PCI resource problems caused by improper address rounding

[Chuck Ebbert <cebbert@redhat.com>]

Looks like a commit that I can't find in git due to the arch merge
has broken PCI address assignment. This patch by Richard Henderson
against 2.6.23 fixes it for x86_64:

--- linux-2.6.23.x86_64/arch/x86_64/kernel/e820.c	2007-10-09 13:31:38.000000000 -0700
+++ linux-2.6.23.x86_64-rth/arch/x86_64/kernel/e820.c	2007-12-15 12:37:44.000000000 -0800
@@ -718,8 +718,8 @@ __init void e820_setup_gap(void)
 	while ((gapsize >> 4) > round)
 		round += round;
 	/* Fun with two's complement */
-	pci_mem_start = (gapstart + round) & -round;
+	pci_mem_start = (gapstart + round - 1) & -round;
 
 	printk(KERN_INFO "Allocating PCI resources starting at %lx (gap: %lx:%lx)\n",
 		pci_mem_start, gapstart, gapsize);


Here is the original changeset, taken from the Mercurial repo. It was
merged in 2.6.14:

# HG changeset patch
# User Daniel Ritz <daniel.ritz@gmx.ch>
# Date 1126304746 -700
# Node ID 51367d6e0b839be0b425a8f67c29f625b670f126
# Parent f4852c862b04efc9f8e2c7913191f5f7d140d895
[PATCH] Update PCI IOMEM allocation start

This fixes the problem with "Averatec 6240 pcmcia_socket0: unable to
apply power", which was due to the CardBus IOMEM register region being
allocated at an address that was actually inside the RAM window that had
been reserved for video frame-buffers in an UMA setup.

The BIOS _should_ have marked that region reserved in the e820 memory
descriptor tables, but did not.

It is fixed by rounding up the default starting address of PCI memory
allocations, so that we leave a bigger gap after the final known memory
location.  The amount of rounding depends on how big the unused memory
gap is that we can allocate IOMEM from.

Based on example code by Linus.

Acked-by: Greg KH <greg@kroah.com>
Acked-by: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

committer: Linus Torvalds <torvalds@g5.osdl.org> 1126304746 -0700


--- a/arch/i386/kernel/setup.c	Fri Sep 09 22:28:40 2005 +0011
+++ b/arch/i386/kernel/setup.c	Fri Sep 09 22:37:26 2005 +0011
@@ -1300,7 +1300,7 @@ legacy_init_iomem_resources(struct resou
  */
 static void __init register_memory(void)
 {
-	unsigned long gapstart, gapsize;
+	unsigned long gapstart, gapsize, round;
 	unsigned long long last;
 	int	      i;
 
@@ -1345,14 +1345,14 @@ static void __init register_memory(void)
 	}
 
 	/*
-	 * Start allocating dynamic PCI memory a bit into the gap,
-	 * aligned up to the nearest megabyte.
-	 *
-	 * Question: should we try to pad it up a bit (do something
-	 * like " + (gapsize >> 3)" in there too?). We now have the
-	 * technology.
+	 * See how much we want to round up: start off with
+	 * rounding to the next 1MB area.
 	 */
-	pci_mem_start = (gapstart + 0xfffff) & ~0xfffff;
+	round = 0x100000;
+	while ((gapsize >> 4) > round)
+		round += round;
+	/* Fun with two's complement */
+	pci_mem_start = (gapstart + round) & -round;
 
 	printk("Allocating PCI resources starting at %08lx (gap: %08lx:%08lx)\n",
 		pci_mem_start, gapstart, gapsize);
--- a/arch/x86_64/kernel/e820.c	Fri Sep 09 22:28:40 2005 +0011
+++ b/arch/x86_64/kernel/e820.c	Fri Sep 09 22:37:26 2005 +0011
@@ -567,7 +567,7 @@ unsigned long pci_mem_start = 0xaeedbabe
  */
 __init void e820_setup_gap(void)
 {
-	unsigned long gapstart, gapsize;
+	unsigned long gapstart, gapsize, round;
 	unsigned long last;
 	int i;
 	int found = 0;
@@ -604,14 +604,14 @@ __init void e820_setup_gap(void)
 	}
 
 	/*
-	 * Start allocating dynamic PCI memory a bit into the gap,
-	 * aligned up to the nearest megabyte.
-	 *
-	 * Question: should we try to pad it up a bit (do something
-	 * like " + (gapsize >> 3)" in there too?). We now have the
-	 * technology.
+	 * See how much we want to round up: start off with
+	 * rounding to the next 1MB area.
 	 */
-	pci_mem_start = (gapstart + 0xfffff) & ~0xfffff;
+	round = 0x100000;
+	while ((gapsize >> 4) > round)
+		round += round;
+	/* Fun with two's complement */
+	pci_mem_start = (gapstart + round) & -round;
 
 	printk(KERN_INFO "Allocating PCI resources starting at %lx (gap: %lx:%lx)\n",
 		pci_mem_start, gapstart, gapsize);