lguest: Patches for 2.6.24 (and patchbomb test)

diff -r bcbb99849fd9 Documentation/lguest/lguest.c
@@ -882,7 +882,7 @@ static u32 handle_block_output(int fd, c
 		 * of the block file (possibly extending it). */
 		if (off + len > device_len) {
 			/* Trim it back to the correct length */
-			ftruncate(dev->fd, device_len);
+			ftruncate64(dev->fd, device_len);
 			/* Die, bad Guest, die. */
 			errx(1, "Write past end %llu+%u", off, len);
 		}

@@ -46,7 +46,7 @@ typedef uint32_t u32;
 typedef uint16_t u16;
 typedef uint8_t u8;
 #include "../../include/linux/lguest_launcher.h"
-#include "../../include/asm-i386/e820.h"
+#include "../../include/asm/e820.h"
 /*:*/
 
 #define PAGE_PRESENT 0x7 	/* Present, RW, Execute */

@@ -159,7 +159,7 @@ static unsigned long entry_point(void *s
 		if (memcmp(p, "GenuineLguest", strlen("GenuineLguest")) == 0)
 			return (long)p + strlen("GenuineLguest") + page_offset;
 
-	err(1, "Is this image a genuine lguest?");
+	errx(1, "Is this image a genuine lguest?");
 }
 
 /* This routine takes an open vmlinux image, which is in ELF, and maps it into

@@ -858,7 +858,7 @@ static u32 handle_block_output(int fd, c
 	/* We first check that the read or write is within the length of the
 	 * block file. */
 	if (off >= device_len)
-		err(1, "Bad offset %llu vs %llu", off, device_len);
+		errx(1, "Bad offset %llu vs %llu", off, device_len);
 	/* Move to the right location in the block file.  This shouldn't fail,
 	 * but best to check. */
 	if (lseek64(dev->fd, off, SEEK_SET) != off)

@@ -243,6 +243,6 @@ void write_timestamp(struct lguest *lg)
 {
 	struct timespec now;
 	ktime_get_real_ts(&now);
-	if (put_user(now, &lg->lguest_data->time))
+	if (copy_to_user(&lg->lguest_data->time, &now, sizeof(struct timespec)))
 		kill_guest(lg, "Writing timestamp");
 }

@@ -45,8 +45,8 @@ typedef uint32_t u32;
 typedef uint32_t u32;
 typedef uint16_t u16;
 typedef uint8_t u8;
-#include "../../include/linux/lguest_launcher.h"
-#include "../../include/asm/e820.h"
+#include "linux/lguest_launcher.h"
+#include "asm/e820.h"
 /*:*/
 
 #define PAGE_PRESENT 0x7 	/* Present, RW, Execute */

@@ -1,6 +1,7 @@
 #ifndef _ASM_LGUEST_USER
 #define _ASM_LGUEST_USER
 /* Everything the "lguest" userspace program needs to know. */
+#include <linux/types.h>
 /* They can register up to 32 arrays of lguest_dma. */
 #define LGUEST_MAX_DMA		32
 /* At most we can dma 16 lguest_dma in one op. */

@@ -37,9 +38,9 @@ struct lguest_dma
 struct lguest_dma
 {
 	/* 0 if free to be used, filled by the Host. */
- 	u32 used_len;
+ 	__u32 used_len;
+	__u16 len[LGUEST_MAX_DMA_SECTIONS];
 	unsigned long addr[LGUEST_MAX_DMA_SECTIONS];
-	u16 len[LGUEST_MAX_DMA_SECTIONS];
 };
 /*:*/
 

@@ -52,11 +53,11 @@ struct lguest_block_page
 {
 	/* 0 is a read, 1 is a write. */
 	int type;
-	u32 sector; 	/* Offset in device = sector * 512. */
-	u32 bytes;	/* Length expected to be read/written in bytes */
+	__u32 sector; 	/* Offset in device = sector * 512. */
+	__u32 bytes;	/* Length expected to be read/written in bytes */
 	/* 0 = pending, 1 = done, 2 = done, error */
 	int result;
-	u32 num_sectors; /* Disk length = num_sectors * 512 */
+	__u32 num_sectors; /* Disk length = num_sectors * 512 */
 };
 
 /*D:520 The network device is basically a memory page where all the Guests on

@@ -86,21 +87,21 @@ struct lguest_net
  */
 struct lguest_device_desc {
 	/* The device type: console, network, disk etc. */
-	u16 type;
+	__u16 type;
 #define LGUEST_DEVICE_T_CONSOLE	1
 #define LGUEST_DEVICE_T_NET	2
 #define LGUEST_DEVICE_T_BLOCK	3
 
 	/* The specific features of this device: these depends on device type
 	 * except for LGUEST_DEVICE_F_RANDOMNESS. */
-	u16 features;
+	__u16 features;
 #define LGUEST_NET_F_NOCSUM		0x4000 /* Don't bother checksumming */
 #define LGUEST_DEVICE_F_RANDOMNESS	0x8000 /* IRQ is fairly random */
 
 	/* This is how the Guest reports status of the device: the Host can set
 	 * LGUEST_DEVICE_S_REMOVED to indicate removal, but the rest are only
 	 * ever manipulated by the Guest, and only ever set. */
-	u16 status;
+	__u16 status;
 /* 256 and above are device specific. */
 #define LGUEST_DEVICE_S_ACKNOWLEDGE	1 /* We have seen device. */
 #define LGUEST_DEVICE_S_DRIVER		2 /* We have found a driver */

@@ -111,8 +112,8 @@ struct lguest_device_desc {
 
 	/* Each device exists somewhere in Guest physical memory, over some
 	 * number of pages. */
-	u16 num_pages;
-	u32 pfn;
+	__u16 num_pages;
+	__u32 pfn;
 };
 /*:*/
 

@@ -87,7 +87,6 @@ struct lguest_data lguest_data = {
 	.noirq_end = (u32)lguest_noirq_end,
 	.blocked_interrupts = { 1 }, /* Block timer interrupts */
 };
-struct lguest_device_desc *lguest_devices;
 static cycle_t clock_base;
 
 /*G:035 Notice the lazy_hcall() above, rather than hcall().  This is our first

@@ -6,6 +6,8 @@
 #include <linux/lguest_bus.h>
 #include <asm/io.h>
 #include <asm/paravirt.h>
+
+struct lguest_device_desc *lguest_devices;
 
 static ssize_t type_show(struct device *_dev,
                          struct device_attribute *attr, char *buf)

103 lines not on this page · raw message

@@ -0,0 +1,67 @@
+/* Architecture specific portion of the lguest hypercalls */
+#ifndef _I386_LGUEST_HCALL_H
+#define _I386_LGUEST_HCALL_H
+
+#define LHCALL_FLUSH_ASYNC	0
+#define LHCALL_LGUEST_INIT	1
+#define LHCALL_CRASH		2
+#define LHCALL_LOAD_GDT		3
+#define LHCALL_NEW_PGTABLE	4
+#define LHCALL_FLUSH_TLB	5
+#define LHCALL_LOAD_IDT_ENTRY	6
+#define LHCALL_SET_STACK	7
+#define LHCALL_TS		8
+#define LHCALL_SET_CLOCKEVENT	9
+#define LHCALL_HALT		10
+#define LHCALL_BIND_DMA		12
+#define LHCALL_SEND_DMA		13
+#define LHCALL_SET_PTE		14
+#define LHCALL_SET_PMD		15
+#define LHCALL_LOAD_TLS		16
+
+/*G:031 First, how does our Guest contact the Host to ask for privileged
+ * operations?  There are two ways: the direct way is to make a "hypercall",
+ * to make requests of the Host Itself.
+ *
+ * Our hypercall mechanism uses the highest unused trap code (traps 32 and
+ * above are used by real hardware interrupts).  Seventeen hypercalls are
+ * available: the hypercall number is put in the %eax register, and the
+ * arguments (when required) are placed in %edx, %ebx and %ecx.  If a return
+ * value makes sense, it's returned in %eax.
+ *
+ * Grossly invalid calls result in Sudden Death at the hands of the vengeful
+ * Host, rather than returning failure.  This reflects Winston Churchill's
+ * definition of a gentleman: "someone who is only rude intentionally". */
+#define LGUEST_TRAP_ENTRY 0x1F
+
+static inline unsigned long
+hcall(unsigned long call,
+      unsigned long arg1, unsigned long arg2, unsigned long arg3)
+{
+	/* "int" is the Intel instruction to trigger a trap. */
+	asm volatile("int $" __stringify(LGUEST_TRAP_ENTRY)
+		       /* The call is in %eax (aka "a"), and can be replaced */
+		     : "=a"(call)
+		       /* The other arguments are in %eax, %edx, %ebx & %ecx */
+		     : "a"(call), "d"(arg1), "b"(arg2), "c"(arg3)
+		       /* "memory" means this might write somewhere in memory.
+			* This isn't true for all calls, but it's safe to tell
+			* gcc that it might happen so it doesn't get clever. */
+		     : "memory");
+	return call;
+}
+/*:*/
+
+void async_hcall(unsigned long call,
+		 unsigned long arg1, unsigned long arg2, unsigned long arg3);
+
+/* Can't use our min() macro here: needs to be a constant */
+#define LGUEST_IRQS (NR_IRQS < 32 ? NR_IRQS: 32)
+
+#define LHCALL_RING_SIZE 64
+struct hcall_ring
+{
+	u32 eax, edx, ebx, ecx;
+};
+
+#endif	/* _I386_LGUEST_HCALL_H */

84 lines not on this page · raw message

1,101 lines not on this page · raw message

1,101 lines not on this page · raw message

@@ -15,6 +15,7 @@
 #include <stdlib.h>
 #include <elf.h>
 #include <sys/mman.h>
+#include <sys/param.h>
 #include <sys/types.h>
 #include <sys/stat.h>
 #include <sys/wait.h>

@@ -162,6 +163,30 @@ static unsigned long entry_point(void *s
 	errx(1, "Is this image a genuine lguest?");
 }
 
+/* This routine is used to load the kernel or initrd.  It tries mmap, but if
+ * that fails (Plan 9's kernel file isn't nicely aligned on page boundaries),
+ * it falls back to reading the memory in. */
+static void map_at(int fd, void *addr, unsigned long offset, unsigned long len)
+{
+	ssize_t r;
+
+	/* We map writable even though for some segments are marked read-only.
+	 * The kernel really wants to be writable: it patches its own
+	 * instructions.
+	 *
+	 * MAP_PRIVATE means that the page won't be copied until a write is
+	 * done to it.  This allows us to share untouched memory between
+	 * Guests. */
+	if (mmap(addr, len, PROT_READ|PROT_WRITE|PROT_EXEC,
+		 MAP_FIXED|MAP_PRIVATE, fd, offset) != MAP_FAILED)
+		return;
+
+	/* pread does a seek and a read in one shot: saves a few lines. */
+	r = pread(fd, addr, len, offset);
+	if (r != len)
+		err(1, "Reading offset %lu len %lu gave %zi", offset, len, r);
+}
+
 /* This routine takes an open vmlinux image, which is in ELF, and maps it into
  * the Guest memory.  ELF = Embedded Linking Format, which is the format used
  * by all modern binaries on Linux including the kernel.

