lguest: move the initial guest page table creation code to the host

This patch moves the initial guest page table creation code to the host,
so the launcher keeps working with PAE enabled configs.

Signed-off-by: Matias Zabaljauregui <zabaljauregui@gmail.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

6 files changed, 83 insertions, 81 deletions

diff --git a/Documentation/lguest/lguest.c b/Documentation/lguest/lguest.c
index aa2574ca94c7..f2dbbf3bdeab 100644
--- a/Documentation/lguest/lguest.c
+++ b/Documentation/lguest/lguest.c
@@ -481,51 +481,6 @@ static unsigned long load_initrd(const char *name, unsigned long mem)
 	/* We return the initrd size. */
 	return len;
 }
-
-/* Once we know how much memory we have we can construct simple linear page
- * tables which set virtual == physical which will get the Guest far enough
- * into the boot to create its own.
- *
- * We lay them out of the way, just below the initrd (which is why we need to
- * know its size here). */
-static unsigned long setup_pagetables(unsigned long mem,
-				      unsigned long initrd_size)
-{
-	unsigned long *pgdir, *linear;
-	unsigned int mapped_pages, i, linear_pages;
-	unsigned int ptes_per_page = getpagesize()/sizeof(void *);
-
-	mapped_pages = mem/getpagesize();
-
-	/* Each PTE page can map ptes_per_page pages: how many do we need? */
-	linear_pages = (mapped_pages + ptes_per_page-1)/ptes_per_page;
-
-	/* We put the toplevel page directory page at the top of memory. */
-	pgdir = from_guest_phys(mem) - initrd_size - getpagesize();
-
-	/* Now we use the next linear_pages pages as pte pages */
-	linear = (void *)pgdir - linear_pages*getpagesize();
-
-	/* Linear mapping is easy: put every page's address into the mapping in
-	 * order.  PAGE_PRESENT contains the flags Present, Writable and
-	 * Executable. */
-	for (i = 0; i < mapped_pages; i++)
-		linear[i] = ((i * getpagesize()) | PAGE_PRESENT);
-
-	/* The top level points to the linear page table pages above. */
-	for (i = 0; i < mapped_pages; i += ptes_per_page) {
-		pgdir[i/ptes_per_page]
-			= ((to_guest_phys(linear) + i*sizeof(void *))
-			   | PAGE_PRESENT);
-	}
-
-	verbose("Linear mapping of %u pages in %u pte pages at %#lx\n",
-		mapped_pages, linear_pages, to_guest_phys(linear));
-
-	/* We return the top level (guest-physical) address: the kernel needs
-	 * to know where it is. */
-	return to_guest_phys(pgdir);
-}
 /*:*/
 
 /* Simple routine to roll all the commandline arguments together with spaces
@@ -548,13 +503,13 @@ static void concat(char *dst, char *args[])
 
 /*L:185 This is where we actually tell the kernel to initialize the Guest.  We
  * saw the arguments it expects when we looked at initialize() in lguest_user.c:
- * the base of Guest "physical" memory, the top physical page to allow, the
- * top level pagetable and the entry point for the Guest. */
-static int tell_kernel(unsigned long pgdir, unsigned long start)
+ * the base of Guest "physical" memory, the top physical page to allow and the
+ * entry point for the Guest. */
+static int tell_kernel(unsigned long start)
 {
 	unsigned long args[] = { LHREQ_INITIALIZE,
 				 (unsigned long)guest_base,
-				 guest_limit / getpagesize(), pgdir, start };
+				 guest_limit / getpagesize(), start };
 	int fd;
 
 	verbose("Guest: %p - %p (%#lx)\n",
@@ -1941,7 +1896,7 @@ int main(int argc, char *argv[])
 {
 	/* Memory, top-level pagetable, code startpoint and size of the
 	 * (optional) initrd. */
-	unsigned long mem = 0, pgdir, start, initrd_size = 0;
+	unsigned long mem = 0, start, initrd_size = 0;
 	/* Two temporaries and the /dev/lguest file descriptor. */
 	int i, c, lguest_fd;
 	/* The boot information for the Guest. */
@@ -2040,9 +1995,6 @@ int main(int argc, char *argv[])
 		boot->hdr.type_of_loader = 0xFF;
 	}
 
-	/* Set up the initial linear pagetables, starting below the initrd. */
-	pgdir = setup_pagetables(mem, initrd_size);
-
 	/* The Linux boot header contains an "E820" memory map: ours is a
 	 * simple, single region. */
 	boot->e820_entries = 1;
@@ -2064,7 +2016,7 @@ int main(int argc, char *argv[])
 
 	/* We tell the kernel to initialize the Guest: this returns the open
 	 * /dev/lguest file descriptor. */
-	lguest_fd = tell_kernel(pgdir, start);
+	lguest_fd = tell_kernel(start);
 
