[PATCH] x86_64: Use function pointers to call DMA mapping functions

AK: I hacked Muli's original patch a lot and there were a lot
of changes - all bugs are probably to blame on me now.
There were also some changes in the fall back behaviour
for swiotlb - in particular it doesn't try to use GFP_DMA
now anymore. Also all DMA mapping operations use the
same core dma_alloc_coherent code with proper fallbacks now.
And various other changes and cleanups.

Known problems: iommu=force swiotlb=force together breaks
                needs more testing.

This patch cleans up x86_64's DMA mapping dispatching code. Right now
we have three possible IOMMU types: AGP GART, swiotlb and nommu, and
in the future we will also have Xen's x86_64 swiotlb and other HW
IOMMUs for x86_64. In order to support all of them cleanly, this
patch:

- introduces a struct dma_mapping_ops with function pointers for each
  of the DMA mapping operations of gart (AMD HW IOMMU), swiotlb
  (software IOMMU) and nommu (no IOMMU).

- gets rid of:

  if (swiotlb)
      return swiotlb_xxx();

- PCI_DMA_BUS_IS_PHYS is now checked against the dma_ops being set
This makes swiotlb faster by avoiding double copying in some cases.

Signed-Off-By: Muli Ben-Yehuda <[email protected]>
Signed-Off-By: Jon D. Mason <[email protected]>
Signed-off-by: Andi Kleen <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

arch/x86_64/Kconfig

@@ -351,32 +351,24 @@ config HPET_EMULATE_RTC
 	depends on HPET_TIMER && RTC=y
 
 config GART_IOMMU
-	bool "IOMMU support"
+	bool "K8 GART IOMMU support"
 	default y
+	select SWIOTLB
 	depends on PCI
 	help
 	  Support the IOMMU. Needed to run systems with more than 3GB of memory
 	  properly with 32-bit PCI devices that do not support DAC (Double Address
 	  Cycle). The IOMMU can be turned off at runtime with the iommu=off parameter.
 	  Normally the kernel will take the right choice by itself.
-	  This option includes a driver for the AMD Opteron/Athlon64 IOMMU
-	  and a software emulation used on some other systems.
+	  This option includes a driver for the AMD Opteron/Athlon64 northbridge IOMMU
+	  and a software emulation used on other systems.
 	  If unsure, say Y.
 
 # need this always enabled with GART_IOMMU for the VIA workaround
 config SWIOTLB
-       bool
-       depends on GART_IOMMU
-       default y
-
-config DUMMY_IOMMU
 	bool
-	depends on !GART_IOMMU && !SWIOTLB
 	default y
-	help
-	  Don't use IOMMU code. This will cause problems when you have more than 4GB
-	  of memory and any 32-bit devices. Don't turn on unless you know what you
-	  are doing.
+	depends on GART_IOMMU
 
 config X86_MCE
 	bool "Machine check support" if EMBEDDED

arch/x86_64/kernel/Makefile

@@ -8,7 +8,7 @@ obj-y	:= process.o signal.o entry.o traps.o irq.o \
 		ptrace.o time.o ioport.o ldt.o setup.o i8259.o sys_x86_64.o \
 		x8664_ksyms.o i387.o syscall.o vsyscall.o \
 		setup64.o bootflag.o e820.o reboot.o quirks.o i8237.o \
-		dmi_scan.o
+		dmi_scan.o pci-dma.o pci-nommu.o
 
 obj-$(CONFIG_X86_MCE)         += mce.o
 obj-$(CONFIG_X86_MCE_INTEL)	+= mce_intel.o
@@ -29,7 +29,7 @@ obj-$(CONFIG_SOFTWARE_SUSPEND)	+= suspend_asm.o
 obj-$(CONFIG_CPU_FREQ)		+= cpufreq/
 obj-$(CONFIG_EARLY_PRINTK)	+= early_printk.o
 obj-$(CONFIG_GART_IOMMU)	+= pci-gart.o aperture.o
-obj-$(CONFIG_DUMMY_IOMMU)	+= pci-nommu.o pci-dma.o
+obj-$(CONFIG_SWIOTLB)		+= pci-swiotlb.o
 obj-$(CONFIG_KPROBES)		+= kprobes.o
 obj-$(CONFIG_X86_PM_TIMER)	+= pmtimer.o
 

arch/x86_64/kernel/pci-dma.c

@@ -8,53 +8,259 @@
 #include <linux/pci.h>
 #include <linux/module.h>
 #include <asm/io.h>
+#include <asm/proto.h>
 