@@ -176,7 +201,6 @@ static unsigned long map_elf(int elf_fd,
 static unsigned long map_elf(int elf_fd, const Elf32_Ehdr *ehdr,
 			     unsigned long *page_offset)
 {
-	void *addr;
 	Elf32_Phdr phdr[ehdr->e_phnum];
 	unsigned int i;
 	unsigned long start = -1UL, end = 0;

@@ -227,23 +251,9 @@ static unsigned long map_elf(int elf_fd,
 		if (phdr[i].p_paddr + phdr[i].p_filesz > end)
 			end = phdr[i].p_paddr + phdr[i].p_filesz;
 
-		/* We map this section of the file at its physical address.  We
-		 * map it read & write even if the header says this segment is
-		 * read-only.  The kernel really wants to be writable: it
-		 * patches its own instructions which would normally be
-		 * read-only.
-		 *
-		 * MAP_PRIVATE means that the page won't be copied until a
-		 * write is done to it.  This allows us to share much of the
-		 * kernel memory between Guests. */
-		addr = mmap((void *)phdr[i].p_paddr,
-			    phdr[i].p_filesz,
-			    PROT_READ|PROT_WRITE|PROT_EXEC,
-			    MAP_FIXED|MAP_PRIVATE,
-			    elf_fd, phdr[i].p_offset);
-		if (addr != (void *)phdr[i].p_paddr)
-			err(1, "Mmaping vmlinux seg %i gave %p not %p",
-			    i, addr, (void *)phdr[i].p_paddr);
+		/* We map this section of the file at its physical address. */
+		map_at(elf_fd, (void *)phdr[i].p_paddr,
+		       phdr[i].p_offset, phdr[i].p_filesz);
 	}
 
 	return entry_point((void *)start, (void *)end, *page_offset);

@@ -402,27 +412,20 @@ static unsigned long load_initrd(const c
 	int ifd;
 	struct stat st;
 	unsigned long len;
-	void *iaddr;
 
 	ifd = open_or_die(name, O_RDONLY);
 	/* fstat() is needed to get the file size. */
 	if (fstat(ifd, &st) < 0)
 		err(1, "fstat() on initrd '%s'", name);
 
-	/* The length needs to be rounded up to a page size: mmap needs the
-	 * address to be page aligned. */
+	/* We map the initrd at the top of memory, but mmap wants it to be
+	 * page-aligned, so we round the size up for that. */
 	len = page_align(st.st_size);
-	/* We map the initrd at the top of memory. */
-	iaddr = mmap((void *)mem - len, st.st_size,
-		     PROT_READ|PROT_EXEC|PROT_WRITE,
-		     MAP_FIXED|MAP_PRIVATE, ifd, 0);
-	if (iaddr != (void *)mem - len)
-		err(1, "Mmaping initrd '%s' returned %p not %p",
-		    name, iaddr, (void *)mem - len);
+	map_at(ifd, (void *)mem - len, 0, st.st_size);
 	/* Once a file is mapped, you can close the file descriptor.  It's a
 	 * little odd, but quite useful. */
 	close(ifd);
-	verbose("mapped initrd %s size=%lu @ %p\n", name, st.st_size, iaddr);
+	verbose("mapped initrd %s size=%lu @ %p\n", name, len, (void*)mem-len);
 
 	/* We return the initrd size. */
 	return len;

368 lines not on this page · raw message

@@ -325,8 +325,8 @@ static int emulate_insn(struct lguest *l
  * Dealing With Guest Memory.
  *
  * When the Guest gives us (what it thinks is) a physical address, we can use
- * the normal copy_from_user() & copy_to_user() on that address: remember,
- * Guest physical == Launcher virtual.
+ * the normal copy_from_user() & copy_to_user() on the corresponding place in
+ * the memory region allocated by the Launcher.
  *
  * But we can't trust the Guest: it might be trying to access the Launcher
  * code.  We have to check that the range is below the pfn_limit the Launcher

@@ -348,8 +348,8 @@ u32 lgread_u32(struct lguest *lg, unsign
 
 	/* Don't let them access lguest binary. */
 	if (!lguest_address_ok(lg, addr, sizeof(val))
-	    || get_user(val, (u32 __user *)addr) != 0)
-		kill_guest(lg, "bad read address %#lx", addr);
+	    || get_user(val, (u32 *)(lg->mem_base + addr)) != 0)
+		kill_guest(lg, "bad read address %#lx: pfn_limit=%u membase=%p", addr, lg->pfn_limit, lg->mem_base);
 	return val;
 }
 

@@ -357,7 +357,7 @@ void lgwrite_u32(struct lguest *lg, unsi
 void lgwrite_u32(struct lguest *lg, unsigned long addr, u32 val)
 {
 	if (!lguest_address_ok(lg, addr, sizeof(val))
-	    || put_user(val, (u32 __user *)addr) != 0)
+	    || put_user(val, (u32 *)(lg->mem_base + addr)) != 0)
 		kill_guest(lg, "bad write address %#lx", addr);
 }
 

@@ -367,7 +367,7 @@ void lgread(struct lguest *lg, void *b, 
 void lgread(struct lguest *lg, void *b, unsigned long addr, unsigned bytes)
 {
 	if (!lguest_address_ok(lg, addr, bytes)
-	    || copy_from_user(b, (void __user *)addr, bytes) != 0) {
+	    || copy_from_user(b, 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);

@@ -379,7 +379,7 @@ void lgwrite(struct lguest *lg, unsigned
 	     unsigned bytes)
 {
 	if (!lguest_address_ok(lg, addr, bytes)
-	    || copy_to_user((void __user *)addr, b, bytes) != 0)
+	    || copy_to_user(lg->mem_base + addr, b, bytes) != 0)
 		kill_guest(lg, "bad write address %#lx len %u", addr, bytes);
 }
 /* (end of memory access helper routines) :*/

@@ -616,11 +616,9 @@ int run_guest(struct lguest *lg, unsigne
 			 *
 			 * Note that if the Guest were really messed up, this
 			 * could happen before it's done the INITIALIZE
-			 * hypercall, so lg->lguest_data will be NULL, so
-			 * &lg->lguest_data->cr2 will be address 8.  Writing
-			 * into that address won't hurt the Host at all,
-			 * though. */
-			if (put_user(cr2, &lg->lguest_data->cr2))
+			 * hypercall, so lg->lguest_data will be NULL */
+			if (lg->lguest_data
+			    && put_user(cr2, &lg->lguest_data->cr2))
 				kill_guest(lg, "Writing cr2");
 			break;
 		case 7: /* We've intercepted a Device Not Available fault. */

@@ -205,16 +205,19 @@ static void initialize(struct lguest *lg
 		tsc_speed = 0;
 
 	/* The pointer to the Guest's "struct lguest_data" is the only
-	 * argument. */
-	lg->lguest_data = (struct lguest_data __user *)lg->regs->edx;
-	/* If we check the address they gave is OK now, we can simply
-	 * copy_to_user/from_user from now on rather than using lgread/lgwrite.
-	 * I put this in to show that I'm not immune to writing stupid
-	 * optimizations. */
+	 * argument.  We check that address now. */
 	if (!lguest_address_ok(lg, lg->regs->edx, sizeof(*lg->lguest_data))) {
 		kill_guest(lg, "bad guest page %p", lg->lguest_data);
 		return;
 	}
+
+	/* Having checked it, we simply set lg->lguest_data to point straight
+	 * into the Launcher's memory at the right place and then use
+	 * 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 + lg->regs->edx;
+
 	/* 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)

@@ -186,7 +186,7 @@ int bind_dma(struct lguest *lg,
 	 * we're doing this. */
 	mutex_lock(&lguest_lock);
 	down_read(fshared);
-	if (get_futex_key((u32 __user *)ukey, fshared, &key) != 0) {
+	if (get_futex_key(lg->mem_base + ukey, fshared, &key) != 0) {
 		kill_guest(lg, "bad dma key %#lx", ukey);
 		goto unlock;
 	}

@@ -247,7 +247,8 @@ static int lgread_other(struct lguest *l
 			void *buf, u32 addr, unsigned bytes)
 {
 	if (!lguest_address_ok(lg, addr, bytes)
-	    || access_process_vm(lg->tsk, addr, buf, bytes, 0) != bytes) {
+	    || access_process_vm(lg->tsk, (unsigned long)lg->mem_base + addr,
+				 buf, bytes, 0) != bytes) {
 		memset(buf, 0, bytes);
 		kill_guest(lg, "bad address in registered DMA struct");
 		return 0;

@@ -261,8 +262,8 @@ static int lgwrite_other(struct lguest *
 			 const void *buf, unsigned bytes)
 {
 	if (!lguest_address_ok(lg, addr, bytes)
-	    || (access_process_vm(lg->tsk, addr, (void *)buf, bytes, 1)
-		!= bytes)) {
+	    || access_process_vm(lg->tsk, (unsigned long)lg->mem_base + addr,
+				 (void *)buf, bytes, 1) != bytes) {
 		kill_guest(lg, "bad address writing to registered DMA");
 		return 0;
 	}

@@ -318,7 +319,7 @@ static u32 copy_data(struct lguest *srcl
 		 * copy_to_user_page(), and some arch's seem to need special
 		 * flushes.  x86 is fine. */
 		if (copy_from_user(maddr + (dst->addr[di] + dstoff)%PAGE_SIZE,
-				   (void __user *)src->addr[si], len) != 0) {
+				   srclg->mem_base+src->addr[si], len) != 0) {
 			/* If a copy failed, it's the source's fault. */
 			kill_guest(srclg, "bad address in sending DMA");
 			totlen = 0;

@@ -377,7 +378,8 @@ static u32 do_dma(struct lguest *srclg, 
 		 * number of pages.  Note that we're holding the destination's
 		 * mmap_sem, as get_user_pages() requires. */
 		if (get_user_pages(dstlg->tsk, dstlg->mm,
-				   dst->addr[i], 1, 1, 1, pages+i, NULL)
+				   (unsigned long)dstlg->mem_base+dst->addr[i],
+				   1, 1, 1, pages+i, NULL)
 		    != 1) {
 			/* This means the destination gave us a bogus buffer */
 			kill_guest(dstlg, "Error mapping DMA pages");

@@ -493,7 +495,7 @@ again:
 	mutex_lock(&lguest_lock);
 	down_read(fshared);
 	/* Get the futex key for the key the Guest gave us */
-	if (get_futex_key((u32 __user *)ukey, fshared, &key) != 0) {
+	if (get_futex_key(lg->mem_base + ukey, fshared, &key) != 0) {
 		kill_guest(lg, "bad sending DMA key");
 		goto unlock;
 	}

@@ -584,7 +586,7 @@ unsigned long get_dma_buffer(struct lgue
 
 	/* This can fail if it's not a valid address, or if the address is not
 	 * divisible by 4 (the futex code needs that, we don't really). */
-	if (get_futex_key((u32 __user *)ukey, fshared, &key) != 0) {
+	if (get_futex_key(lg->mem_base + ukey, fshared, &key) != 0) {
 		kill_guest(lg, "bad registered DMA buffer");
 		goto unlock;
 	}

@@ -143,6 +143,9 @@ struct lguest
 	struct mm_struct *mm; 	/* == tsk->mm, but that becomes NULL on exit */
 	u16 guestid;
 	u32 pfn_limit;
+	/* This provides the offset to the base of guest-physical
+	 * memory in the Launcher. */
+	void __user *mem_base;
 	u32 page_offset;
 	u32 cr2;
 	int halted;

@@ -1,9 +1,9 @@
 /*P:200 This contains all the /dev/lguest code, whereby the userspace launcher
  * controls and communicates with the Guest.  For example, the first write will
- * tell us the memory size, pagetable, entry point and kernel address offset.
- * A read will run the Guest until a signal is pending (-EINTR), or the Guest
- * does a DMA out to the Launcher.  Writes are also used to get a DMA buffer
- * registered by the Guest and to send the Guest an interrupt. :*/
+ * tell us the Guest's memory layout, pagetable, entry point and kernel address
+ * offset.  A read will run the Guest until something happens, such as a signal
+ * or the Guest doing a DMA out to the Launcher.  Writes are also used to get a
+ * DMA buffer registered by the Guest and to send the Guest an interrupt. :*/
 #include <linux/uaccess.h>
 #include <linux/miscdevice.h>
 #include <linux/fs.h>

@@ -142,8 +142,10 @@ static ssize_t read(struct file *file, c
 	return run_guest(lg, (unsigned long __user *)user);
 }
 
-/*L:020 The initialization write supplies 4 32-bit values (in addition to the
+/*L:020 The initialization write supplies 5 32-bit values (in addition to the
  * 32-bit LHREQ_INITIALIZE value).  These are:
+ *
+ * base: The start of the Guest-physical memory inside the Launcher memory.
  *
  * pfnlimit: The highest (Guest-physical) page number the Guest should be
  * allowed to access.  The Launcher has to live in Guest memory, so it sets

