Fix double mutex_init bug in send code

It was possible to cause a kernel panic in the send code by 
initializing an already-initialized mutex, if a record was created 
with type DATA, destroyed with a different type (bypassing the 
mutex_destroy call) and then re-allocated as a DATA record again.

We tweak the logic to not change the type of a record once it has 
been created, avoiding the issue.

Reviewed-by: Brian Behlendorf <[email protected]>
Reviewed-by: Jorgen Lundman <[email protected]>
Signed-off-by: Paul Dagnelie <[email protected]>
Closes #10374 

0 files changed, 0 insertions, 0 deletions