-/* Map a set of buffers described by scatterlist in streaming
- * mode for DMA.  This is the scatter-gather version of the
- * above pci_map_single interface.  Here the scatter gather list
- * elements are each tagged with the appropriate dma address
- * and length.  They are obtained via sg_dma_{address,length}(SG).
- *
- * NOTE: An implementation may be able to use a smaller number of
- *       DMA address/length pairs than there are SG table elements.
- *       (for example via virtual mapping capabilities)
- *       The routine returns the number of addr/length pairs actually
- *       used, at most nents.
- *
- * Device ownership issues as mentioned above for pci_map_single are
- * the same here.
+int iommu_merge __read_mostly = 0;
+EXPORT_SYMBOL(iommu_merge);
+
+dma_addr_t bad_dma_address __read_mostly;
+EXPORT_SYMBOL(bad_dma_address);
+
+/* This tells the BIO block layer to assume merging. Default to off
+   because we cannot guarantee merging later. */
+int iommu_bio_merge __read_mostly = 0;
+EXPORT_SYMBOL(iommu_bio_merge);
+
+int iommu_sac_force __read_mostly = 0;
+EXPORT_SYMBOL(iommu_sac_force);
+
+int no_iommu __read_mostly;
+#ifdef CONFIG_IOMMU_DEBUG
+int panic_on_overflow __read_mostly = 1;
+int force_iommu __read_mostly = 1;
+#else
+int panic_on_overflow __read_mostly = 0;
+int force_iommu __read_mostly= 0;
+#endif
+
+/* Dummy device used for NULL arguments (normally ISA). Better would
+   be probably a smaller DMA mask, but this is bug-to-bug compatible
+   to i386. */
+struct device fallback_dev = {
+	.bus_id = "fallback device",
+	.coherent_dma_mask = 0xffffffff,
+	.dma_mask = &fallback_dev.coherent_dma_mask,
+};
+
+/* Allocate DMA memory on node near device */
+noinline static void *
+dma_alloc_pages(struct device *dev, gfp_t gfp, unsigned order)
+{
+	struct page *page;
+	int node;
+	if (dev->bus == &pci_bus_type)
+		node = pcibus_to_node(to_pci_dev(dev)->bus);
+	else
+		node = numa_node_id();
+	page = alloc_pages_node(node, gfp, order);
+	return page ? page_address(page) : NULL;
+}
+
+/*
+ * Allocate memory for a coherent mapping.
  */
-int dma_map_sg(struct device *hwdev, struct scatterlist *sg,
-	       int nents, int direction)
+void *
+dma_alloc_coherent(struct device *dev, size_t size, dma_addr_t *dma_handle,
+		   gfp_t gfp)
 {
-	int i;
-
-	BUG_ON(direction == DMA_NONE);
- 	for (i = 0; i < nents; i++ ) {
-		struct scatterlist *s = &sg[i];
-		BUG_ON(!s->page); 
-		s->dma_address = virt_to_bus(page_address(s->page) +s->offset);
-		s->dma_length = s->length;
+	void *memory;
+	unsigned long dma_mask = 0;
+	u64 bus;
+
+	if (!dev)
+		dev = &fallback_dev;
+	dma_mask = dev->coherent_dma_mask;
+	if (dma_mask == 0)
+		dma_mask = 0xffffffff;
+
+	/* Kludge to make it bug-to-bug compatible with i386. i386
+	   uses the normal dma_mask for alloc_coherent. */
+	dma_mask &= *dev->dma_mask;
+
+	/* Why <=? Even when the mask is smaller than 4GB it is often
+	   larger than 16MB and in this case we have a chance of
+	   finding fitting memory in the next higher zone first. If
+	   not retry with true GFP_DMA. -AK */
+	if (dma_mask <= 0xffffffff)
+		gfp |= GFP_DMA32;
+
+ again:
+	memory = dma_alloc_pages(dev, gfp, get_order(size));
+	if (memory == NULL)
+		return NULL;
+
+	{
+		int high, mmu;
+		bus = virt_to_bus(memory);
+	        high = (bus + size) >= dma_mask;
+		mmu = high;
+		if (force_iommu && !(gfp & GFP_DMA))
+			mmu = 1;
+		else if (high) {
+			free_pages((unsigned long)memory,
+				   get_order(size));
+
+			/* Don't use the 16MB ZONE_DMA unless absolutely
+			   needed. It's better to use remapping first. */
+			if (dma_mask < 0xffffffff && !(gfp & GFP_DMA)) {
+				gfp = (gfp & ~GFP_DMA32) | GFP_DMA;
+				goto again;
+			}
+
+			if (dma_ops->alloc_coherent)
+				return dma_ops->alloc_coherent(dev, size,
+							   dma_handle, gfp);
+			return NULL;
+		}
+
+		memset(memory, 0, size);
+		if (!mmu) {
+			*dma_handle = virt_to_bus(memory);
+			return memory;
+		}
+	}
+
+	if (dma_ops->alloc_coherent) {
+		free_pages((unsigned long)memory, get_order(size));
+		gfp &= ~(GFP_DMA|GFP_DMA32);
+		return dma_ops->alloc_coherent(dev, size, dma_handle, gfp);
+	}
+
+	if (dma_ops->map_simple) {
+		*dma_handle = dma_ops->map_simple(dev, memory,
+					      size,
+					      PCI_DMA_BIDIRECTIONAL);
+		if (*dma_handle != bad_dma_address)
+			return memory;
 	}
-	return nents;
-}
 
