When dumping vma information the pagemap_read routine calculates

the minimum of what the user asks for and the end of the vma.

Unfortunately the code uses vma->vm_start rather than vma->vm_end

which can result in the end address being before the start, and

a nasty never-ending loop in the kernel.
Diffed against 2.6.23-rc1-mm2
Signed-off-by: Dave Boutcher <boutcher@cs.umn.edu>

---

 fs/proc/task_mmu.c |    2 +-

 1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c

index b2baeab..b12740c 100644

--- a/fs/proc/task_mmu.c

+++ b/fs/proc/task_mmu.c

@@ -671,7 +671,7 @@ static ssize_t pagemap_read(struct file *file, char __user *buf,

 			ret = -EIO;

 			goto out_mm;

 		}

-		vend = min(vma->vm_start - 1, end - 1) + 1;

+		vend = min(vma->vm_end - 1, end - 1) + 1;

 		ret = pagemap_fill(&pm, vend);

 		if (ret || !pm.count)

 			break;

--

1.4.4.2
-

Thanks, Dave. I've added this fix to my local tree. It's still in a

broken state at the moment, so Andrew, feel free to pick this up.
--

Mathematics is the supreme nostalgia of our time.

-