Re: [PATCH v2] Bluetooth: Fix l2cap_tx_window_full

Hi Andrei,

* Andrei Emeltchenko [off-list ref] [2012-02-08 15:59:11 +0200]:

Hi all,

On Wed, Feb 08, 2012 at 11:54:54AM -0200, Ulisses Furquim wrote:

quoted

Hi Luiz,

On Wed, Feb 8, 2012 at 11:47 AM, Luiz Augusto von Dentz
[off-list ref] wrote:

quoted

From: Luiz Augusto von Dentz <redacted>

l2cap_tx_window_full is not checking the window limit properly, first it
computes based on sequence numbers which doesn't take into account the
ReqSeq and always assume 64 not the real window size.

To fix this now it just checks if the number of unacked frames is >= of
tx window which is much simpler.

Signed-off-by: Luiz Augusto von Dentz <redacted>
---
Add debug

 include/net/bluetooth/l2cap.h |   10 +++-------
 1 files changed, 3 insertions(+), 7 deletions(-)

diff --git a/include/net/bluetooth/l2cap.h b/include/net/bluetooth/l2cap.h
index 42fdbb8..bcfddb2 100644
--- a/include/net/bluetooth/l2cap.h
+++ b/include/net/bluetooth/l2cap.h

@@ -661,14 +661,10 @@ static inline __u16 __next_seq(struct l2cap_chan *chan, __u16 seq)

 static inline int l2cap_tx_window_full(struct l2cap_chan *ch)
 {
-       int sub;
+       BT_DBG("chan %p unacked %d tx_win %d", ch, ch->unacked_frames,
+                                                       ch->remote_tx_win);

-       sub = (ch->next_tx_seq - ch->expected_ack_seq) % 64;
-
-       if (sub < 0)
-               sub += 64;
-
-       return sub == ch->remote_tx_win;
+       return ch->unacked_frames >= ch->remote_tx_win;
 }

 static inline __u16 __get_reqseq(struct l2cap_chan *chan, __u32 ctrl)
--
1.7.7.6

Looks good to me.

Same here, I am wondering why it was done in a such complex way? Maybe
there is some logic behind?

I followed the specification, next_tx_seq minus expecpted_ack_seq is the
definition of the current size of the tx_window.

Anyway,

Acked-by: Gustavo F. Padovan <redacted>

	Gustavo