via-cuda: Fix re-initialization of reply_ptr and reading_reply
authorFinn Thain <fthain@telegraphics.com.au>
Sun, 1 Jan 2017 00:56:26 +0000 (19:56 -0500)
committerMichael Ellerman <mpe@ellerman.id.au>
Tue, 7 Feb 2017 05:56:20 +0000 (16:56 +1100)
When reading_reply is set, reply_ptr points into an adb_request struct.
Conversely, when reply_ptr instead points into the global cuda_rbuf,
reading_reply must be false.

Unfortunately, this rule can be violated because re-initialization
of reply_ptr and reading_reply presently depends on the TREQ input.

Fix this by re-initializing reply_ptr and reading_reply as soon as they
are known to be invalid.

Tested-by: Stan Johnson <userm57@yahoo.com>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
drivers/macintosh/via-cuda.c

index 1cf1467cf6e54f6af2a7f6e5520b2a7cf834bb7e..ae3da6b952295e36e4023b354e46b1736d9d8f41 100644 (file)
@@ -592,6 +592,7 @@ cuda_interrupt(int irq, void *arg)
            }
            current_req = req->next;
            complete = 1;
+           reading_reply = 0;
        } else {
            /* This is tricky. We must break the spinlock to call
             * cuda_input. However, doing so means we might get
@@ -603,11 +604,10 @@ cuda_interrupt(int irq, void *arg)
            ibuf_len = reply_ptr - cuda_rbuf;
            memcpy(ibuf, cuda_rbuf, ibuf_len);
        }
+       reply_ptr = cuda_rbuf;
        if (TREQ_asserted(status)) {
            assert_TIP();
            cuda_state = reading;
-           reply_ptr = cuda_rbuf;
-           reading_reply = 0;
        } else {
            cuda_state = idle;
            cuda_start();