@@ -166,7 +168,7 @@ static int initialize(struct file *file,
 	 * Guest. */
 	struct lguest *lg;
 	int err, i;
-	u32 args[4];
+	u32 args[5];
 
 	/* We grab the Big Lguest lock, which protects the global array
 	 * "lguests" and multiple simultaneous initializations. */

@@ -194,8 +196,9 @@ static int initialize(struct file *file,
 
 	/* Populate the easy fields of our "struct lguest" */
 	lg->guestid = i;
-	lg->pfn_limit = args[0];
-	lg->page_offset = args[3];
+	lg->mem_base = (void __user *)(long)args[0];
+	lg->pfn_limit = args[1];
+	lg->page_offset = args[4];
 
 	/* 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. */

@@ -210,13 +213,13 @@ static int initialize(struct file *file,
 	/* 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[1]);
+	err = init_guest_pagetable(lg, args[2]);
 	if (err)
 		goto free_regs;
 
 	/* Now we initialize the Guest's registers, handing it the start
 	 * address. */
-	setup_regs(lg->regs, args[2]);
+	setup_regs(lg->regs, args[3]);
 
 	/* There are a couple of GDT entries the Guest expects when first
 	 * booting. */

@@ -152,7 +152,7 @@ static spte_t gpte_to_spte(struct lguest
 static spte_t gpte_to_spte(struct lguest *lg, gpte_t gpte, int write)
 {
 	spte_t spte;
-	unsigned long pfn;
+	unsigned long pfn, base;
 
 	/* The Guest sets the global flag, because it thinks that it is using
 	 * PGE.  We only told it to use PGE so it would tell us whether it was

@@ -160,11 +160,14 @@ static spte_t gpte_to_spte(struct lguest
 	 * use the global bit, so throw it away. */
 	spte.flags = (gpte.flags & ~_PAGE_GLOBAL);
 
+	/* The Guest's pages are offset inside the Launcher. */
+	base = (unsigned long)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
 	 * fit in spte.pfn.  get_pfn() finds the real physical number of the
 	 * page, given the virtual number. */
-	pfn = get_pfn(gpte.pfn, write);
+	pfn = get_pfn(base + gpte.pfn, write);
 	if (pfn == -1UL) {
 		kill_guest(lg, "failed to get page %u", gpte.pfn);
 		/* When we destroy the Guest, we'll go through the shadow page

@@ -46,10 +46,6 @@ static struct {
 /* This One Big lock protects all inter-guest data structures. */
 DEFINE_MUTEX(lguest_lock);
 static DEFINE_PER_CPU(struct lguest *, last_guest);
-
-/* FIXME: Make dynamic. */
-#define MAX_LGUEST_GUESTS 16
-struct lguest lguests[MAX_LGUEST_GUESTS];
 
 /* Offset from where switcher.S was compiled to where we've copied it */
 static unsigned long switcher_offset(void)

@@ -660,16 +656,6 @@ int run_guest(struct lguest *lg, unsigne
  * deliver_trap() and demand_page().  After all those, we'll be ready to
  * examine the Switcher, and our philosophical understanding of the Host/Guest
  * duality will be complete. :*/
-
-int find_free_guest(void)
-{
-	unsigned int i;
-	for (i = 0; i < MAX_LGUEST_GUESTS; i++)
-		if (!lguests[i].tsk)
-			return i;
-	return -1;
-}
-
 static void adjust_pge(void *on)
 {
 	if (on)

@@ -225,9 +225,7 @@ static void initialize(struct lguest *lg
 	    /* 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(tsc_speed, &lg->lguest_data->tsc_khz)
-	    /* We also give the Guest a unique id, as used in lguest_net.c. */
-	    || put_user(lg->guestid, &lg->lguest_data->guestid))
+	    || put_user(tsc_speed, &lg->lguest_data->tsc_khz))
 		kill_guest(lg, "bad guest page %p", lg->lguest_data);
 
 	/* We write the current time into the Guest's data page once now. */

@@ -212,7 +212,7 @@ int bind_dma(struct lguest *lg,
 			lg->dma[i].num_dmas = numdmas;
 			lg->dma[i].next_dma = 0;
 			lg->dma[i].key = key;
-			lg->dma[i].guestid = lg->guestid;
+			lg->dma[i].owner = lg;
 			lg->dma[i].interrupt = interrupt;
 
 			/* Now we add it to the hash table: the position

@@ -412,7 +412,7 @@ static int dma_transfer(struct lguest *s
 
 	/* From the "struct lguest_dma_info" we found in the hash, grab the
 	 * Guest. */
-	dstlg = &lguests[dst->guestid];
+	dstlg = dst->owner;
 	/* Read in the source "struct lguest_dma" handed to SEND_DMA. */
 	lgread(srclg, &src_dma, udma, sizeof(src_dma));
 

@@ -506,8 +506,8 @@ again:
 		struct lguest_dma_info *i;
 		/* Look through the hash for other Guests. */
 		list_for_each_entry(i, &dma_hash[hash(&key)], list) {
-			/* Don't send to ourselves. */
-			if (i->guestid == lg->guestid)
+			/* Don't send to ourselves (would deadlock). */
+			if (i->owner->mm == lg->mm)
 				continue;
 			if (!key_eq(&key, &i->key))
 				continue;

@@ -594,7 +594,7 @@ unsigned long get_dma_buffer(struct lgue
 	 * send to its own Guest for the moment, so the entry must be for this
 	 * Guest) */
 	list_for_each_entry(i, &dma_hash[hash(&key)], list) {
-		if (key_eq(&key, &i->key) && i->guestid == lg->guestid) {
+		if (key_eq(&key, &i->key) && i->owner == lg) {
 			unsigned int j;
 			/* Look through the registered DMA array for an
 			 * available buffer. */

@@ -52,9 +52,9 @@ struct lguest_dma_info
 	struct list_head list;
 	union futex_key key;
 	unsigned long dmas;
+	struct lguest *owner;
 	u16 next_dma;
 	u16 num_dmas;
-	u16 guestid;
 	u8 interrupt; 	/* 0 when not registered */
 };
 

@@ -141,7 +141,6 @@ struct lguest
 	struct lguest_data __user *lguest_data;
 	struct task_struct *tsk;
 	struct mm_struct *mm; 	/* == tsk->mm, but that becomes NULL on exit */
-	u16 guestid;
 	u32 pfn_limit;
 	/* This provides the offset to the base of guest-physical
 	 * memory in the Launcher. */

@@ -196,7 +195,6 @@ struct lguest
 	DECLARE_BITMAP(irqs_pending, LGUEST_IRQS);
 };
 
-extern struct lguest lguests[];
 extern struct mutex lguest_lock;
 
 /* core.c: */

@@ -204,7 +202,6 @@ void lgwrite_u32(struct lguest *lg, unsi
 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);

@@ -167,11 +167,11 @@ static int initialize(struct file *file,
 	/* "struct lguest" contains everything we (the Host) know about a
 	 * Guest. */
 	struct lguest *lg;
-	int err, i;
+	int err;
 	u32 args[5];
 
-	/* We grab the Big Lguest lock, which protects the global array
-	 * "lguests" and multiple simultaneous initializations. */
+	/* We grab the Big Lguest lock, which protects against multiple
+	 * simultaneous initializations. */
 	mutex_lock(&lguest_lock);
 	/* You can't initialize twice!  Close the device and start again... */
 	if (file->private_data) {

@@ -184,18 +184,13 @@ static int initialize(struct file *file,
 		goto unlock;
 	}
 
-	/* Find an unused guest. */
-	i = find_free_guest();
-	if (i < 0) {
-		err = -ENOSPC;
+	lg = kzalloc(sizeof(*lg), GFP_KERNEL);
+	if (!lg) {
+		err = -ENOMEM;
 		goto unlock;
 	}
-	/* OK, we have an index into the "lguest" array: "lg" is a convenient
-	 * pointer. */
-	lg = &lguests[i];
 
 	/* Populate the easy fields of our "struct lguest" */
-	lg->guestid = i;
 	lg->mem_base = (void __user *)(long)args[0];
 	lg->pfn_limit = args[1];
 	lg->page_offset = args[4];

@@ -465,12 +465,7 @@ static int lguestnet_probe(struct lguest
 	/* Ethernet defaults with some changes */
 	ether_setup(dev);
 	dev->set_mac_address = NULL;
-
-	dev->dev_addr[0] = 0x02; /* set local assignment bit (IEEE802) */
-	dev->dev_addr[1] = 0x00;
-	memcpy(&dev->dev_addr[2], &lguest_data.guestid, 2);
-	dev->dev_addr[4] = 0x00;
-	dev->dev_addr[5] = 0x00;
+	random_ether_addr(dev->dev_addr);
 
 	dev->open = lguestnet_open;
 	dev->stop = lguestnet_close;

@@ -41,8 +41,6 @@ struct lguest_data
 /* Fields initialized by the Host at boot: */
 	/* Memory not to try to access */
 	unsigned long reserve_mem;
-	/* ID of this Guest (used by network driver to set ethernet address) */
-	u16 guestid;
 	/* KHz for the TSC clock. */
 	u32 tsc_khz;
 

@@ -215,10 +215,9 @@ int deliver_trap(struct lguest *lg, unsi
  * system calls down from 1750ns to 270ns.  Plus, if lguest didn't do it, all
  * the other hypervisors would tease it.
  *
- * This routine determines if a trap can be delivered directly. */
-static int direct_trap(const struct lguest *lg,
-		       const struct desc_struct *trap,
-		       unsigned int num)
+ * This routine indicates if a particular trap number could be delivered
+ * directly. */
+static int direct_trap(unsigned int num)
 {
 	/* Hardware interrupts don't go to the Guest at all (except system
 	 * call). */

@@ -229,14 +228,7 @@ static int direct_trap(const struct lgue
 	 * fault address), general protection faults (in/out emulation) and
 	 * device not available (TS handling), and of course, the hypercall
 	 * trap. */
-	if (num == 14 || num == 13 || num == 7 || num == LGUEST_TRAP_ENTRY)
-		return 0;
-
-	/* Only trap gates (type 15) can go direct to the Guest.  Interrupt
-	 * gates (type 14) disable interrupts as they are entered, which we
-	 * never let the Guest do.  Not present entries (type 0x0) also can't
-	 * go direct, of course 8) */
-	return idt_type(trap->a, trap->b) == 0xF;
+	return num != 14 && num != 13 && num != 7 && num != LGUEST_TRAP_ENTRY;
 }
 /*:*/
 

@@ -342,15 +334,11 @@ void load_guest_idt_entry(struct lguest 
 	 * to copy this again. */
 	lg->changed |= CHANGED_IDT;
 
-	/* The IDT which we keep in "struct lguest" only contains 32 entries
-	 * for the traps and LGUEST_IRQS (32) entries for interrupts.  We
-	 * ignore attempts to set handlers for higher interrupt numbers, except
-	 * for the system call "interrupt" at 128: we have a special IDT entry
-	 * for that. */
-	if (num < ARRAY_SIZE(lg->idt))
+	/* Check that the Guest doesn't try to step outside the bounds. */
+	if (num >= ARRAY_SIZE(lg->idt))
+		kill_guest(lg, "Setting idt entry %u", num);
+	else
 		set_trap(lg, &lg->idt[num], num, lo, hi);
-	else if (num == SYSCALL_VECTOR)
-		set_trap(lg, &lg->syscall_idt, num, lo, hi);
 }
 