-EXPORT_SYMBOL(dma_map_sg);
+	if (panic_on_overflow)
+		panic("dma_alloc_coherent: IOMMU overflow by %lu bytes\n",size);
+	free_pages((unsigned long)memory, get_order(size));
+	return NULL;
+}
+EXPORT_SYMBOL(dma_alloc_coherent);
 
-/* Unmap a set of streaming mode DMA translations.
- * Again, cpu read rules concerning calls here are the same as for
- * pci_unmap_single() above.
+/*
+ * Unmap coherent memory.
+ * The caller must ensure that the device has finished accessing the mapping.
  */
-void dma_unmap_sg(struct device *dev, struct scatterlist *sg,
-		  int nents, int dir)
+void dma_free_coherent(struct device *dev, size_t size,
+			 void *vaddr, dma_addr_t bus)
+{
+	if (dma_ops->unmap_single)
+		dma_ops->unmap_single(dev, bus, size, 0);
+	free_pages((unsigned long)vaddr, get_order(size));
+}
+EXPORT_SYMBOL(dma_free_coherent);
+
+int dma_supported(struct device *dev, u64 mask)
+{
+	if (dma_ops->dma_supported)
+		return dma_ops->dma_supported(dev, mask);
+
+	/* Copied from i386. Doesn't make much sense, because it will
+	   only work for pci_alloc_coherent.
+	   The caller just has to use GFP_DMA in this case. */
+        if (mask < 0x00ffffff)
+                return 0;
+
+	/* Tell the device to use SAC when IOMMU force is on.  This
+	   allows the driver to use cheaper accesses in some cases.
+
+	   Problem with this is that if we overflow the IOMMU area and
+	   return DAC as fallback address the device may not handle it
+	   correctly.
+
+	   As a special case some controllers have a 39bit address
+	   mode that is as efficient as 32bit (aic79xx). Don't force
+	   SAC for these.  Assume all masks <= 40 bits are of this
+	   type. Normally this doesn't make any difference, but gives
+	   more gentle handling of IOMMU overflow. */
+	if (iommu_sac_force && (mask >= 0xffffffffffULL)) {
+		printk(KERN_INFO "%s: Force SAC with mask %Lx\n", dev->bus_id,mask);
+		return 0;
+	}
+
+	return 1;
+}
+EXPORT_SYMBOL(dma_supported);
+
+int dma_set_mask(struct device *dev, u64 mask)
 {
-	int i;
-	for (i = 0; i < nents; i++) { 
-		struct scatterlist *s = &sg[i];
-		BUG_ON(s->page == NULL); 
-		BUG_ON(s->dma_address == 0); 
-		dma_unmap_single(dev, s->dma_address, s->dma_length, dir);
-	} 
+	if (!dev->dma_mask || !dma_supported(dev, mask))
+		return -EIO;
