[PATCH] x86_64: Clean and enhance up K8 northbridge access code

 - Factor out the duplicated access/cache code into a single file
   * Shared between i386/x86-64.
 - Share flush code between AGP and IOMMU
   * Fix a bug: AGP didn't wait for end of flush before
 - Drop 8 northbridges limit and allocate dynamically
 - Add lock to serialize AGP and IOMMU GART flushes
 - Add PCI ID for next AMD northbridge
 - Random related cleanups

The old K8 NUMA discovery code is unchanged. New systems
should all use SRAT for this.

Cc: "Navin Boppuri" <navin.boppuri@newisys.com>
Cc: Dave Jones <davej@redhat.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

1 files changed, 118 insertions, 0 deletions

diff --git a/arch/x86_64/kernel/k8.c b/arch/x86_64/kernel/k8.c
new file mode 100644
index 00000000000..6416682d33d
--- /dev/null
+++ b/arch/x86_64/kernel/k8.c
@@ -0,0 +1,118 @@
+/*
+ * Shared support code for AMD K8 northbridges and derivates.
+ * Copyright 2006 Andi Kleen, SUSE Labs. Subject to GPLv2.
+ */
+#include <linux/gfp.h>
+#include <linux/types.h>
+#include <linux/init.h>
+#include <linux/errno.h>
+#include <linux/module.h>
+#include <linux/spinlock.h>
+#include <asm/k8.h>
+
+int num_k8_northbridges;
+EXPORT_SYMBOL(num_k8_northbridges);
+
+static u32 *flush_words;
+
+struct pci_device_id k8_nb_ids[] = {
+	{ PCI_DEVICE(PCI_VENDOR_ID_AMD, 0x1103) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_AMD, 0x1203) },
+	{}
+};
+EXPORT_SYMBOL(k8_nb_ids);
+
+struct pci_dev **k8_northbridges;
+EXPORT_SYMBOL(k8_northbridges);
+
+static struct pci_dev *next_k8_northbridge(struct pci_dev *dev)
+{
+	do {
+		dev = pci_get_device(PCI_ANY_ID, PCI_ANY_ID, dev);
+		if (!dev)
+			break;
+	} while (!pci_match_id(&k8_nb_ids[0], dev));
+	return dev;
+}
+
+int cache_k8_northbridges(void)
+{
+	int i;
+	struct pci_dev *dev;
+	if (num_k8_northbridges)
+		return 0;
+
+	num_k8_northbridges = 0;
+	dev = NULL;
+	while ((dev = next_k8_northbridge(dev)) != NULL)
+		num_k8_northbridges++;
+
+	k8_northbridges = kmalloc((num_k8_northbridges + 1) * sizeof(void *),
+				  GFP_KERNEL);
+	if (!k8_northbridges)
+		return -ENOMEM;
+
+	flush_words = kmalloc(num_k8_northbridges * sizeof(u32), GFP_KERNEL);
+	if (!flush_words) {
+		kfree(k8_northbridges);
+		return -ENOMEM;
+	}
+
+	dev = NULL;
+	i = 0;
+	while ((dev = next_k8_northbridge(dev)) != NULL) {
+		k8_northbridges[i++] = dev;
+		pci_read_config_dword(dev, 0x9c, &flush_words[i]);
+	}
+	k8_northbridges[i] = NULL;
+	return 0;
+}
+EXPORT_SYMBOL_GPL(cache_k8_northbridges);
+
+/* Ignores subdevice/subvendor but as far as I can figure out
+   they're useless anyways */
+int __init early_is_k8_nb(u32 device)
+{
+	struct pci_device_id *id;
+	u32 vendor = device & 0xffff;
+	device >>= 16;
+	for (id = k8_nb_ids; id->vendor; id++)
+		if (vendor == id->vendor && device == id->device)
+			return 1;
+	return 0;
+}
+
+void k8_flush_garts(void)
+{
+	int flushed, i;
+	unsigned long flags;
+	static DEFINE_SPINLOCK(gart_lock);
+
+	/* Avoid races between AGP and IOMMU. In theory it's not needed
+	   but I'm not sure if the hardware won't lose flush requests
+	   when another is pending. This whole thing is so expensive anyways
+	   that it doesn't matter to serialize more. -AK */
+	spin_lock_irqsave(&gart_lock, flags);
+	flushed = 0;
+	for (i = 0; i < num_k8_northbridges; i++) {
+		pci_write_config_dword(k8_northbridges[i], 0x9c,
+				       flush_words[i]|1);
+		flushed++;
+	}
+	for (i = 0; i < num_k8_northbridges; i++) {
+		u32 w;
+		/* Make sure the hardware actually executed the flush*/
+		for (;;) {
+			pci_read_config_dword(k8_northbridges[i],
+					      0x9c, &w);
+			if (!(w & 1))
+				break;
+			cpu_relax();
+		}
+	}
+	spin_unlock_irqrestore(&gart_lock, flags);
+	if (!flushed)
+		printk("nothing to flush?\n");
+}
+EXPORT_SYMBOL_GPL(k8_flush_garts);
+