnum == 14 ? lg->arch.last_pagefault
+			   : cpu->regs->errcode);
 }
 
 /* Now we can look at each of the routines this calls, in increasing order of
@@ -557,9 +557,9 @@ int lguest_arch_init_hypercalls(struct lg_cpu *cpu)
  *
  * Most of the Guest's registers are left alone: we used get_zeroed_page() to
  * allocate the structure, so they will be 0. */
-void lguest_arch_setup_regs(struct lguest *lg, unsigned long start)
+void lguest_arch_setup_regs(struct lg_cpu *cpu, unsigned long start)
 {
-	struct lguest_regs *regs = lg->regs;
+	struct lguest_regs *regs = cpu->regs;
 
 	/* There are four "segment" registers which the Guest needs to boot:
 	 * The "code segment" register (cs) refers to the kernel code segment
@@ -586,5 +586,5 @@ void lguest_arch_setup_regs(struct lguest *lg, unsigned long start)
 
 	/* There are a couple of GDT entries the Guest expects when first
 	 * booting. */
-	setup_guest_gdt(lg);
+	setup_guest_gdt(cpu->lg);
 }