resource: Fix generic page_is_ram() for partial RAM pages

Wu Fengguang · H. Peter Anvin · commit 37b99dd5372c · 2010-03-01T10:18:32.000-08:00
The System RAM walk shall skip partial RAM pages and avoid calling
func() on them. So that page_is_ram() return 0 for a partial RAM page.

In particular, it shall not call func() with len=0.
This fixes a boot time bug reported by Sachin and root caused by Thomas:

&gt; &gt;&gt;&gt; WARNING: at arch/x86/mm/ioremap.c:111 __ioremap_caller+0x169/0x2f1()
&gt; &gt;&gt;&gt; Hardware name: BladeCenter LS21 -[79716AA]-
&gt; &gt;&gt;&gt; Modules linked in:
&gt; &gt;&gt;&gt; Pid: 0, comm: swapper Not tainted 2.6.33-git6-autotest #1
&gt; &gt;&gt;&gt; Call Trace:
&gt; &gt;&gt;&gt; [&lt;ffffffff81047cff&gt;] ? __ioremap_caller+0x169/0x2f1
&gt; &gt;&gt;&gt; [&lt;ffffffff81063b7d&gt;] warn_slowpath_common+0x77/0xa4
&gt; &gt;&gt;&gt; [&lt;ffffffff81063bb9&gt;] warn_slowpath_null+0xf/0x11
&gt; &gt;&gt;&gt; [&lt;ffffffff81047cff&gt;] __ioremap_caller+0x169/0x2f1
&gt; &gt;&gt;&gt; [&lt;ffffffff813747a3&gt;] ? acpi_os_map_memory+0x12/0x1b
&gt; &gt;&gt;&gt; [&lt;ffffffff81047f10&gt;] ioremap_nocache+0x12/0x14
&gt; &gt;&gt;&gt; [&lt;ffffffff813747a3&gt;] acpi_os_map_memory+0x12/0x1b
&gt; &gt;&gt;&gt; [&lt;ffffffff81282fa0&gt;] acpi_tb_verify_table+0x29/0x5b
&gt; &gt;&gt;&gt; [&lt;ffffffff812827f0&gt;] acpi_load_tables+0x39/0x15a
&gt; &gt;&gt;&gt; [&lt;ffffffff8191c8f8&gt;] acpi_early_init+0x60/0xf5
&gt; &gt;&gt;&gt; [&lt;ffffffff818f2cad&gt;] start_kernel+0x397/0x3a7
&gt; &gt;&gt;&gt; [&lt;ffffffff818f2295&gt;] x86_64_start_reservations+0xa5/0xa9
&gt; &gt;&gt;&gt; [&lt;ffffffff818f237a&gt;] x86_64_start_kernel+0xe1/0xe8
&gt; &gt;&gt;&gt; ---[ end trace 4eaa2a86a8e2da22 ]---
&gt; &gt;&gt;&gt; ioremap reserve_memtype failed -22

The return code is -EINVAL, so it failed in the is_ram check, which is
not too surprising

&gt; BIOS-provided physical RAM map:
&gt;  BIOS-e820: 0000000000000000 - 000000000009c000 (usable)
&gt;  BIOS-e820: 000000000009c000 - 00000000000a0000 (reserved)
&gt;  BIOS-e820: 00000000000e0000 - 0000000000100000 (reserved)
&gt;  BIOS-e820: 0000000000100000 - 00000000cffa3900 (usable)
&gt;  BIOS-e820: 00000000cffa3900 - 00000000cffa7400 (ACPI data)

The ACPI data is not starting on a page boundary and neither does the
usable RAM area end on a page boundary. Very useful !

&gt; ACPI: DSDT 00000000cffa3900 036CE (v01 IBM    SERLEWIS 00001000 INTL 20060912)

ACPI is trying to map DSDT at cffa3900, which results in a check
vs. cffa3000 which is the relevant page boundary. The generic is_ram
check correctly identifies that as RAM because it's in the usable
resource area. The old e820 based is_ram check does not take
overlapping resource areas into account. That's why it works.

CC: Sachin Sant &lt;sachinp@in.ibm.com&gt;
CC: Thomas Gleixner &lt;tglx@linutronix.de&gt;
CC: KAMEZAWA Hiroyuki &lt;kamezawa.hiroyu@jp.fujitsu.com&gt;
Signed-off-by: Wu Fengguang &lt;fengguang.wu@intel.com&gt;
LKML-Reference: &lt;20100301135551.GA9998@localhost&gt;
Signed-off-by: H. Peter Anvin &lt;hpa@zytor.com&gt;
diff --git a/kernel/resource.c b/kernel/resource.c
@@ -274,7 +274,7 @@ int walk_system_ram_range(unsigned long start_pfn, unsigned long nr_pages,
 		void *arg, int (*func)(unsigned long, unsigned long, void *))
 {
 	struct resource res;
-	unsigned long pfn, len;
+	unsigned long pfn, end_pfn;
 	u64 orig_end;
 	int ret = -1;
 
@@ -284,9 +284,10 @@ int walk_system_ram_range(unsigned long start_pfn, unsigned long nr_pages,
 	orig_end = res.end;
 	while ((res.start < res.end) &&
 		(find_next_system_ram(&res, "System RAM") >= 0)) {
-		pfn = (unsigned long)(res.start >> PAGE_SHIFT);
-		len = (unsigned long)((res.end + 1 - res.start) >> PAGE_SHIFT);
-		ret = (*func)(pfn, len, arg);
+		pfn = (res.start + PAGE_SIZE - 1) >> PAGE_SHIFT;
+		end_pfn = (res.end + 1) >> PAGE_SHIFT;
+		if (end_pfn > pfn)
+		    ret = (*func)(pfn, end_pfn - pfn, arg);
 		if (ret)
 			break;
 		res.start = res.end + 1;
