Pull rework-memory-attribute-aliasing into release branch

7 files changed, 359 insertions, 73 deletions

diff --git a/Documentation/ia64/aliasing.txt b/Documentation/ia64/aliasing.txt
new file mode 100644
index 000000000000..38f9a52d1820
--- /dev/null
+++ b/Documentation/ia64/aliasing.txt
@@ -0,0 +1,208 @@
+	         MEMORY ATTRIBUTE ALIASING ON IA-64
+
+			   Bjorn Helgaas
+		       <bjorn.helgaas@hp.com>
+			    May 4, 2006
+
+
+MEMORY ATTRIBUTES
+
+    Itanium supports several attributes for virtual memory references.
+    The attribute is part of the virtual translation, i.e., it is
+    contained in the TLB entry.  The ones of most interest to the Linux
+    kernel are:
+
+	WB		Write-back (cacheable)
+	UC		Uncacheable
+	WC		Write-coalescing
+
+    System memory typically uses the WB attribute.  The UC attribute is
+    used for memory-mapped I/O devices.  The WC attribute is uncacheable
+    like UC is, but writes may be delayed and combined to increase
+    performance for things like frame buffers.
+
+    The Itanium architecture requires that we avoid accessing the same
+    page with both a cacheable mapping and an uncacheable mapping[1].
+
+    The design of the chipset determines which attributes are supported
+    on which regions of the address space.  For example, some chipsets
+    support either WB or UC access to main memory, while others support
+    only WB access.
+
+MEMORY MAP
+
+    Platform firmware describes the physical memory map and the
+    supported attributes for each region.  At boot-time, the kernel uses
+    the EFI GetMemoryMap() interface.  ACPI can also describe memory
+    devices and the attributes they support, but Linux/ia64 currently
+    doesn't use this information.
+
+    The kernel uses the efi_memmap table returned from GetMemoryMap() to
+    learn the attributes supported by each region of physical address
+    space.  Unfortunately, this table does not completely describe the
+    address space because some machines omit some or all of the MMIO
+    regions from the map.
+
+    The kernel maintains another table, kern_memmap, which describes the
+    memory Linux is actually using and the attribute for each region.
+    This contains only system memory; it does not contain MMIO space.
+
+    The kern_memmap table typically contains only a subset of the system
+    memory described by the efi_memmap.  Linux/ia64 can't use all memory
+    in the system because of constraints imposed by the identity mapping
+    scheme.
+
+    The efi_memmap table is preserved unmodified because the original
+    boot-time information is required for kexec.
+
+KERNEL IDENTITY MAPPINGS
+
+    Linux/ia64 identity mappings are done with large pages, currently
+    either 16MB or 64MB, referred to as "granules."  Cacheable mappings
+    are speculative[2], so the processor can read any location in the
+    page at any time, independent of the programmer's intentions.  This
+    means that to avoid attribute aliasing, Linux can create a cacheable
+    identity mapping only when the entire granule supports cacheable
+    access.
+
+    Therefore, kern_memmap contains only full granule-sized regions that
+    can referenced safely by an identity mapping.
+
+    Uncacheable mappings are not speculative, so the processor will
+    generate UC accesses only to locations explicitly referenced by
+    software.  This allows UC identity mappings to cover granules that
+    are only partially populated, or populated with a combination of UC
+    and WB regions.
+
+USER MAPPINGS
+
+    User mappings are typically done with 16K or 64K pages.  The smaller
+    page size allows more flexibility because only 16K or 64K has to be
+    homogeneous with respect to memory attributes.
+
+POTENTIAL ATTRIBUTE ALIASING CASES
+
+    There are several ways the kernel creates new mappings:
+
+    mmap of /dev/mem
+
+	This uses remap_pfn_range(), which creates user mappings.  These
+	mappings may be either WB or UC.  If the region being mapped
+	happens to be in kern_memmap, meaning that it may also be mapped
+	by a kernel identity mapping, the user mapping must use the same
+	attribute as the kernel mapping.
+
+	If the region is not in kern_memmap, the user mapping should use
+	an attribute reported as being supported in the EFI memory map.
+
+	Since the EFI memory map does not describe MMIO on some
+	machines, this should use an uncacheable mapping as a fallback.
+
+    mmap of /sys/class/pci_bus/.../legacy_mem
+
+	This is very similar to mmap of /dev/mem, except that legacy_mem
+	only allows mmap of the one megabyte "legacy MMIO" area for a
+	specific PCI bus.  Typically this is the first megabyte of
+	physical address space, but it may be different on machines with
+	several VGA devices.
+
