Re: [linux-usb-devel] [PATCH] base/class.c: prevent ooops due to insert/remove race (v3)

[Mark Lord <lkml@rtr.ca>]

(resending .. somebody trimmed the CC: list earlier)

Greg KH wrote:
>>>> Mark Lord wrote:
>>>>> ..
>>>>>
>>>>> While doing insert/remove (quickly) tests on USB,
>>>>> I managed to trigger an Oops on 2.6.23.8 on a call
>>>>> to strlen() in make_class_name().
...

> I'll hold off on adding this patch for now.
..

Why?

Bugs that result in Oops in core code (class.c) can bite
just about any subsystem that does hotplug, and should get
prompt attention.  Or so one might think.

* * * *

Patch reproduced below for full CC:

While doing insert/remove (quickly) tests on USB, I managed to trigger
an Oops on 2.6.23.1 on the call to strlen() in make_class_name().

This patch prevents the oops, but still keeps the bug visible.

There is still the larger problem of the overall race
that caused this in the first place, but much of the rest
of the code in class.c appears to also do NULL checks to
avoid Oops'ing, so this continues the tradition.

Signed-off-by:  Mark Lord <mlord@pobox.com> ---

Patch applies to both 2.6.24 and 2.6.23.

--- old/drivers/base/class.c    2007-11-29 10:51:43.000000000 -0500
+++ linux/drivers/base/class.c    2007-11-29 13:00:15.000000000 -0500
@@ -352,9 +352,22 @@
char *make_class_name(const char *name, struct kobject *kobj)
{
    char *class_name;
+    const char *kname;
    int size;

-    size = strlen(name) + strlen(kobject_name(kobj)) + 2;
+    /* Rapidly inserting/removing a USB device (others?)
+     * can trigger an Oops on the strlen() call.
+     * Cause unknown yet, so prevent the Oops
+     * but don't mask the issue.
+     */
+    kname = kobject_name(kobj);
+    if (!kname || !name) {
+        printk(KERN_ERR "make_class_name: name=%p kname=%p\n",
+                        name, kname);
+        BUG_ON(1);
+        return NULL;
+    }
+    size = strlen(name) + strlen(kname) + 2;

    class_name = kmalloc(size, GFP_KERNEL);
    if (!class_name)