On Mar 22 Williams, Dan J wrote:

On Thu, Mar 22, 2012 at 7:39 AM, Greg Kroah-Hartman
[off-list ref] wrote:

quoted

On Wed, Mar 21, 2012 at 11:32:14PM -0700, Dan Williams wrote:

[..]

quoted

quoted

diff --git a/fs/sysfs/dir.c b/fs/sysfs/dir.c
index 7fdf6a7..86521ee 100644
--- a/fs/sysfs/dir.c
+++ b/fs/sysfs/dir.c

@@ -714,6 +714,9 @@ int sysfs_create_dir(struct kobject * kobj)

      else
              parent_sd = &sysfs_root;

+     if (!parent_sd)
+             return -ENOENT;
+
      if (sysfs_ns_type(parent_sd))
              ns = kobj->ktype->namespace(kobj);
      type = sysfs_read_ns_type(kobj);

So what happens if this is true?  Does this patch fix the oops?

This patch downgrades the oops by turning it into a device_add()
failure, but the patches that *fix* this warning are here [1] and here
[2].

[...]

[1]: http://marc.info/?l=linux-scsi&m=133239707903443&w=2
[2]: http://marc.info/?l=linux-scsi&m=133239709603452&w=2

Isn't this something which is to be accomplished by counting references to
the parent device?
-- 
Stefan Richter
-=====-===-- --== =-==-
http://arcgraph.de/sr/