From: Larry Woodman <lwoodman@redhat.com>
Date: Tue, 7 Jul 2009 16:37:25 -0400
Subject: [mm] prevent softlockups in copy_hugetlb_page_range
Message-id: 1246999045.31419.56.camel@dhcp-100-19-198.bos.redhat.com
O-Subject: [RHEL5 patch] prevent softlockups in copy_hugetlb_page_range()
Bugzilla: 508919
RH-Acked-by: Prarit Bhargava <prarit@redhat.com>
RH-Acked-by: Rik van Riel <riel@redhat.com>
RH-Acked-by: Dean Nelson <dnelson@redhat.com>

We made several fixes to copy_hugetlb_page_range(), hugetlb_cow() and
copy_huge_page() to prevent races.  This involved holding spinlocks and
preventing reschedules while those lock are held.  Since we removed the
cond_resched() calls in copy_huge_page() we opened ourselves to
softlockups if the hugepage range being copied is very large.  To
prevent this I've added a cond_resched() call in
copy_hugetlb_page_range() when the pagetable spinlocks have been
dropped.

Fixes BZ 508919

diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index 4ac0821..21f2097 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -432,6 +432,8 @@ int copy_hugetlb_page_range(struct mm_struct *dst, struct mm_struct *src,
 		spin_unlock(&dst->page_table_lock);
 		if (oom)
 			goto nomem;
+		if (forcecow)
+			cond_resched();
 	}
 	return 0;