 	/* We clone off a thread, which wakes the Launcher whenever one of the
 	 * input file descriptors needs attention.  We call this the Waker, and
diff --git a/arch/x86/lguest/i386_head.S b/arch/x86/lguest/i386_head.S
index 5c7cef34c9e7..10b9bd35a8ff 100644
--- a/arch/x86/lguest/i386_head.S
+++ b/arch/x86/lguest/i386_head.S
@@ -30,21 +30,6 @@ ENTRY(lguest_entry)
 	movl $lguest_data - __PAGE_OFFSET, %edx
 	int $LGUEST_TRAP_ENTRY
 
-	/* The Host put the toplevel pagetable in lguest_data.pgdir.  The movsl
-	 * instruction uses %esi implicitly as the source for the copy we're
-	 * about to do. */
-	movl lguest_data - __PAGE_OFFSET + LGUEST_DATA_pgdir, %esi
-
-	/* Copy first 32 entries of page directory to __PAGE_OFFSET entries.
-	 * This means the first 128M of kernel memory will be mapped at
-	 * PAGE_OFFSET where the kernel expects to run.  This will get it far
-	 * enough through boot to switch to its own pagetables. */
-	movl $32, %ecx
-	movl %esi, %edi
-	addl $((__PAGE_OFFSET >> 22) * 4), %edi
-	rep
-	movsl
-
 	/* Set up the initial stack so we can run C code. */
 	movl $(init_thread_union+THREAD_SIZE),%esp
 
diff --git a/drivers/lguest/lg.h b/drivers/lguest/lg.h
index 5faefeaf6790..f2c641e0bdde 100644
--- a/drivers/lguest/lg.h
+++ b/drivers/lguest/lg.h
@@ -164,7 +164,7 @@ 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);
+int init_guest_pagetable(struct lguest *lg);
 void free_guest_pagetable(struct lguest *lg);
 void guest_new_pagetable(struct lg_cpu *cpu, unsigned long pgtable);
 void guest_set_pmd(struct lguest *lg, unsigned long gpgdir, u32 i);
diff --git a/drivers/lguest/lguest_user.c b/drivers/lguest/lguest_user.c
index e73a000473cc..34bc017b8b3c 100644
--- a/drivers/lguest/lguest_user.c
+++ b/drivers/lguest/lguest_user.c
@@ -146,7 +146,7 @@ static int lg_cpu_start(struct lg_cpu *cpu, unsigned id, unsigned long start_ip)
 	return 0;
 }
 
-/*L:020 The initialization write supplies 4 pointer sized (32 or 64 bit)
+/*L:020 The initialization write supplies 3 pointer sized (32 or 64 bit)
  * values (in addition to the LHREQ_INITIALIZE value).  These are:
  *
  * base: The start of the Guest-physical memory inside the Launcher memory.
@@ -155,9 +155,6 @@ static int lg_cpu_start(struct lg_cpu *cpu, unsigned id, unsigned long start_ip)
  * allowed to access.  The Guest memory lives inside the Launcher, so it sets
  * this to ensure the Guest can only reach its own memory.
  *
- * pgdir: The (Guest-physical) address of the top of the initial Guest
- * pagetables (which are set up by the Launcher).
- *
  * start: The first instruction to execute ("eip" in x86-speak).
  */
 static int initialize(struct file *file, const unsigned long __user *input)
@@ -166,7 +163,7 @@ static int initialize(struct file *file, const unsigned long __user *input)
 	 * Guest. */
 	struct lguest *lg;
 	int err;
-	unsigned long args[4];
+	unsigned long args[3];
 
 	/* We grab the Big Lguest lock, which protects against multiple
 	 * simultaneous initializations. */
@@ -192,14 +189,14 @@ static int initialize(struct file *file, const unsigned long __user *input)
 	lg->mem_base = (void __user *)args[0];
 	lg->pfn_limit = args[1];
 
-	/* This is the first cpu (cpu 0) and it will start booting at args[3] */
-	err = lg_cpu_start(&lg->cpus[0], 0, args[3]);
+	/* This is the first cpu (cpu 0) and it will start booting at args[2] */
+	err = lg_cpu_start(&lg->cpus[0], 0, args[2]);
 	if (err)
 		goto release_guest;
 
 	/* Initialize the Guest's shadow page tables, using the toplevel
 	 * address the Launcher gave us.  This allocates memory, so can fail. */
-	err = init_guest_pagetable(lg, args[2]);
+	err = init_guest_pagetable(lg);
 	if (err)
 		goto free_regs;
 
diff --git a/drivers/lguest/page_tables.c b/drivers/lguest/page_tables.c
index 81d0c6053447..576a8318221c 100644
--- a/drivers/lguest/page_tables.c
+++ b/drivers/lguest/page_tables.c
@@ -14,6 +14,7 @@
 #include <linux/percpu.h>
 #include <asm/tlbflush.h>
 #include <asm/uaccess.h>
+#include <asm/bootparam.h>
 #include "lg.h"
 