+	"X" uses this to access VGA frame buffers.  Using legacy_mem
+	rather than /dev/mem allows multiple instances of X to talk to
+	different VGA cards.
+
+	The /dev/mem mmap constraints apply.
+
+	However, since this is for mapping legacy MMIO space, WB access
+	does not make sense.  This matters on machines without legacy
+	VGA support: these machines may have WB memory for the entire
+	first megabyte (or even the entire first granule).
+
+	On these machines, we could mmap legacy_mem as WB, which would
+	be safe in terms of attribute aliasing, but X has no way of
+	knowing that it is accessing regular memory, not a frame buffer,
+	so the kernel should fail the mmap rather than doing it with WB.
+
+    read/write of /dev/mem
+
+	This uses copy_from_user(), which implicitly uses a kernel
+	identity mapping.  This is obviously safe for things in
+	kern_memmap.
+
+	There may be corner cases of things that are not in kern_memmap,
+	but could be accessed this way.  For example, registers in MMIO
+	space are not in kern_memmap, but could be accessed with a UC
+	mapping.  This would not cause attribute aliasing.  But
+	registers typically can be accessed only with four-byte or
+	eight-byte accesses, and the copy_from_user() path doesn't allow
+	any control over the access size, so this would be dangerous.
+
+    ioremap()
+
+	This returns a kernel identity mapping for use inside the
+	kernel.
+
+	If the region is in kern_memmap, we should use the attribute
+	specified there.  Otherwise, if the EFI memory map reports that
+	the entire granule supports WB, we should use that (granules
+	that are partially reserved or occupied by firmware do not appear
+	in kern_memmap).  Otherwise, we should use a UC mapping.
+
+PAST PROBLEM CASES
+
+    mmap of various MMIO regions from /dev/mem by "X" on Intel platforms
+
+      The EFI memory map may not report these MMIO regions.
+
+      These must be allowed so that X will work.  This means that
+      when the EFI memory map is incomplete, every /dev/mem mmap must
+      succeed.  It may create either WB or UC user mappings, depending
+      on whether the region is in kern_memmap or the EFI memory map.
+
+    mmap of 0x0-0xA0000 /dev/mem by "hwinfo" on HP sx1000 with VGA enabled
+
+      See https://bugzilla.novell.com/show_bug.cgi?id=140858.
+
+      The EFI memory map reports the following attributes:
+        0x00000-0x9FFFF WB only
+        0xA0000-0xBFFFF UC only (VGA frame buffer)
+        0xC0000-0xFFFFF WB only
+
+      This mmap is done with user pages, not kernel identity mappings,
+      so it is safe to use WB mappings.
+
+      The kernel VGA driver may ioremap the VGA frame buffer at 0xA0000,
+      which will use a granule-sized UC mapping covering 0-0xFFFFF.  This
+      granule covers some WB-only memory, but since UC is non-speculative,
+      the processor will never generate an uncacheable reference to the
+      WB-only areas unless the driver explicitly touches them.
+
+    mmap of 0x0-0xFFFFF legacy_mem by "X"
+
+      If the EFI memory map reports this entire range as WB, there
+      is no VGA MMIO hole, and the mmap should fail or be done with
+      a WB mapping.
+
+      There's no easy way for X to determine whether the 0xA0000-0xBFFFF
+      region is a frame buffer or just memory, so I think it's best to
+      just fail this mmap request rather than using a WB mapping.  As
+      far as I know, there's no need to map legacy_mem with WB
+      mappings.
+
+      Otherwise, a UC mapping of the entire region is probably safe.
+      The VGA hole means the region will not be in kern_memmap.  The
+      HP sx1000 chipset doesn't support UC access to the memory surrounding
+      the VGA hole, but X doesn't need that area anyway and should not
+      reference it.
+
+    mmap of 0xA0000-0xBFFFF legacy_mem by "X" on HP sx1000 with VGA disabled
+
+      The EFI memory map reports the following attributes:
+        0x00000-0xFFFFF WB only (no VGA MMIO hole)
+
+      This is a special case of the previous case, and the mmap should
+      fail for the same reason as above.
+
+NOTES
+
+    [1] SDM rev 2.2, vol 2, sec 4.4.1.
+    [2] SDM rev 2.2, vol 2, sec 4.4.6.
diff --git a/arch/ia64/kernel/efi.c b/arch/ia64/kernel/efi.c
index 12cfedce73b1..c33d0ba7e300 100644
--- a/arch/ia64/kernel/efi.c
+++ b/arch/ia64/kernel/efi.c
@@ -8,6 +8,8 @@
  * Copyright (C) 1999-2003 Hewlett-Packard Co.
  *	David Mosberger-Tang <davidm@hpl.hp.com>
  *	Stephane Eranian <eranian@hpl.hp.com>
+ * (c) Copyright 2006 Hewlett-Packard Development Company, L.P.
+ *	Bjorn Helgaas <bjorn.helgaas@hp.com>
  *
  * All EFI Runtime Services are not implemented yet as EFI only
  * supports physical mode addressing on SoftSDV. This is to be fixed
@@ -622,28 +624,20 @@ efi_get_iobase (void)
 	return 0;
 }
 