 /* The default entry for each interrupt points into the Switcher routines which

@@ -393,20 +381,21 @@ void copy_traps(const struct lguest *lg,
 
 	/* We can simply copy the direct traps, otherwise we use the default
 	 * ones in the Switcher: they will return to the Host. */
-	for (i = 0; i < FIRST_EXTERNAL_VECTOR; i++) {
-		if (direct_trap(lg, &lg->idt[i], i))
+	for (i = 0; i < ARRAY_SIZE(lg->idt); i++) {
+		/* If no Guest can ever override this trap, leave it alone. */
+		if (!direct_trap(i))
+			continue;
+
+		/* Only trap gates (type 15) can go direct to the Guest.
+		 * Interrupt gates (type 14) disable interrupts as they are
+		 * entered, which we never let the Guest do.  Not present
+		 * entries (type 0x0) also can't go direct, of course. */
+		if (idt_type(lg->idt[i].a, lg->idt[i].b) == 0xF)
 			idt[i] = lg->idt[i];
 		else
+			/* Reset it to the default. */
 			default_idt_entry(&idt[i], i, def[i]);
 	}
-
-	/* Don't forget the system call trap!  The IDT entries for other
-	 * interupts never change, so no need to copy them. */
-	i = SYSCALL_VECTOR;
-	if (direct_trap(lg, &lg->syscall_idt, i))
-		idt[i] = lg->syscall_idt;
-	else
-		default_idt_entry(&idt[i], i, def[i]);
 }
 
 void guest_set_clockevent(struct lguest *lg, unsigned long delta)

@@ -183,8 +183,7 @@ struct lguest
 	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;
+	struct desc_struct idt[IDT_ENTRIES];
 
 	/* Virtual clock device */
 	struct hrtimer hrt;

diff -r 46a53d30a8c0 arch/i386/lguest/boot.c
@@ -65,6 +65,7 @@
 #include <asm/e820.h>
 #include <asm/mce.h>
 #include <asm/io.h>
+#include <asm/i387.h>
 
 /*G:010 Welcome to the Guest!
  *

538 lines not on this page · raw message

476 lines not on this page · raw message

diff -r 46a53d30a8c0 drivers/lguest/interrupts_and_traps.c
@@ -165,7 +165,7 @@ void maybe_do_interrupt(struct lguest *l
 	/* Look at the IDT entry the Guest gave us for this interrupt.  The
 	 * first 32 (FIRST_EXTERNAL_VECTOR) entries are for traps, so we skip
 	 * over them. */
-	idt = &lg->idt[FIRST_EXTERNAL_VECTOR+irq];
+	idt = &lg->arch.idt[FIRST_EXTERNAL_VECTOR+irq];
 	/* If they don't have a handler (yet?), we just ignore it */
 	if (idt_present(idt->a, idt->b)) {
 		/* OK, mark it no longer pending and deliver it. */

@@ -197,14 +197,14 @@ int deliver_trap(struct lguest *lg, unsi
 {
 	/* Trap numbers are always 8 bit, but we set an impossible trap number
 	 * for traps inside the Switcher, so check that here. */
-	if (num >= ARRAY_SIZE(lg->idt))
+	if (num >= ARRAY_SIZE(lg->arch.idt))
 		return 0;
 
 	/* Early on the Guest hasn't set the IDT entries (or maybe it put a
 	 * bogus one in): if we fail here, the Guest will be killed. */
-	if (!idt_present(lg->idt[num].a, lg->idt[num].b))
+	if (!idt_present(lg->arch.idt[num].a, lg->arch.idt[num].b))
 		return 0;
-	set_guest_interrupt(lg, lg->idt[num].a, lg->idt[num].b, has_err(num));
+	set_guest_interrupt(lg, lg->arch.idt[num].a, lg->arch.idt[num].b, has_err(num));
 	return 1;
 }
 

@@ -341,10 +341,10 @@ void load_guest_idt_entry(struct lguest 
 	lg->changed |= CHANGED_IDT;
 
 	/* Check that the Guest doesn't try to step outside the bounds. */
-	if (num >= ARRAY_SIZE(lg->idt))
+	if (num >= ARRAY_SIZE(lg->arch.idt))
 		kill_guest(lg, "Setting idt entry %u", num);
 	else
-		set_trap(lg, &lg->idt[num], num, lo, hi);
+		set_trap(lg, &lg->arch.idt[num], num, lo, hi);
 }
 
 /* The default entry for each interrupt points into the Switcher routines which

@@ -387,7 +387,7 @@ void copy_traps(const struct lguest *lg,
 
 	/* We can simply copy the direct traps, otherwise we use the default
 	 * ones in the Switcher: they will return to the Host. */
-	for (i = 0; i < ARRAY_SIZE(lg->idt); i++) {
+	for (i = 0; i < ARRAY_SIZE(lg->arch.idt); i++) {
 		/* If no Guest can ever override this trap, leave it alone. */
 		if (!direct_trap(i))
 			continue;

@@ -396,8 +396,8 @@ void copy_traps(const struct lguest *lg,
 		 * Interrupt gates (type 14) disable interrupts as they are
 		 * entered, which we never let the Guest do.  Not present
 		 * entries (type 0x0) also can't go direct, of course. */
-		if (idt_type(lg->idt[i].a, lg->idt[i].b) == 0xF)
-			idt[i] = lg->idt[i];
+		if (idt_type(lg->arch.idt[i].a, lg->arch.idt[i].b) == 0xF)
+			idt[i] = lg->arch.idt[i];
 		else
 			/* Reset it to the default. */
 			default_idt_entry(&idt[i], i, def[i]);

97 lines not on this page · raw message

diff -r 46a53d30a8c0 drivers/lguest/segments.c
@@ -73,14 +73,14 @@ static void fixup_gdt_table(struct lgues
 		/* Segment descriptors contain a privilege level: the Guest is
 		 * sometimes careless and leaves this as 0, even though it's
 		 * running at privilege level 1.  If so, we fix it here. */
-		if ((lg->gdt[i].b & 0x00006000) == 0)
-			lg->gdt[i].b |= (GUEST_PL << 13);
+		if ((lg->arch.gdt[i].b & 0x00006000) == 0)
+			lg->arch.gdt[i].b |= (GUEST_PL << 13);
 
 		/* Each descriptor has an "accessed" bit.  If we don't set it
 		 * now, the CPU will try to set it when the Guest first loads
 		 * that entry into a segment register.  But the GDT isn't
 		 * writable by the Guest, so bad things can happen. */
-		lg->gdt[i].b |= 0x00000100;
+		lg->arch.gdt[i].b |= 0x00000100;
 	}
 }
 

@@ -106,12 +106,12 @@ void setup_guest_gdt(struct lguest *lg)
 void setup_guest_gdt(struct lguest *lg)
 {
 	/* Start with full 0-4G segments... */
-	lg->gdt[GDT_ENTRY_KERNEL_CS] = FULL_EXEC_SEGMENT;
-	lg->gdt[GDT_ENTRY_KERNEL_DS] = FULL_SEGMENT;
+	lg->arch.gdt[GDT_ENTRY_KERNEL_CS] = FULL_EXEC_SEGMENT;
+	lg->arch.gdt[GDT_ENTRY_KERNEL_DS] = FULL_SEGMENT;
 	/* ...except the Guest is allowed to use them, so set the privilege
 	 * level appropriately in the flags. */
-	lg->gdt[GDT_ENTRY_KERNEL_CS].b |= (GUEST_PL << 13);
-	lg->gdt[GDT_ENTRY_KERNEL_DS].b |= (GUEST_PL << 13);
+	lg->arch.gdt[GDT_ENTRY_KERNEL_CS].b |= (GUEST_PL << 13);
+	lg->arch.gdt[GDT_ENTRY_KERNEL_DS].b |= (GUEST_PL << 13);
 }
 
 /* Like the IDT, we never simply use the GDT the Guest gives us.  We set up the

@@ -126,7 +126,7 @@ void copy_gdt_tls(const struct lguest *l
 	unsigned int i;
 
 	for (i = GDT_ENTRY_TLS_MIN; i <= GDT_ENTRY_TLS_MAX; i++)
-		gdt[i] = lg->gdt[i];
+		gdt[i] = lg->arch.gdt[i];
 }
 
 /* This is the full version */

@@ -138,7 +138,7 @@ void copy_gdt(const struct lguest *lg, s
 	 * replaced.  See ignored_gdt() above. */
 	for (i = 0; i < GDT_ENTRIES; i++)
 		if (!ignored_gdt(i))
-			gdt[i] = lg->gdt[i];
+			gdt[i] = lg->arch.gdt[i];
 }
 
 /* This is where the Guest asks us to load a new GDT (LHCALL_LOAD_GDT). */

@@ -146,12 +146,12 @@ void load_guest_gdt(struct lguest *lg, u
 {
 	/* 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(lg->gdt))
+	if (num > ARRAY_SIZE(lg->arch.gdt))
 		kill_guest(lg, "too many gdt entries %i", num);
 
 	/* We read the whole thing in, then fix it up. */
-	lgread(lg, lg->gdt, table, num * sizeof(lg->gdt[0]));
-	fixup_gdt_table(lg, 0, ARRAY_SIZE(lg->gdt));
+	lgread(lg, lg->arch.gdt, table, num * sizeof(lg->arch.gdt[0]));
+	fixup_gdt_table(lg, 0, ARRAY_SIZE(lg->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. */
 	lg->changed |= CHANGED_GDT;

@@ -159,7 +159,7 @@ void load_guest_gdt(struct lguest *lg, u
 
 void guest_load_tls(struct lguest *lg, unsigned long gtls)
 {
-	struct desc_struct *tls = &lg->gdt[GDT_ENTRY_TLS_MIN];
+	struct desc_struct *tls = &lg->arch.gdt[GDT_ENTRY_TLS_MIN];
 
 	lgread(lg, tls, gtls, sizeof(*tls)*GDT_ENTRY_TLS_ENTRIES);
 	fixup_gdt_table(lg, GDT_ENTRY_TLS_MIN, GDT_ENTRY_TLS_MAX+1);

87 lines not on this page · raw message

@@ -26,10 +26,9 @@ extern unsigned long default_idt_entries
 struct lguest_regs
 {
 	/* Manually saved part. */
-	unsigned long ebx, ecx, edx;
+	unsigned long eax, 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 */

@@ -198,10 +198,10 @@ int run_guest(struct lguest *lg, unsigne
 {
 	/* We stop running once the Guest is dead. */
 	while (!lg->dead) {
-		/* First we run any hypercalls the Guest wants done: either in
-		 * the hypercall ring in "struct lguest_data", or directly by
-		 * using int 31 (LGUEST_TRAP_ENTRY). */
-		do_hypercalls(lg);
+		/* First we run any hypercalls the Guest wants done. */
+		if (lg->hcall)
+			do_hypercalls(lg);
+
 		/* It's possible the Guest did a SEND_DMA hypercall to the
 		 * Launcher, in which case we return from the read() now. */
 		if (lg->dma_is_pending) {

67 lines not on this page · raw message

@@ -316,13 +316,14 @@ void lguest_arch_handle_trap(struct lgue
 			return;
 		break;
 	case 32 ... 255:
-		/* These values mean a real interrupt occurred, in
-		 * which case the Host handler has already been run.
-		 * We just do a friendly check if another process
-		 * should now be run, then fall through to loop
-		 * around: */
+		/* These values mean a real interrupt occurred, in which case
+		 * the Host handler has already been run.  We just do a
+		 * friendly check if another process should now be run, then
+		 * return to run the Guest again */
 		cond_resched();
-	case LGUEST_TRAP_ENTRY: /* Handled before re-entering Guest */
+		return;
+	case LGUEST_TRAP_ENTRY:
+		lg->hcall = lg->regs;
 		return;
 	}
 

@@ -111,6 +111,9 @@ struct lguest
 	u32 esp1;
 	u8 ss1;
 
+	/* If a hypercall was asked for, this points to the arguments. */
+	struct lguest_regs *hcall;
+
 	/* Do we need to stop what we're doing and return to userspace? */
 	int break_out;
 	wait_queue_head_t break_wq;

diff -r 5da9ffae9f41 arch/i386/lguest/boot.c
@@ -158,10 +158,10 @@ void async_hcall(unsigned long call,
 		/* Table full, so do normal hcall which will flush table. */
 		hcall(call, arg1, arg2, arg3);
 	} else {
-		lguest_data.hcalls[next_call].eax = call;
-		lguest_data.hcalls[next_call].edx = arg1;
-		lguest_data.hcalls[next_call].ebx = arg2;
-		lguest_data.hcalls[next_call].ecx = arg3;
+		lguest_data.hcalls[next_call].arg0 = call;
+		lguest_data.hcalls[next_call].arg1 = arg1;
+		lguest_data.hcalls[next_call].arg2 = arg2;
+		lguest_data.hcalls[next_call].arg3 = arg3;
 		/* Arguments must all be written before we mark it to go */
 		wmb();
 		lguest_data.hcall_status[next_call] = 0;

