summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorRobin Getz <robin.getz@analog.com>2009-11-05 15:44:44 +0000
committerMike Frysinger <vapier@gentoo.org>2009-12-15 00:15:28 -0500
commitdbc5e6989e2261c965bae5269d26ed1641e1534c (patch)
treea2f8f1f8d4aaae9e189bf7a71ac27798ddd7f2a6
parente92d32fcb9925884a0ad9e1ad8f10d837d7de7a6 (diff)
Blackfin: don't walk VMAs when oopsing
If we're double faulting, then we have to assume the VMAs are not safe as broken pointers here will prevent full trace output for the double fault. Shouldn't be a big problem though as rarely is a double fault caused by code in userspace. Signed-off-by: Robin Getz <robin.getz@analog.com> Signed-off-by: Mike Frysinger <vapier@gentoo.org>
-rw-r--r--arch/blackfin/kernel/traps.c9
1 files changed, 9 insertions, 0 deletions
diff --git a/arch/blackfin/kernel/traps.c b/arch/blackfin/kernel/traps.c
index 89835ac58a8a..427294c47f1b 100644
--- a/arch/blackfin/kernel/traps.c
+++ b/arch/blackfin/kernel/traps.c
@@ -119,6 +119,15 @@ static void decode_address(char *buf, unsigned long address)
return;
}
+ /*
+ * Don't walk any of the vmas if we are oopsing, it has been known
+ * to cause problems - corrupt vmas (kernel crashes) cause double faults
+ */
+ if (oops_in_progress) {
+ strcat(buf, "/* kernel dynamic memory (maybe user-space) */");
+ return;
+ }
+
/* looks like we're off in user-land, so let's walk all the
* mappings of all our processes and see if we can't be a whee
* bit more specific