-static efi_memory_desc_t *
-efi_memory_descriptor (unsigned long phys_addr)
+static struct kern_memdesc *
+kern_memory_descriptor (unsigned long phys_addr)
 {
-	void *efi_map_start, *efi_map_end, *p;
-	efi_memory_desc_t *md;
-	u64 efi_desc_size;
-
-	efi_map_start = __va(ia64_boot_param->efi_memmap);
-	efi_map_end   = efi_map_start + ia64_boot_param->efi_memmap_size;
-	efi_desc_size = ia64_boot_param->efi_memdesc_size;
+	struct kern_memdesc *md;
 
-	for (p = efi_map_start; p < efi_map_end; p += efi_desc_size) {
-		md = p;
-
-		if (phys_addr - md->phys_addr < (md->num_pages << EFI_PAGE_SHIFT))
+	for (md = kern_memmap; md->start != ~0UL; md++) {
+		if (phys_addr - md->start < (md->num_pages << EFI_PAGE_SHIFT))
 			 return md;
 	}
 	return 0;
 }
 
-static int
-efi_memmap_has_mmio (void)
+static efi_memory_desc_t *
+efi_memory_descriptor (unsigned long phys_addr)
 {
 	void *efi_map_start, *efi_map_end, *p;
 	efi_memory_desc_t *md;
@@ -656,8 +650,8 @@ efi_memmap_has_mmio (void)
 	for (p = efi_map_start; p < efi_map_end; p += efi_desc_size) {
 		md = p;
 
-		if (md->type == EFI_MEMORY_MAPPED_IO)
-			return 1;
+		if (phys_addr - md->phys_addr < (md->num_pages << EFI_PAGE_SHIFT))
+			 return md;
 	}
 	return 0;
 }
@@ -683,71 +677,125 @@ efi_mem_attributes (unsigned long phys_addr)
 }
 EXPORT_SYMBOL(efi_mem_attributes);
 
-/*
- * Determines whether the memory at phys_addr supports the desired
- * attribute (WB, UC, etc).  If this returns 1, the caller can safely
- * access size bytes at phys_addr with the specified attribute.
- */
-int
-efi_mem_attribute_range (unsigned long phys_addr, unsigned long size, u64 attr)
+u64
+efi_mem_attribute (unsigned long phys_addr, unsigned long size)
 {
 	unsigned long end = phys_addr + size;
 	efi_memory_desc_t *md = efi_memory_descriptor(phys_addr);
+	u64 attr;
+
+	if (!md)
+		return 0;
+
+	/*
+	 * EFI_MEMORY_RUNTIME is not a memory attribute; it just tells
+	 * the kernel that firmware needs this region mapped.
+	 */
+	attr = md->attribute & ~EFI_MEMORY_RUNTIME;
+	do {
+		unsigned long md_end = efi_md_end(md);
+
+		if (end <= md_end)
+			return attr;
+
+		md = efi_memory_descriptor(md_end);
+		if (!md || (md->attribute & ~EFI_MEMORY_RUNTIME) != attr)
+			return 0;
+	} while (md);
+	return 0;
+}
+
+u64
+kern_mem_attribute (unsigned long phys_addr, unsigned long size)
+{
+	unsigned long end = phys_addr + size;
+	struct kern_memdesc *md;
+	u64 attr;
 