 /*M:008 We hold reference to pages, which prevents them from being swapped.
@@ -581,15 +582,82 @@ void guest_set_pmd(struct lguest *lg, unsigned long gpgdir, u32 idx)
 		release_pgd(lg, lg->pgdirs[pgdir].pgdir + idx);
 }
 
+/* Once we know how much memory we have we can construct simple identity
+ * (which set virtual == physical) and linear mappings
+ * which will get the Guest far enough into the boot to create its own.
+ *
+ * We lay them out of the way, just below the initrd (which is why we need to
+ * know its size here). */
+static unsigned long setup_pagetables(struct lguest *lg,
+				      unsigned long mem,
+				      unsigned long initrd_size)
+{
+	pgd_t __user *pgdir;
+	pte_t __user *linear;
+	unsigned int mapped_pages, i, linear_pages, phys_linear;
+	unsigned long mem_base = (unsigned long)lg->mem_base;
+
+	/* We have mapped_pages frames to map, so we need
+	 * linear_pages page tables to map them. */
+	mapped_pages = mem / PAGE_SIZE;
+	linear_pages = (mapped_pages + PTRS_PER_PTE - 1) / PTRS_PER_PTE;
+
+	/* We put the toplevel page directory page at the top of memory. */
+	pgdir = (pgd_t *)(mem + mem_base - initrd_size - PAGE_SIZE);
+
+	/* Now we use the next linear_pages pages as pte pages */
+	linear = (void *)pgdir - linear_pages * PAGE_SIZE;
+
+	/* Linear mapping is easy: put every page's address into the
+	 * mapping in order. */
+	for (i = 0; i < mapped_pages; i++) {
+		pte_t pte;
+		pte = pfn_pte(i, __pgprot(_PAGE_PRESENT|_PAGE_RW|_PAGE_USER));
+		if (copy_to_user(&linear[i], &pte, sizeof(pte)) != 0)
+			return -EFAULT;
+	}
+
+	/* The top level points to the linear page table pages above.
+	 * We setup the identity and linear mappings here. */
+	phys_linear = (unsigned long)linear - mem_base;
+	for (i = 0; i < mapped_pages; i += PTRS_PER_PTE) {
+		pgd_t pgd;
+		pgd = __pgd((phys_linear + i * sizeof(pte_t)) |
+			    (_PAGE_PRESENT | _PAGE_RW | _PAGE_USER));
+
+		if (copy_to_user(&pgdir[i / PTRS_PER_PTE], &pgd, sizeof(pgd))
+		    || copy_to_user(&pgdir[pgd_index(PAGE_OFFSET)
+					   + i / PTRS_PER_PTE],
+				    &pgd, sizeof(pgd)))
+			return -EFAULT;
+	}
+
+	/* We return the top level (guest-physical) address: remember where
+	 * this is. */
+	return (unsigned long)pgdir - mem_base;
+}
+
 /*H:500 (vii) Setting up the page tables initially.
  *
  * When a Guest is first created, the Launcher tells us where the toplevel of
  * its first page table is.  We set some things up here: */
-int init_guest_pagetable(struct lguest *lg, unsigned long pgtable)
+int init_guest_pagetable(struct lguest *lg)
 {
+	u64 mem;
+	u32 initrd_size;
+	struct boot_params __user *boot = (struct boot_params *)lg->mem_base;
+
+	/* Get the Guest memory size and the ramdisk size from the boot header
+	 * located at lg->mem_base (Guest address 0). */
+	if (copy_from_user(&mem, &boot->e820_map[0].size, sizeof(mem))
+	    || get_user(initrd_size, &boot->hdr.ramdisk_size))
+		return -EFAULT;
+
 	/* We start on the first shadow page table, and give it a blank PGD
 	 * page. */
-	lg->pgdirs[0].gpgdir = pgtable;
+	lg->pgdirs[0].gpgdir = setup_pagetables(lg, mem, initrd_size);
+	if (IS_ERR_VALUE(lg->pgdirs[0].gpgdir))
+		return lg->pgdirs[0].gpgdir;
 	lg->pgdirs[0].pgdir = (pgd_t *)get_zeroed_page(GFP_KERNEL);
 	if (!lg->pgdirs[0].pgdir)
 		return -ENOMEM;
diff --git a/include/linux/lguest_launcher.h b/include/linux/lguest_launcher.h
index bd0eba760522..a53407a4165c 100644
--- a/include/linux/lguest_launcher.h
+++ b/include/linux/lguest_launcher.h
@@ -54,7 +54,7 @@ struct lguest_vqconfig {
 /* Write command first word is a request. */
 enum lguest_req
 {
-	LHREQ_INITIALIZE, /* + base, pfnlimit, pgdir, start */
+	LHREQ_INITIALIZE, /* + base, pfnlimit, start */
 	LHREQ_GETDMA, /* No longer used */
 	LHREQ_IRQ, /* + irq */
 	LHREQ_BREAK, /* + on/off flag (on blocks until someone does off) */