210 lines not on this page · raw message

71 lines not on this page · raw message

diff -r 5da9ffae9f41 drivers/lguest/lg.h
@@ -108,7 +108,7 @@ struct lguest
 	u8 ss1;
 
 	/* If a hypercall was asked for, this points to the arguments. */
-	struct lguest_regs *hcall;
+	struct hcall_args *hcall;
 
 	/* Do we need to stop what we're doing and return to userspace? */
 	int break_out;

@@ -198,6 +198,8 @@ void lguest_arch_host_fini(void);
 void lguest_arch_host_fini(void);
 void lguest_arch_run_guest(struct lguest *lg);
 void lguest_arch_handle_trap(struct lguest *lg);
+int lguest_arch_init_hypercalls(struct lguest *lg);
+int lguest_arch_do_hcall(struct lguest *lg, struct hcall_args *args);
 
 /* <arch>_switcher.S: */
 extern char start_switcher_text[], end_switcher_text[], switch_to_guest[];

diff -r 5da9ffae9f41 include/asm-i386/lguest_hcall.h
@@ -1,6 +1,8 @@
 /* Architecture specific portion of the lguest hypercalls */
 #ifndef _I386_LGUEST_HCALL_H
 #define _I386_LGUEST_HCALL_H
+
+#include <irq_vectors.h>
 
 #define LHCALL_FLUSH_ASYNC	0
 #define LHCALL_LGUEST_INIT	1

@@ -59,9 +61,9 @@ void async_hcall(unsigned long call,
 #define LGUEST_IRQS (NR_IRQS < 32 ? NR_IRQS: 32)
 
 #define LHCALL_RING_SIZE 64
-struct hcall_ring
+struct hcall_args
 {
-	u32 eax, edx, ebx, ecx;
+	/* These map directly onto eax, ebx, ecx, edx in struct lguest_regs */
+	unsigned long arg0, arg2, arg3, arg1;
 };
-
 #endif	/* _I386_LGUEST_HCALL_H */

diff -r 5da9ffae9f41 include/linux/lguest.h
@@ -36,7 +36,7 @@ struct lguest_data
 	/* 0xFF == done (set by Host), 0 == pending (set by Guest). */
 	u8 hcall_status[LHCALL_RING_SIZE];
 	/* The actual registers for the hypercalls. */
-	struct hcall_ring hcalls[LHCALL_RING_SIZE];
+	struct hcall_args hcalls[LHCALL_RING_SIZE];
 
 /* Fields initialized by the Host at boot: */
 	/* Memory not to try to access */

@@ -473,9 +473,9 @@ static unsigned long setup_pagetables(un
 				      unsigned long initrd_size,
 				      unsigned long page_offset)
 {
-	u32 *pgdir, *linear;
+	unsigned long *pgdir, *linear;
 	unsigned int mapped_pages, i, linear_pages;
-	unsigned int ptes_per_page = getpagesize()/sizeof(u32);
+	unsigned int ptes_per_page = getpagesize()/sizeof(void *);
 
 	/* Ideally we map all physical memory starting at page_offset.
 	 * However, if page_offset is 0xC0000000 we can only map 1G of physical

@@ -505,7 +505,7 @@ static unsigned long setup_pagetables(un
 	 * continue from there. */
 	for (i = 0; i < mapped_pages; i += ptes_per_page) {
 		pgdir[(i + page_offset/getpagesize())/ptes_per_page]
-			= ((to_guest_phys(linear) + i*sizeof(u32))
+			= ((to_guest_phys(linear) + i*sizeof(void *))
 			   | PAGE_PRESENT);
 	}
 

@@ -537,12 +537,13 @@ static void concat(char *dst, char *args
  * the base of guest "physical" memory, the top physical page to allow, the
  * top level pagetable, the entry point and the page_offset constant for the
  * Guest. */
-static int tell_kernel(u32 pgdir, u32 start, u32 page_offset)
-{
-	u32 args[] = { LHREQ_INITIALIZE,
-		       (unsigned long)guest_base,
-		       guest_limit / getpagesize(),
-		       pgdir, start, page_offset };
+static int tell_kernel(unsigned long pgdir, unsigned long start,
+		       unsigned long page_offset)
+{
+	unsigned long args[] = { LHREQ_INITIALIZE,
+				 (unsigned long)guest_base,
+				 guest_limit / getpagesize(),
+				 pgdir, start, page_offset };
 	int fd;
 
 	verbose("Guest: %p - %p (%#lx)\n",

@@ -586,7 +587,7 @@ static void wake_parent(int pipefd, int 
 
 	for (;;) {
 		fd_set rfds = devices->infds;
-		u32 args[] = { LHREQ_BREAK, 1 };
+		unsigned long args[] = { LHREQ_BREAK, 1 };
 
 		/* Wait until input is ready from one of the devices. */
 		select(devices->max_infd+1, &rfds, NULL, NULL, NULL);

@@ -684,7 +685,7 @@ static u32 *get_dma_buffer(int fd, void 
 static u32 *get_dma_buffer(int fd, void *key,
 			   struct iovec iov[], unsigned int *num, u32 *irq)
 {
-	u32 buf[] = { LHREQ_GETDMA, to_guest_phys(key) };
+	unsigned long buf[] = { LHREQ_GETDMA, to_guest_phys(key) };
 	unsigned long udma;
 	u32 *res;
 

@@ -705,7 +706,7 @@ static u32 *get_dma_buffer(int fd, void 
 /* This is a convenient routine to send the Guest an interrupt. */
 static void trigger_irq(int fd, u32 irq)
 {
-	u32 buf[] = { LHREQ_IRQ, irq };
+	unsigned long buf[] = { LHREQ_IRQ, irq };
 	if (write(fd, buf, sizeof(buf)) != 0)
 		err(1, "Triggering irq %i", irq);
 }

@@ -787,7 +788,7 @@ static bool handle_console_input(int fd,
 			struct timeval now;
 			gettimeofday(&now, NULL);
 			if (now.tv_sec <= abort->start.tv_sec+1) {
-				u32 args[] = { LHREQ_BREAK, 0 };
+				unsigned long args[] = { LHREQ_BREAK, 0 };
 				/* Close the fd so Waker will know it has to
 				 * exit. */
 				close(waker_fd);

@@ -1365,7 +1366,7 @@ run_guest(int lguest_fd, struct device_l
 run_guest(int lguest_fd, struct device_list *device_list)
 {
 	for (;;) {
-		u32 args[] = { LHREQ_BREAK, 0 };
+		unsigned long args[] = { LHREQ_BREAK, 0 };
 		unsigned long arr[2];
 		int readval;
 

@@ -43,7 +43,7 @@ static void setup_regs(struct lguest_reg
 /*L:310 To send DMA into the Guest, the Launcher needs to be able to ask for a
  * DMA buffer.  This is done by writing LHREQ_GETDMA and the key to
  * /dev/lguest. */
-static long user_get_dma(struct lguest *lg, const u32 __user *input)
+static long user_get_dma(struct lguest *lg, const unsigned long __user *input)
 {
 	unsigned long key, udma, irq;
 

@@ -67,7 +67,7 @@ static long user_get_dma(struct lguest *
 /*L:315 To force the Guest to stop running and return to the Launcher, the
  * Waker sets writes LHREQ_BREAK and the value "1" to /dev/lguest.  The
  * Launcher then writes LHREQ_BREAK and "0" to release the Waker. */
-static int break_guest_out(struct lguest *lg, const u32 __user *input)
+static int break_guest_out(struct lguest *lg, const unsigned long __user *input)
 {
 	unsigned long on;
 

@@ -90,9 +90,9 @@ static int break_guest_out(struct lguest
 
 /*L:050 Sending an interrupt is done by writing LHREQ_IRQ and an interrupt
  * number to /dev/lguest. */
-static int user_send_irq(struct lguest *lg, const u32 __user *input)
-{
-	u32 irq;
+static int user_send_irq(struct lguest *lg, const unsigned long __user *input)
+{
+	unsigned long irq;
 
 	if (get_user(irq, input) != 0)
 		return -EFAULT;

@@ -142,8 +142,8 @@ static ssize_t read(struct file *file, c
 	return run_guest(lg, (unsigned long __user *)user);
 }
 
-/*L:020 The initialization write supplies 5 32-bit values (in addition to the
- * 32-bit LHREQ_INITIALIZE value).  These are:
+/*L:020 The initialization write supplies 5 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.
  *

@@ -162,13 +162,13 @@ static ssize_t read(struct file *file, c
  * quickly converted from physical to virtual by adding PAGE_OFFSET.  It's
  * 0xC0000000 (3G) by default, but it's configurable at kernel build time.
  */
-static int initialize(struct file *file, const u32 __user *input)
+static int initialize(struct file *file, const unsigned long __user *input)
 {
 	/* "struct lguest" contains everything we (the Host) know about a
 	 * Guest. */
 	struct lguest *lg;
 	int err;
-	u32 args[5];
+	unsigned long args[5];
 
 	/* We grab the Big Lguest lock, which protects against multiple
 	 * simultaneous initializations. */

@@ -259,17 +259,18 @@ unlock:
  * start with a 32 bit number: for the first write this must be
  * LHREQ_INITIALIZE to set up the Guest.  After that the Launcher can use
  * writes of other values to get DMA buffers and send interrupts. */
-static ssize_t write(struct file *file, const char __user *input,
+static ssize_t write(struct file *file, const char __user *in,
 		     size_t size, loff_t *off)
 {
 	/* Once the guest is initialized, we hold the "struct lguest" in the
 	 * file private data. */
 	struct lguest *lg = file->private_data;
-	u32 req;
+	const unsigned long __user *input = (const unsigned long __user *)in;
+	unsigned long req;
 
 	if (get_user(req, input) != 0)
 		return -EFAULT;
-	input += sizeof(req);
+	input++;
 
 	/* If you haven't initialized, you must do that first. */
 	if (req != LHREQ_INITIALIZE && !lg)

@@ -285,13 +286,13 @@ static ssize_t write(struct file *file, 
 
 	switch (req) {
 	case LHREQ_INITIALIZE:
-		return initialize(file, (const u32 __user *)input);
+		return initialize(file, input);
 	case LHREQ_GETDMA:
-		return user_get_dma(lg, (const u32 __user *)input);
+		return user_get_dma(lg, input);
 	case LHREQ_IRQ:
-		return user_send_irq(lg, (const u32 __user *)input);
+		return user_send_irq(lg, input);
 	case LHREQ_BREAK:
-		return break_guest_out(lg, (const u32 __user *)input);
+		return break_guest_out(lg, input);
 	default:
 		return -EINVAL;
 	}

@@ -535,3 +535,39 @@ int lguest_arch_init_hypercalls(struct l
 /* Now we've examined the hypercall code; our Guest can make requests.  There
  * is one other way we can do things for the Guest, as we see in
  * emulate_insn(). :*/
+
+/*L:030 lguest_arch_setup_regs()
+ *
+ * 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)
+{
+	struct lguest_regs *regs = lg->regs;
+
+	/* There are four "segment" registers which the Guest needs to boot:
+	 * The "code segment" register (cs) refers to the kernel code segment
+	 * __KERNEL_CS, and the "data", "extra" and "stack" segment registers
+	 * refer to the kernel data segment __KERNEL_DS.
+	 *
+	 * The privilege level is packed into the lower bits.  The Guest runs
+	 * at privilege level 1 (GUEST_PL).*/
+	regs->ds = regs->es = regs->ss = __KERNEL_DS|GUEST_PL;
+	regs->cs = __KERNEL_CS|GUEST_PL;
+
+	/* The "eflags" register contains miscellaneous flags.  Bit 1 (0x002)
+	 * is supposed to always be "1".  Bit 9 (0x200) controls whether
+	 * interrupts are enabled.  We always leave interrupts enabled while
+	 * running the Guest. */
+	regs->eflags = 0x202;
+
+	/* The "Extended Instruction Pointer" register says where the Guest is
+	 * running. */
+	regs->eip = start;
+
+	/* %esi points to our boot information, at physical address 0, so don't
+	 * touch it. */
+	/* There are a couple of GDT entries the Guest expects when first
+	 * booting. */
+
+	setup_guest_gdt(lg);
+}