 	/*
-	 * Some firmware doesn't report MMIO regions in the EFI memory
-	 * map.  The Intel BigSur (a.k.a. HP i2000) has this problem.
-	 * On those platforms, we have to assume UC is valid everywhere.
+	 * This is a hack for ioremap calls before we set up kern_memmap.
+	 * Maybe we should do efi_memmap_init() earlier instead.
 	 */
-	if (!md || (md->attribute & attr) != attr) {
-		if (attr == EFI_MEMORY_UC && !efi_memmap_has_mmio())
-			return 1;
+	if (!kern_memmap) {
+		attr = efi_mem_attribute(phys_addr, size);
+		if (attr & EFI_MEMORY_WB)
+			return EFI_MEMORY_WB;
 		return 0;
 	}
 
+	md = kern_memory_descriptor(phys_addr);
+	if (!md)
+		return 0;
+
+	attr = md->attribute;
 	do {
-		unsigned long md_end = efi_md_end(md);
+		unsigned long md_end = kmd_end(md);
 
 		if (end <= md_end)
-			return 1;
+			return attr;
 
-		md = efi_memory_descriptor(md_end);
-		if (!md || (md->attribute & attr) != attr)
+		md = kern_memory_descriptor(md_end);
+		if (!md || md->attribute != attr)
 			return 0;
 	} while (md);
 	return 0;
 }
+EXPORT_SYMBOL(kern_mem_attribute);
 
-/*
- * For /dev/mem, we only allow read & write system calls to access
- * write-back memory, because read & write don't allow the user to
- * control access size.
- */
 int
 valid_phys_addr_range (unsigned long phys_addr, unsigned long size)
 {
-	return efi_mem_attribute_range(phys_addr, size, EFI_MEMORY_WB);
+	u64 attr;
+
+	/*
+	 * /dev/mem reads and writes use copy_to_user(), which implicitly
+	 * uses a granule-sized kernel identity mapping.  It's really
+	 * only safe to do this for regions in kern_memmap.  For more
+	 * details, see Documentation/ia64/aliasing.txt.
+	 */
+	attr = kern_mem_attribute(phys_addr, size);
+	if (attr & EFI_MEMORY_WB || attr & EFI_MEMORY_UC)
+		return 1;
+	return 0;
 }
 
-/*
- * We allow mmap of anything in the EFI memory map that supports
- * either write-back or uncacheable access.  For uncacheable regions,
- * the supported access sizes are system-dependent, and the user is
- * responsible for using the correct size.
- *
- * Note that this doesn't currently allow access to hot-added memory,
- * because that doesn't appear in the boot-time EFI memory map.
- */
 int
 valid_mmap_phys_addr_range (unsigned long phys_addr, unsigned long size)
 {
-	if (efi_mem_attribute_range(phys_addr, size, EFI_MEMORY_WB))
-		return 1;
+	/*
+	 * MMIO regions are often missing from the EFI memory map.
+	 * We must allow mmap of them for programs like X, so we
+	 * currently can't do any useful validation.
+	 */
+	return 1;
+}
 
-	if (efi_mem_attribute_range(phys_addr, size, EFI_MEMORY_UC))
-		return 1;
+pgprot_t
+phys_mem_access_prot(struct file *file, unsigned long pfn, unsigned long size,
+		     pgprot_t vma_prot)
+{
+	unsigned long phys_addr = pfn << PAGE_SHIFT;
+	u64 attr;
 
