From: Hugh Dickins <hugh@veritas.com>
Date: Tue, 17 May 2005 04:53:09 +0000 (-0700)
Subject: [PATCH] fix impossible VmallocChunk
X-Git-Tag: v2.6.12-rc5~164
X-Git-Url: https://openfabrics.org/gitweb/?a=commitdiff_plain;h=64d13c00cf1f7c3d2c1ff449e2a0500ab568d319;p=~emulex%2Finfiniband.git

[PATCH] fix impossible VmallocChunk

VmallocTotal: 34359738367 kB
VmallocUsed:    266288 kB
VmallocChunk: 18014366299193295 kB
is unsettling - x86_64 and some other architectures keep a separate address
range for modules in vmalloc's vmlist, which /proc/meminfo should pass over.

Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
---

diff --git a/fs/proc/mmu.c b/fs/proc/mmu.c
index a7041038ad5..25d2d9c6e32 100644
--- a/fs/proc/mmu.c
+++ b/fs/proc/mmu.c
@@ -50,13 +50,23 @@ void get_vmalloc_info(struct vmalloc_info *vmi)
 		read_lock(&vmlist_lock);
 
 		for (vma = vmlist; vma; vma = vma->next) {
+			unsigned long addr = (unsigned long) vma->addr;
+
+			/*
+			 * Some archs keep another range for modules in vmlist
+			 */
+			if (addr < VMALLOC_START)
+				continue;
+			if (addr >= VMALLOC_END)
+				break;
+
 			vmi->used += vma->size;
 
-			free_area_size = (unsigned long) vma->addr - prev_end;
+			free_area_size = addr - prev_end;
 			if (vmi->largest_chunk < free_area_size)
 				vmi->largest_chunk = free_area_size;
 
-			prev_end = vma->size + (unsigned long) vma->addr;
+			prev_end = vma->size + addr;
 		}
 
 		if (VMALLOC_END - prev_end > vmi->largest_chunk)