@@ -204,6 +204,7 @@ void lguest_arch_handle_trap(struct lgue
 void lguest_arch_handle_trap(struct lguest *lg);
 int lguest_arch_init_hypercalls(struct lguest *lg);
 int lguest_arch_do_hcall(struct lguest *lg, struct hcall_args *args);
+void lguest_arch_setup_regs(struct lguest *lg, unsigned long start);
 
 /* <arch>_switcher.S: */
 extern char start_switcher_text[], end_switcher_text[], switch_to_guest[];

@@ -9,37 +9,6 @@
 #include <linux/fs.h>
 #include "lg.h"
 
-/*L:030 setup_regs() doesn't really belong in this file, but it gives us an
- * early glimpse deeper into the Host so it's worth having here.
- *
- * Most of the Guest's registers are left alone: we used get_zeroed_page() to
- * allocate the structure, so they will be 0. */
-static void setup_regs(struct lguest_regs *regs, unsigned long start)
-{
-	/* There are four "segment" registers which the Guest needs to boot:
-	 * The "code segment" register (cs) refers to the kernel code segment
-	 * __KERNEL_CS, and the "data", "extra" and "stack" segment registers
-	 * refer to the kernel data segment __KERNEL_DS.
-	 *
-	 * The privilege level is packed into the lower bits.  The Guest runs
-	 * at privilege level 1 (GUEST_PL).*/
-	regs->ds = regs->es = regs->ss = __KERNEL_DS|GUEST_PL;
-	regs->cs = __KERNEL_CS|GUEST_PL;
-
-	/* The "eflags" register contains miscellaneous flags.  Bit 1 (0x002)
-	 * is supposed to always be "1".  Bit 9 (0x200) controls whether
-	 * interrupts are enabled.  We always leave interrupts enabled while
-	 * running the Guest. */
-	regs->eflags = 0x202;
-
-	/* The "Extended Instruction Pointer" register says where the Guest is
-	 * running. */
-	regs->eip = start;
-
-	/* %esi points to our boot information, at physical address 0, so don't
-	 * touch it. */
-}
-
 /*L:310 To send DMA into the Guest, the Launcher needs to be able to ask for a
  * DMA buffer.  This is done by writing LHREQ_GETDMA and the key to
  * /dev/lguest. */

@@ -214,11 +183,7 @@ static int initialize(struct file *file,
 
 	/* Now we initialize the Guest's registers, handing it the start
 	 * address. */
-	setup_regs(lg->regs, args[3]);
-
-	/* There are a couple of GDT entries the Guest expects when first
-	 * booting. */
-	setup_guest_gdt(lg);
+	lguest_arch_setup_regs(lg, args[3]);
 
 	/* The timer for lguest's clock needs initialization. */
 	init_clockdev(lg);

@@ -4,6 +4,7 @@
 #define _LINUX_LGUEST_H
 
 #ifndef __ASSEMBLY__
+#include <linux/time.h>
 #include <asm/irq.h>
 #include <asm/lguest_hcall.h>
 

@@ -83,7 +83,7 @@ static void do_hcall(struct lguest *lg, 
 		guest_set_stack(lg, args->arg1, args->arg2, args->arg3);
 		break;
 	case LHCALL_SET_PTE:
-		guest_set_pte(lg, args->arg1, args->arg2, mkgpte(args->arg3));
+		guest_set_pte(lg, args->arg1, args->arg2, __pte(args->arg3));
 		break;
 	case LHCALL_SET_PMD:
 		guest_set_pmd(lg, args->arg1, args->arg2);

66 lines not on this page · raw message

454 lines not on this page · raw message

@@ -35,7 +35,7 @@ struct lguest_dma_info
 
 struct pgdir
 {
-	unsigned long cr3;
+	unsigned long gpgdir;
 	pgd_t *pgdir;
 };
 

@@ -159,10 +159,10 @@ int init_guest_pagetable(struct lguest *
 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_set_pmd(struct lguest *lg, unsigned long gpgdir, 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,
+void guest_set_pte(struct lguest *lg, unsigned long gpgdir,
 		   unsigned long vaddr, pte_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);

@@ -96,7 +96,7 @@ static unsigned long gpgd_addr(struct lg
 static unsigned long gpgd_addr(struct lguest *lg, unsigned long vaddr)
 {
 	unsigned int index = vaddr >> (PGDIR_SHIFT);
-	return lg->pgdirs[lg->pgdidx].cr3 + index * sizeof(pgd_t);
+	return lg->pgdirs[lg->pgdidx].gpgdir + index * sizeof(pgd_t);
 }
 
 static unsigned long gpte_addr(struct lguest *lg,

@@ -369,7 +369,7 @@ static unsigned int find_pgdir(struct lg
 {
 	unsigned int i;
 	for (i = 0; i < ARRAY_SIZE(lg->pgdirs); i++)
-		if (lg->pgdirs[i].cr3 == pgtable)
+		if (lg->pgdirs[i].gpgdir == pgtable)
 			break;
 	return i;
 }

@@ -378,7 +378,7 @@ static unsigned int find_pgdir(struct lg
  * allocate a new one (and so the kernel parts are not there), we set
  * blank_pgdir. */
 static unsigned int new_pgdir(struct lguest *lg,
-			      unsigned long cr3,
+			      unsigned long gpgdir,
 			      int *blank_pgdir)
 {
 	unsigned int next;

@@ -398,7 +398,7 @@ static unsigned int new_pgdir(struct lgu
 			*blank_pgdir = 1;
 	}
 	/* Record which Guest toplevel this shadows. */
-	lg->pgdirs[next].cr3 = cr3;
+	lg->pgdirs[next].gpgdir = gpgdir;
 	/* Release all the non-kernel mappings. */
 	flush_user_mappings(lg, next);
 

@@ -499,7 +499,7 @@ static void do_set_pte(struct lguest *lg
  * 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,
-		   unsigned long cr3, unsigned long vaddr, pte_t gpte)
+		   unsigned long gpgdir, unsigned long vaddr, pte_t gpte)
 {
 	/* Kernel mappings must be changed on all top levels.  Slow, but
 	 * doesn't happen often. */

@@ -510,7 +510,7 @@ void guest_set_pte(struct lguest *lg,
 				do_set_pte(lg, i, vaddr, gpte);
 	} else {
 		/* Is this page table one we have a shadow for? */
-		int pgdir = find_pgdir(lg, cr3);
+		int pgdir = find_pgdir(lg, gpgdir);
 		if (pgdir != ARRAY_SIZE(lg->pgdirs))
 			/* If so, do the update. */
 			do_set_pte(lg, pgdir, vaddr, gpte);

@@ -531,7 +531,7 @@ void guest_set_pte(struct lguest *lg,
  *
  * So with that in mind here's our code to to update a (top-level) PGD entry:
  */
-void guest_set_pmd(struct lguest *lg, unsigned long cr3, u32 idx)
+void guest_set_pmd(struct lguest *lg, unsigned long gpgdir, u32 idx)
 {
 	int pgdir;
 

@@ -541,7 +541,7 @@ void guest_set_pmd(struct lguest *lg, un
 		return;
 
 	/* If they're talking about a page table we have a shadow for... */
-	pgdir = find_pgdir(lg, cr3);
+	pgdir = find_pgdir(lg, gpgdir);
 	if (pgdir < ARRAY_SIZE(lg->pgdirs))
 		/* ... throw it away. */
 		release_pgd(lg, lg->pgdirs[pgdir].pgdir + idx);

@@ -561,7 +561,7 @@ int init_guest_pagetable(struct lguest *
 	/* We start on the first shadow page table, and give it a blank PGD
 	 * page. */
 	lg->pgdidx = 0;
-	lg->pgdirs[lg->pgdidx].cr3 = pgtable;
+	lg->pgdirs[lg->pgdidx].gpgdir = pgtable;
 	lg->pgdirs[lg->pgdidx].pgdir = (pgd_t*)get_zeroed_page(GFP_KERNEL);
 	if (!lg->pgdirs[lg->pgdidx].pgdir)
 		return -ENOMEM;

@@ -400,7 +400,8 @@ void __init native_init_IRQ(void)
 		int vector = FIRST_EXTERNAL_VECTOR + i;
 		if (i >= NR_IRQS)
 			break;
-		if (vector != SYSCALL_VECTOR) 
+		/* SYSCALL_VECTOR was reserved in trap_init. */
+		if (!test_bit(vector, used_vectors))
 			set_intr_gate(vector, interrupt[i]);
 	}
 

@@ -1198,7 +1198,7 @@ static int __assign_irq_vector(int irq)
 static int __assign_irq_vector(int irq)
 {
 	static int current_vector = FIRST_DEVICE_VECTOR, current_offset = 0;
-	int vector, offset, i;
+	int vector, offset;
 
 	BUG_ON((unsigned)irq >= NR_IRQ_VECTORS);
 

@@ -1215,11 +1215,8 @@ next:
 	}
 	if (vector == current_vector)
 		return -ENOSPC;
-	if (vector == SYSCALL_VECTOR)
+	if (test_and_set_bit(vector, used_vectors))
 		goto next;
-	for (i = 0; i < NR_IRQ_VECTORS; i++)
-		if (irq_vector[i] == vector)
-			goto next;
 
 	current_vector = vector;
 	current_offset = offset;

@@ -2290,6 +2287,12 @@ static inline void __init check_timer(vo
 
 void __init setup_IO_APIC(void)
 {
+	int i;
+
+	/* Reserve all the system vectors. */
+	for (i = FIRST_SYSTEM_VECTOR; i < NR_VECTORS; i++)
+		set_bit(i, used_vectors);
+
 	enable_IO_APIC();
 
 	if (acpi_ioapic)

@@ -64,6 +64,9 @@
 #include "mach_traps.h"
 
 int panic_on_unrecovered_nmi;
+
+DECLARE_BITMAP(used_vectors, NR_VECTORS);
+EXPORT_SYMBOL_GPL(used_vectors);
 
 asmlinkage int system_call(void);
 

@@ -1156,6 +1159,8 @@ static void __init set_task_gate(unsigne
 
 void __init trap_init(void)
 {
+	int i;
+
 #ifdef CONFIG_EISA
 	void __iomem *p = ioremap(0x0FFFD9, 4);
 	if (readl(p) == 'E'+('I'<<8)+('S'<<16)+('A'<<24)) {

@@ -1215,6 +1220,11 @@ void __init trap_init(void)
 
 	set_system_gate(SYSCALL_VECTOR,&system_call);
 
+	/* Reserve all the builtin and the syscall vector. */
+	for (i = 0; i < FIRST_EXTERNAL_VECTOR; i++)
+		set_bit(i, used_vectors);
+	set_bit(SYSCALL_VECTOR, used_vectors);
+
 	/*
 	 * Should be a barrier for any external CPU state.
 	 */

@@ -45,4 +45,7 @@ void init_IRQ(void);
 void init_IRQ(void);
 void __init native_init_IRQ(void);
 
+/* Interrupt vector management */
+extern DECLARE_BITMAP(used_vectors, NR_VECTORS);
+
 #endif /* _ASM_IRQ_H */

@@ -281,37 +281,47 @@ static int __init init(void)
 	/* First we put the Switcher up in very high virtual memory. */
 	err = map_switcher();
 	if (err)
-		return err;
+		goto out;
 
 	/* Now we set up the pagetable implementation for the Guests. */
 	err = init_pagetables(switcher_page, SHARED_SWITCHER_PAGES);
-	if (err) {
-		unmap_switcher();
-		return err;
-	}
+	if (err)
+		goto unmap;
 
 	/* The I/O subsystem needs some things initialized. */
 	lguest_io_init();
 
