Re: [PATCH] farsync: fix invalid memory accesses in fst_add_one() and fst_init_card()

From: Alexey Khoroshilov <redacted>
Date: Tue, 08 Jul 2014 18:40:32 -0400

On 08.07.2014 18:20, David Miller wrote:

quoted

From: Alexey Khoroshilov <redacted>
Date: Sat,  5 Jul 2014 03:35:50 +0400

quoted

-                }
+			card->nports = i;
+			return (card->nports == 0) ? err : 0;
+		}

I don't think this is the right thing to do.

This will cause the caller to not free the IRQ or any of the
other resources.

My understanding of the existing code is to proceed if at least one port
is available.
So I return error code if no ports available at all, otherwise
initialization continues and can succeed.
If something else goes wrong, all resources are deallocated.

Do you suggest to return error code unconditionally?

Yes, that is my suggestion.

I'm also weary of so many changes to a driver that gets so little
usage and probably next to no testing at all.