-	return 0;
+	/*
+	 * For /dev/mem mmap, we use user mappings, but if the region is
+	 * in kern_memmap (and hence may be covered by a kernel mapping),
+	 * we must use the same attribute as the kernel mapping.
+	 */
+	attr = kern_mem_attribute(phys_addr, size);
+	if (attr & EFI_MEMORY_WB)
+		return pgprot_cacheable(vma_prot);
+	else if (attr & EFI_MEMORY_UC)
+		return pgprot_noncached(vma_prot);
+
+	/*
+	 * Some chipsets don't support UC access to memory.  If
+	 * WB is supported, we prefer that.
+	 */
+	if (efi_mem_attribute(phys_addr, size) & EFI_MEMORY_WB)
+		return pgprot_cacheable(vma_prot);
+
+	return pgprot_noncached(vma_prot);
 }
 
 int __init
diff --git a/arch/ia64/mm/ioremap.c b/arch/ia64/mm/ioremap.c
index 643ccc6960ce..07bd02b6c372 100644
--- a/arch/ia64/mm/ioremap.c
+++ b/arch/ia64/mm/ioremap.c
@@ -11,6 +11,7 @@
 #include <linux/module.h>
 #include <linux/efi.h>
 #include <asm/io.h>
+#include <asm/meminit.h>
 
 static inline void __iomem *
 __ioremap (unsigned long offset, unsigned long size)
@@ -21,16 +22,29 @@ __ioremap (unsigned long offset, unsigned long size)
 void __iomem *
 ioremap (unsigned long offset, unsigned long size)
 {
-	if (efi_mem_attribute_range(offset, size, EFI_MEMORY_WB))
-		return phys_to_virt(offset);
+	u64 attr;
+	unsigned long gran_base, gran_size;
 
-	if (efi_mem_attribute_range(offset, size, EFI_MEMORY_UC))
+	/*
+	 * For things in kern_memmap, we must use the same attribute
+	 * as the rest of the kernel.  For more details, see
+	 * Documentation/ia64/aliasing.txt.
+	 */
+	attr = kern_mem_attribute(offset, size);
+	if (attr & EFI_MEMORY_WB)
+		return phys_to_virt(offset);
+	else if (attr & EFI_MEMORY_UC)
 		return __ioremap(offset, size);
 
 	/*
-	 * Someday this should check ACPI resources so we
-	 * can do the right thing for hot-plugged regions.
+	 * Some chipsets don't support UC access to memory.  If
+	 * WB is supported for the whole granule, we prefer that.
 	 */
+	gran_base = GRANULEROUNDDOWN(offset);
+	gran_size = GRANULEROUNDUP(offset + size) - gran_base;
+	if (efi_mem_attribute(gran_base, gran_size) & EFI_MEMORY_WB)
+		return phys_to_virt(offset);
+
 	return __ioremap(offset, size);
 }
 EXPORT_SYMBOL(ioremap);
@@ -38,6 +52,9 @@ EXPORT_SYMBOL(ioremap);
 void __iomem *
 ioremap_nocache (unsigned long offset, unsigned long size)
 {
+	if (kern_mem_attribute(offset, size) & EFI_MEMORY_WB)
+		return 0;
+
 	return __ioremap(offset, size);
 }
 EXPORT_SYMBOL(ioremap_nocache);
diff --git a/arch/ia64/pci/pci.c b/arch/ia64/pci/pci.c
index ab829a22f8a4..30d148f34042 100644
--- a/arch/ia64/pci/pci.c
+++ b/arch/ia64/pci/pci.c
@@ -645,18 +645,31 @@ char *ia64_pci_get_legacy_mem(struct pci_bus *bus)
 int
 pci_mmap_legacy_page_range(struct pci_bus *bus, struct vm_area_struct *vma)
 {
+	unsigned long size = vma->vm_end - vma->vm_start;
+	pgprot_t prot;
 	char *addr;
 