+	/* We might need to reserve an interrupt vector. */
+	err = init_interrupts();
+	if (err)
+		goto free_pgtables;
+
 	/* /dev/lguest needs to be registered. */
 	err = lguest_device_init();
-	if (err) {
-		free_pagetables();
-		unmap_switcher();
-		return err;
-	}
+	if (err)
+		goto free_interrupts;
 
 	/* Finally we do some architecture-specific setup. */
 	lguest_arch_host_init();
 
 	/* All good! */
 	return 0;
+
+free_interrupts:
+	free_interrupts();
+free_pgtables:
+	free_pagetables();
+unmap:
+	unmap_switcher();
+out:
+	return err;
 }
 
 /* Cleaning up is just the same code, backwards.  With a little French. */
 static void __exit fini(void)
 {
 	lguest_device_remove();
+	free_interrupts();
 	free_pagetables();
 	unmap_switcher();
 

@@ -530,6 +530,10 @@ int lguest_arch_init_hypercalls(struct l
 	if (put_user(tsc_speed, &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");
+
 	return 0;
 }
 /* Now we've examined the hypercall code; our Guest can make requests.  There

@@ -87,6 +87,7 @@ struct lguest_data lguest_data = {
 	.noirq_start = (u32)lguest_noirq_start,
 	.noirq_end = (u32)lguest_noirq_end,
 	.blocked_interrupts = { 1 }, /* Block timer interrupts */
+	.syscall_vec = SYSCALL_VECTOR,
 };
 static cycle_t clock_base;
 

68 lines not on this page · raw message

@@ -142,6 +142,9 @@ void copy_traps(const struct lguest *lg,
 		const unsigned long *def);
 void guest_set_clockevent(struct lguest *lg, unsigned long delta);
 void init_clockdev(struct lguest *lg);
+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);

@@ -48,6 +48,9 @@ struct lguest_data
 /* Fields initialized by the Guest at boot: */
 	/* Instruction range to suppress interrupts even if enabled */
 	unsigned long noirq_start, noirq_end;
+
+	/* The vector to try to use for system calls (0x40 or 0x80). */
+	unsigned int syscall_vec;
 };
 extern struct lguest_data lguest_data;
 #endif /* __ASSEMBLY__ */

@@ -178,19 +178,16 @@ static void *get_pages(unsigned int num)
 /* To find out where to start we look for the magic Guest string, which marks
  * the code we see in lguest_asm.S.  This is a hack which we are currently
  * plotting to replace with the normal Linux entry point. */
-static unsigned long entry_point(const void *start, const void *end,
-				 unsigned long page_offset)
+static unsigned long entry_point(const void *start, const void *end)
 {
 	const void *p;
 
-	/* The scan gives us the physical starting address.  We want the
-	 * virtual address in this case, and fortunately, we already figured
-	 * out the physical-virtual difference and passed it here in
-	 * "page_offset". */
+	/* The scan gives us the physical starting address.  We boot with
+	 * pagetables set up with virtual and physical the same, so that's
+	 * OK. */
 	for (p = start; p < end; p++)
 		if (memcmp(p, "GenuineLguest", strlen("GenuineLguest")) == 0)
-			return to_guest_phys(p + strlen("GenuineLguest"))
-				+ page_offset;
+			return to_guest_phys(p + strlen("GenuineLguest"));
 
 	errx(1, "Is this image a genuine lguest?");
 }

@@ -224,14 +221,11 @@ static void map_at(int fd, void *addr, u
  * by all modern binaries on Linux including the kernel.
  *
  * The ELF headers give *two* addresses: a physical address, and a virtual
- * address.  The Guest kernel expects to be placed in memory at the physical
- * address, and the page tables set up so it will correspond to that virtual
- * address.  We return the difference between the virtual and physical
- * addresses in the "page_offset" pointer.
+ * address.  We use the physical address; the Guest will map itself to the
+ * virtual address.
  *
  * We return the starting address. */
-static unsigned long map_elf(int elf_fd, const Elf32_Ehdr *ehdr,
-			     unsigned long *page_offset)
+static unsigned long map_elf(int elf_fd, const Elf32_Ehdr *ehdr)
 {
 	void *start = (void *)-1, *end = NULL;
 	Elf32_Phdr phdr[ehdr->e_phnum];

@@ -255,9 +249,6 @@ static unsigned long map_elf(int elf_fd,
 	if (read(elf_fd, phdr, sizeof(phdr)) != sizeof(phdr))
 		err(1, "Reading program headers");
 
-	/* We don't know page_offset yet. */
-	*page_offset = 0;
-
 	/* Try all the headers: there are usually only three.  A read-only one,
 	 * a read-write one, and a "note" section which isn't loadable. */
 	for (i = 0; i < ehdr->e_phnum; i++) {

@@ -268,14 +259,6 @@ static unsigned long map_elf(int elf_fd,
 		verbose("Section %i: size %i addr %p\n",
 			i, phdr[i].p_memsz, (void *)phdr[i].p_paddr);
 
-		/* We expect a simple linear address space: every segment must
-		 * have the same difference between virtual (p_vaddr) and
-		 * physical (p_paddr) address. */
-		if (!*page_offset)
-			*page_offset = phdr[i].p_vaddr - phdr[i].p_paddr;
-		else if (*page_offset != phdr[i].p_vaddr - phdr[i].p_paddr)
-			errx(1, "Page offset of section %i different", i);
-
 		/* We track the first and last address we mapped, so we can
 		 * tell entry_point() where to scan. */
 		if (from_guest_phys(phdr[i].p_paddr) < start)

@@ -288,50 +271,13 @@ static unsigned long map_elf(int elf_fd,
 		       phdr[i].p_offset, phdr[i].p_filesz);
 	}
 
-	return entry_point(start, end, *page_offset);
-}
-
-/*L:170 Prepare to be SHOCKED and AMAZED.  And possibly a trifle nauseated.
- *
- * We know that CONFIG_PAGE_OFFSET sets what virtual address the kernel expects
- * to be.  We don't know what that option was, but we can figure it out
- * approximately by looking at the addresses in the code.  I chose the common
- * case of reading a memory location into the %eax register:
- *
- *  movl <some-address>, %eax
- *
- * This gets encoded as five bytes: "0xA1 <4-byte-address>".  For example,
- * "0xA1 0x18 0x60 0x47 0xC0" reads the address 0xC0476018 into %eax.
- *
- * In this example can guess that the kernel was compiled with
- * CONFIG_PAGE_OFFSET set to 0xC0000000 (it's always a round number).  If the
- * kernel were larger than 16MB, we might see 0xC1 addresses show up, but our
- * kernel isn't that bloated yet.
- *
- * Unfortunately, x86 has variable-length instructions, so finding this
- * particular instruction properly involves writing a disassembler.  Instead,
- * we rely on statistics.  We look for "0xA1" and tally the different bytes
- * which occur 4 bytes later (the "0xC0" in our example above).  When one of
- * those bytes appears three times, we can be reasonably confident that it
- * forms the start of CONFIG_PAGE_OFFSET.
- *
- * This is amazingly reliable. */
-static unsigned long intuit_page_offset(unsigned char *img, unsigned long len)
-{
-	unsigned int i, possibilities[256] = { 0 };
-
-	for (i = 0; i + 4 < len; i++) {
-		/* mov 0xXXXXXXXX,%eax */
-		if (img[i] == 0xA1 && ++possibilities[img[i+4]] > 3)
-			return (unsigned long)img[i+4] << 24;
-	}
-	errx(1, "could not determine page offset");
+	return entry_point(start, end);
 }
 
 /*L:160 Unfortunately the entire ELF image isn't compressed: the segments
  * which need loading are extracted and compressed raw.  This denies us the
  * information we need to make a fully-general loader. */
-static unsigned long unpack_bzimage(int fd, unsigned long *page_offset)
+static unsigned long unpack_bzimage(int fd)
 {
 	gzFile f;
 	int ret, len = 0;

@@ -352,12 +298,7 @@ static unsigned long unpack_bzimage(int 
 
 	verbose("Unpacked size %i addr %p\n", len, img);
 
-	/* Without the ELF header, we can't tell virtual-physical gap.  This is
-	 * CONFIG_PAGE_OFFSET, and people do actually change it.  Fortunately,
-	 * I have a clever way of figuring it out from the code itself.  */
-	*page_offset = intuit_page_offset(img, len);
-
-	return entry_point(img, img + len, *page_offset);
+	return entry_point(img, img + len);
 }
 
 /*L:150 A bzImage, unlike an ELF file, is not meant to be loaded.  You're

@@ -368,7 +309,7 @@ static unsigned long unpack_bzimage(int 
  * The bzImage is formed by putting the decompressing code in front of the
  * compressed kernel code.  So we can simple scan through it looking for the
  * first "gzip" header, and start decompressing from there. */
-static unsigned long load_bzimage(int fd, unsigned long *page_offset)
+static unsigned long load_bzimage(int fd)
 {
 	unsigned char c;
 	int state = 0;

@@ -396,7 +337,7 @@ static unsigned long load_bzimage(int fd
 			if (c != 0x03)
 				state = -1;
 			else
-				return unpack_bzimage(fd, page_offset);
+				return unpack_bzimage(fd);
 		}
 	}
 	errx(1, "Could not find kernel in bzImage");

@@ -405,7 +346,7 @@ static unsigned long load_bzimage(int fd
 /*L:140 Loading the kernel is easy when it's a "vmlinux", but most kernels
  * come wrapped up in the self-decompressing "bzImage" format.  With some funky
  * coding, we can load those, too. */
-static unsigned long load_kernel(int fd, unsigned long *page_offset)
+static unsigned long load_kernel(int fd)
 {
 	Elf32_Ehdr hdr;
 

@@ -415,10 +356,10 @@ static unsigned long load_kernel(int fd,
 
 	/* If it's an ELF file, it starts with "\177ELF" */
 	if (memcmp(hdr.e_ident, ELFMAG, SELFMAG) == 0)
-		return map_elf(fd, &hdr, page_offset);
+		return map_elf(fd, &hdr);
 
 	/* Otherwise we assume it's a bzImage, and try to unpack it */
-	return load_bzimage(fd, page_offset);
+	return load_bzimage(fd);
 }
 
 /* This is a trivial little helper to align pages.  Andi Kleen hated it because

@@ -463,27 +404,20 @@ static unsigned long load_initrd(const c
 	return len;
 }
 
-/* Once we know the address the Guest kernel expects, we can construct simple
- * linear page tables for all of memory which will get the Guest far enough
+/* 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). */
 static unsigned long setup_pagetables(unsigned long mem,
-				      unsigned long initrd_size,
-				      unsigned long page_offset)
+				      unsigned long initrd_size)
 {
 	unsigned long *pgdir, *linear;
 	unsigned int mapped_pages, i, linear_pages;
 	unsigned int ptes_per_page = getpagesize()/sizeof(void *);
 
-	/* Ideally we map all physical memory starting at page_offset.
-	 * However, if page_offset is 0xC0000000 we can only map 1G of physical
-	 * (0xC0000000 + 1G overflows). */
-	if (mem <= -page_offset)
-		mapped_pages = mem/getpagesize();
-	else
-		mapped_pages = -page_offset/getpagesize();
+	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;

@@ -500,11 +434,9 @@ static unsigned long setup_pagetables(un
 	for (i = 0; i < mapped_pages; i++)
 		linear[i] = ((i * getpagesize()) | PAGE_PRESENT);
 
-	/* The top level points to the linear page table pages above.  The
-	 * entry representing page_offset points to the first one, and they
-	 * continue from there. */
+	/* The top level points to the linear page table pages above. */
 	for (i = 0; i < mapped_pages; i += ptes_per_page) {
-		pgdir[(i + page_offset/getpagesize())/ptes_per_page]
+		pgdir[i/ptes_per_page]
 			= ((to_guest_phys(linear) + i*sizeof(void *))
 			   | PAGE_PRESENT);
 	}