+	*dev->dma_mask = mask;
+	return 0;
 }
+EXPORT_SYMBOL(dma_set_mask);
+
+/* iommu=[size][,noagp][,off][,force][,noforce][,leak][,memaper[=order]][,merge]
+         [,forcesac][,fullflush][,nomerge][,biomerge]
+   size  set size of iommu (in bytes)
+   noagp don't initialize the AGP driver and use full aperture.
+   off   don't use the IOMMU
+   leak  turn on simple iommu leak tracing (only when CONFIG_IOMMU_LEAK is on)
+   memaper[=order] allocate an own aperture over RAM with size 32MB^order.
+   noforce don't force IOMMU usage. Default.
+   force  Force IOMMU.
+   merge  Do lazy merging. This may improve performance on some block devices.
+          Implies force (experimental)
+   biomerge Do merging at the BIO layer. This is more efficient than merge,
+            but should be only done with very big IOMMUs. Implies merge,force.
+   nomerge Don't do SG merging.
+   forcesac For SAC mode for masks <40bits  (experimental)
+   fullflush Flush IOMMU on each allocation (default)
+   nofullflush Don't use IOMMU fullflush
+   allowed  overwrite iommu off workarounds for specific chipsets.
+   soft	 Use software bounce buffering (default for Intel machines)
+   noaperture Don't touch the aperture for AGP.
+*/
+__init int iommu_setup(char *p)
+{
+    iommu_merge = 1;
 
-EXPORT_SYMBOL(dma_unmap_sg);
+    while (*p) {
+	    if (!strncmp(p,"off",3))
+		    no_iommu = 1;
+	    /* gart_parse_options has more force support */
+	    if (!strncmp(p,"force",5))
+		    force_iommu = 1;
+	    if (!strncmp(p,"noforce",7)) {
+		    iommu_merge = 0;
+		    force_iommu = 0;
+	    }
+
+	    if (!strncmp(p, "biomerge",8)) {
+		    iommu_bio_merge = 4096;
+		    iommu_merge = 1;
+		    force_iommu = 1;
+	    }
+	    if (!strncmp(p, "panic",5))
+		    panic_on_overflow = 1;
+	    if (!strncmp(p, "nopanic",7))
+		    panic_on_overflow = 0;
+	    if (!strncmp(p, "merge",5)) {
+		    iommu_merge = 1;
+		    force_iommu = 1;
+	    }
+	    if (!strncmp(p, "nomerge",7))
+		    iommu_merge = 0;
+	    if (!strncmp(p, "forcesac",8))
+		    iommu_sac_force = 1;
+
+#ifdef CONFIG_SWIOTLB
+	    if (!strncmp(p, "soft",4))
+		    swiotlb = 1;
+#endif
+
+#ifdef CONFIG_GART_IOMMU
+	    gart_parse_options(p);
+#endif
+
+	    p += strcspn(p, ",");
+	    if (*p == ',')
+		    ++p;
+    }
+    return 1;
+}