+	/*
+	 * Avoid attribute aliasing.  See Documentation/ia64/aliasing.txt
+	 * for more details.
+	 */
+	if (!valid_mmap_phys_addr_range(vma->vm_pgoff << PAGE_SHIFT, size))
+		return -EINVAL;
+	prot = phys_mem_access_prot(NULL, vma->vm_pgoff, size,
+				    vma->vm_page_prot);
+	if (pgprot_val(prot) != pgprot_val(pgprot_noncached(vma->vm_page_prot)))
+		return -EINVAL;
+
 	addr = pci_get_legacy_mem(bus);
 	if (IS_ERR(addr))
 		return PTR_ERR(addr);
 
 	vma->vm_pgoff += (unsigned long)addr >> PAGE_SHIFT;
-	vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
+	vma->vm_page_prot = prot;
 	vma->vm_flags |= (VM_SHM | VM_RESERVED | VM_IO);
 
 	if (remap_pfn_range(vma, vma->vm_start, vma->vm_pgoff,
-			    vma->vm_end - vma->vm_start, vma->vm_page_prot))
+			    size, vma->vm_page_prot))
 		return -EAGAIN;
 
 	return 0;
diff --git a/include/asm-ia64/io.h b/include/asm-ia64/io.h
index c2e3742108bb..781ee2c7e8c3 100644
--- a/include/asm-ia64/io.h
+++ b/include/asm-ia64/io.h
@@ -88,6 +88,7 @@ phys_to_virt (unsigned long address)
 }
 
 #define ARCH_HAS_VALID_PHYS_ADDR_RANGE
+extern u64 kern_mem_attribute (unsigned long phys_addr, unsigned long size);
 extern int valid_phys_addr_range (unsigned long addr, size_t count); /* efi.c */
 extern int valid_mmap_phys_addr_range (unsigned long addr, size_t count);
 
diff --git a/include/asm-ia64/pgtable.h b/include/asm-ia64/pgtable.h
index eaac08d5e0bd..228981cadf8f 100644
--- a/include/asm-ia64/pgtable.h
+++ b/include/asm-ia64/pgtable.h
@@ -316,22 +316,20 @@ ia64_phys_addr_valid (unsigned long addr)
 #define pte_mkhuge(pte)		(__pte(pte_val(pte)))
 
 /*
- * Macro to a page protection value as "uncacheable".  Note that "protection" is really a
- * misnomer here as the protection value contains the memory attribute bits, dirty bits,
- * and various other bits as well.
+ * Make page protection values cacheable, uncacheable, or write-
+ * combining.  Note that "protection" is really a misnomer here as the
+ * protection value contains the memory attribute bits, dirty bits, and
+ * various other bits as well.
  */
+#define pgprot_cacheable(prot)		__pgprot((pgprot_val(prot) & ~_PAGE_MA_MASK) | _PAGE_MA_WB)
 #define pgprot_noncached(prot)		__pgprot((pgprot_val(prot) & ~_PAGE_MA_MASK) | _PAGE_MA_UC)
-
-/*
- * Macro to make mark a page protection value as "write-combining".
- * Note that "protection" is really a misnomer here as the protection
- * value contains the memory attribute bits, dirty bits, and various
- * other bits as well.  Accesses through a write-combining translation
- * works bypasses the caches, but does allow for consecutive writes to
- * be combined into single (but larger) write transactions.
- */
 #define pgprot_writecombine(prot)	__pgprot((pgprot_val(prot) & ~_PAGE_MA_MASK) | _PAGE_MA_WC)
 
+struct file;
+extern pgprot_t phys_mem_access_prot(struct file *file, unsigned long pfn,
+				     unsigned long size, pgprot_t vma_prot);
+#define __HAVE_PHYS_MEM_ACCESS_PROT
+
 static inline unsigned long
 pgd_index (unsigned long address)
 {
diff --git a/include/linux/efi.h b/include/linux/efi.h
index e203613d3aec..66d621dbcb6c 100644
--- a/include/linux/efi.h
+++ b/include/linux/efi.h
@@ -294,6 +294,7 @@ extern void efi_enter_virtual_mode (void);	/* switch EFI to virtual mode, if pos
 extern u64 efi_get_iobase (void);
 extern u32 efi_mem_type (unsigned long phys_addr);
 extern u64 efi_mem_attributes (unsigned long phys_addr);
+extern u64 efi_mem_attribute (unsigned long phys_addr, unsigned long size);
 extern int efi_mem_attribute_range (unsigned long phys_addr, unsigned long size,
 				    u64 attr);
 extern int __init efi_uart_console_only (void);