@@ -535,15 +467,12 @@ static void concat(char *dst, char *args
 /* 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, the entry point and the page_offset constant for the
- * Guest. */
-static int tell_kernel(unsigned long pgdir, unsigned long start,
-		       unsigned long page_offset)
+ * top level pagetable and the entry point for the Guest. */
+static int tell_kernel(unsigned long pgdir, unsigned long start)
 {
 	unsigned long args[] = { LHREQ_INITIALIZE,
 				 (unsigned long)guest_base,
-				 guest_limit / getpagesize(),
-				 pgdir, start, page_offset };
+				 guest_limit / getpagesize(), pgdir, start };
 	int fd;
 
 	verbose("Guest: %p - %p (%#lx)\n",

@@ -1424,9 +1353,9 @@ static void usage(void)
 /*L:105 The main routine is where the real work begins: */
 int main(int argc, char *argv[])
 {
-	/* Memory, top-level pagetable, code startpoint, PAGE_OFFSET and size
-	 * of the (optional) initrd. */
-	unsigned long mem = 0, pgdir, start, page_offset, initrd_size = 0;
+	/* Memory, top-level pagetable, code startpoint and size of the
+	 * (optional) initrd. */
+	unsigned long mem = 0, pgdir, start, initrd_size = 0;
 	/* A temporary and the /dev/lguest file descriptor. */
 	int i, c, lguest_fd;
 	/* The list of Guest devices, based on command line arguments. */

@@ -1500,8 +1429,7 @@ int main(int argc, char *argv[])
 	setup_console(&device_list);
 
 	/* Now we load the kernel */
-	start = load_kernel(open_or_die(argv[optind+1], O_RDONLY),
-			    &page_offset);
+	start = load_kernel(open_or_die(argv[optind+1], O_RDONLY));
 
 	/* Boot information is stashed at physical address 0 */
 	boot = from_guest_phys(0);

@@ -1518,7 +1446,7 @@ int main(int argc, char *argv[])
 	}
 
 	/* Set up the initial linear pagetables, starting below the initrd. */
-	pgdir = setup_pagetables(mem, initrd_size, page_offset);
+	pgdir = setup_pagetables(mem, initrd_size);
 
 	/* The Linux boot header contains an "E820" memory map: ours is a
 	 * simple, single region. */

@@ -1535,7 +1463,7 @@ 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, page_offset);
+	lguest_fd = tell_kernel(pgdir, start);
 
 	/* We fork off a child process, which wakes the Launcher whenever one
 	 * of the input file descriptors needs attention.  Otherwise we would

@@ -133,6 +133,7 @@ void foo(void)
 #ifdef CONFIG_LGUEST_GUEST
 	BLANK();
 	OFFSET(LGUEST_DATA_irq_enabled, lguest_data, irq_enabled);
+	OFFSET(LGUEST_DATA_pgdir, lguest_data, pgdir);
 	OFFSET(LGUEST_PAGES_host_gdt_desc, lguest_pages, state.host_gdt_desc);
 	OFFSET(LGUEST_PAGES_host_idt_desc, lguest_pages, state.host_idt_desc);
 	OFFSET(LGUEST_PAGES_host_cr3, lguest_pages, state.host_cr3);

@@ -86,6 +86,7 @@ struct lguest_data lguest_data = {
 	.hcall_status = { [0 ... LHCALL_RING_SIZE-1] = 0xFF },
 	.noirq_start = (u32)lguest_noirq_start,
 	.noirq_end = (u32)lguest_noirq_end,
+	.kernel_address = PAGE_OFFSET,
 	.blocked_interrupts = { 1 }, /* Block timer interrupts */
 	.syscall_vec = SYSCALL_VECTOR,
 };

@@ -988,11 +989,7 @@ __init void lguest_init(void *boot)
 
 	/*G:070 Now we've seen all the paravirt_ops, we return to
 	 * lguest_init() where the rest of the fairly chaotic boot setup
-	 * occurs.
-	 *
-	 * The Host expects our first hypercall to tell it where our "struct
-	 * lguest_data" is, so we do that first. */
-	hcall(LHCALL_LGUEST_INIT, __pa(&lguest_data), 0, 0);
+	 * occurs. */
 
 	/* The native boot code sets up initial page tables immediately after
 	 * the kernel itself, and sets init_pg_tables_end so they're not

@@ -181,14 +181,14 @@ static void initialize(struct lguest *lg
 	/* 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)
-	    /* 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))
+	    || get_user(lg->noirq_end, &lg->lguest_data->noirq_end))
 		kill_guest(lg, "bad guest page %p", lg->lguest_data);
 
 	/* We write the current time into the Guest's data page once now. */
 	write_timestamp(lg);
+
+	/* page_tables.c will also do some setup. */
+	page_table_guest_data_init(lg);
 
 	/* 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

@@ -216,9 +216,10 @@ static int emulate_insn(struct lguest *l
 	 * guest_pa just subtracts the Guest's page_offset. */
 	unsigned long physaddr = guest_pa(lg, lg->regs->eip);
 
-	/* The guest_pa() function only works for Guest kernel addresses, but
-	 * that's all we're trying to do anyway. */
-	if (lg->regs->eip < lg->page_offset)
+	/* 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)
 		return 0;
 
 	/* Decoding x86 instructions is icky. */

@@ -62,8 +62,9 @@ static void push_guest_stack(struct lgue
  * it). */
 static void set_guest_interrupt(struct lguest *lg, u32 lo, u32 hi, int has_err)
 {
-	unsigned long gstack;
+	unsigned long gstack, origstack;
 	u32 eflags, ss, irq_enable;
+	unsigned long virtstack;
 
 	/* 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

@@ -71,8 +72,10 @@ static void set_guest_interrupt(struct l
 	if ((lg->regs->ss&0x3) != GUEST_PL) {
 		/* The Guest told us their kernel stack with the SET_STACK
 		 * hypercall: both the virtual address and the segment */
-		gstack = guest_pa(lg, lg->esp1);
+		virtstack = lg->esp1;
 		ss = lg->ss1;
+
+		origstack = gstack = guest_pa(lg, virtstack);
 		/* We push the old stack segment and pointer onto the new
 		 * stack: when the Guest does an "iret" back from the interrupt
 		 * handler the CPU will notice they're dropping privilege

@@ -81,8 +84,10 @@ static void set_guest_interrupt(struct l
 		push_guest_stack(lg, &gstack, lg->regs->esp);
 	} else {
 		/* We're staying on the same Guest (kernel) stack. */
-		gstack = guest_pa(lg, lg->regs->esp);
+		virtstack = lg->regs->esp;
 		ss = lg->regs->ss;
+
+		origstack = gstack = guest_pa(lg, virtstack);
 	}
 
 	/* Remember that we never let the Guest actually disable interrupts, so

@@ -108,7 +113,7 @@ static void set_guest_interrupt(struct l
 	/* 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 = gstack + lg->page_offset;
+	lg->regs->esp = virtstack + (gstack - origstack);
 	lg->regs->cs = (__KERNEL_CS|GUEST_PL);
 	lg->regs->eip = idt_address(lo, hi);
 

@@ -64,7 +64,7 @@ struct lguest
 	/* This provides the offset to the base of guest-physical
 	 * memory in the Launcher. */
 	void __user *mem_base;
-	u32 page_offset;
+	unsigned long kernel_address;
 	u32 cr2;
 	int halted;
 	int ts;

@@ -166,6 +166,8 @@ void map_switcher_in_guest(struct lguest
 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);
+unsigned long guest_pa(struct lguest *lg, unsigned long vaddr);
+void page_table_guest_data_init(struct lguest *lg);
 
 /* <arch>_core.c: */
 void lguest_arch_host_init(void);

@@ -230,9 +232,5 @@ do {								\
 } 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 */

@@ -111,7 +111,7 @@ static ssize_t read(struct file *file, c
 	return run_guest(lg, (unsigned long __user *)user);
 }
 
-/*L:020 The initialization write supplies 5 pointer sized (32 or 64 bit)
+/*L:020 The initialization write supplies 4 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.

@@ -124,12 +124,6 @@ static ssize_t read(struct file *file, c
  * pagetables (which are set up by the Launcher).
  *
  * start: The first instruction to execute ("eip" in x86-speak).
- *
- * page_offset: The PAGE_OFFSET constant in the Guest kernel.  We should
- * probably wean the code off this, but it's a very useful constant!  Any
- * address above this is within the Guest kernel, and any kernel address can
- * quickly converted from physical to virtual by adding PAGE_OFFSET.  It's
- * 0xC0000000 (3G) by default, but it's configurable at kernel build time.
  */
 static int initialize(struct file *file, const unsigned long __user *input)
 {

@@ -137,7 +131,7 @@ static int initialize(struct file *file,
 	 * Guest. */
 	struct lguest *lg;
 	int err;
-	unsigned long args[5];
+	unsigned long args[4];
 
 	/* We grab the Big Lguest lock, which protects against multiple
 	 * simultaneous initializations. */

@@ -162,7 +156,6 @@ static int initialize(struct file *file,
 	/* Populate the easy fields of our "struct lguest" */
 	lg->mem_base = (void __user *)(long)args[0];
 	lg->pfn_limit = args[1];
-	lg->page_offset = args[4];
 
 	/* 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. */

83 lines not on this page · raw message

@@ -36,6 +36,7 @@
  * definition of a gentleman: "someone who is only rude intentionally". */
 #define LGUEST_TRAP_ENTRY 0x1F
 
+#ifndef __ASSEMBLY__
 static inline unsigned long
 hcall(unsigned long call,
       unsigned long arg1, unsigned long arg2, unsigned long arg3)

@@ -71,4 +72,6 @@ struct hcall_args
 	/* These map directly onto eax, ebx, ecx, edx in struct lguest_regs */
 	unsigned long arg0, arg2, arg3, arg1;
 };
+
+#endif /* !__ASSEMBLY__ */
 #endif	/* _I386_LGUEST_HCALL_H */

@@ -44,11 +44,14 @@ struct lguest_data
 	unsigned long reserve_mem;
 	/* KHz for the TSC clock. */
 	u32 tsc_khz;
+	/* Page where the top-level pagetable is */
+	unsigned long pgdir;
 
 /* Fields initialized by the Guest at boot: */
 	/* Instruction range to suppress interrupts even if enabled */
 	unsigned long noirq_start, noirq_end;
-
+	/* Address above which page tables are all identical. */
+	unsigned long kernel_address;
 	/* The vector to try to use for system calls (0x40 or 0x80). */
 	unsigned int syscall_vec;
 };

@@ -0,0 +1,13 @@
+/* Configuration space parsing helpers for virtio.
+ *
+ * The configuration is [type][len][... len bytes ...] fields.
+ *
+ * Copyright 2007 Rusty Russell, IBM Corporation.
+ * GPL v2 or later.
+ */
+#include <linux/err.h>
+#include <linux/virtio.h>
+#include <linux/virtio_config.h>
+#include <linux/bug.h>
+#include <asm/system.h>
+

171 lines not on this page · raw message

@@ -340,4 +340,10 @@ struct parisc_device_id {
 #define PA_HVERSION_ANY_ID	0xffff
 #define PA_SVERSION_ANY_ID	0xffffffff
 
+struct virtio_device_id {
+	__u32 device;
+	__u32 vendor;
+};
+#define VIRTIO_DEV_ANY_ID	0xffffffff
+
 #endif /* LINUX_MOD_DEVICETABLE_H */

110 lines not on this page · raw message

111 lines not on this page · raw message

Index: linux-2.6/include/linux/virtio_config.h
===================================================================
@@ -85,9 +85,9 @@ struct virtio_config_ops
 		     && sizeof(*(v)) != 4 && sizeof(*(v)) != 8);	\
 	if (!_err) {							\
 		switch (sizeof(*(v))) {					\
-		case 2: le16_to_cpus(v); break;				\
-		case 4: le32_to_cpus(v); break;				\
-		case 8: le64_to_cpus(v); break;				\
+		case 2: le16_to_cpus((__u16 *) v); break;		\
+		case 4: le32_to_cpus((__u32 *) v); break;		\
+		case 8: le64_to_cpus((__u64 *) v); break;		\
 		}							\
 	}								\
 	_err;								\