IB: cxgb3: delay freeing mem untill entirely done with it

Sure, it's just the pointer value we use, but the coverity checker
complains about a use-after-free bug and it really does seem cleaner
to delay freeing until we are entirely done with the memory. So,
rearrange the code to move the kfree() later untill we are completely
done.
Trivial and harmless, but nice IMHO.

Signed-off-by: Jesper Juhl <jj@chaosbits.net>
Reviewed-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>

diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c
index 0bdf09a..145d82a 100644 (file)
--- a/drivers/infiniband/hw/cxgb3/iwch_provider.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c
@@ -783,8 +783,8 @@ static int iwch_dealloc_mw(struct ib_mw *mw)
        mmid = (mw->rkey) >> 8;
        cxio_deallocate_window(&rhp->rdev, mhp->attr.stag);
        remove_handle(rhp, &rhp->mmidr, mmid);
-       kfree(mhp);
        PDBG("%s ib_mw %p mmid 0x%x ptr %p\n", __func__, mw, mmid, mhp);
+       kfree(mhp);
        return 0;
 }