From mboxrd@z Thu Jan 1 00:00:00 1970 From: Denis Joseph Barrow Subject: I think I figured out the another bug which manifests itself when low_latency is set to 0 Date: Tue, 13 Jan 2009 11:07:00 +0100 Message-ID: <496C67C4.4070501@option.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit To: Alan Cox , Linux USB kernel mailing list , Linux netdev Mailing list Return-path: Received: from mailer2.option.com ([81.246.70.163]:8104 "EHLO mailer2.option.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752688AbZAMKHE (ORCPT ); Tue, 13 Jan 2009 05:07:04 -0500 Sender: netdev-owner@vger.kernel.org List-ID: Hi Alan, Admittedly there always was a bug when tty_insert_flip_string runs out of memory when low_latency is set to 0, After your change I now need a callback when tty_insert_flip_string has room for more data & I currently I don't get one, This is needed to execute the code in hso_unthrottle_tasklet i.e. I do flow control by not submitting read urbs to the modem when TTY_THROTTLED is set. The only way I can see for getting this callback without modifying the tty layer is setting the TTY_THROTTLED bit if tty_insert_flip_string returns 0, I hope this works. I'm personally a bit frightened that we may have more bugs/problems that will manifest by setting low_latency to 0, especially for the guys here running diagnostics over the serial port requiring very high throughput. -- best regards, D.J. Barrow Linux Kernel Developer Option NV, Gaston Geenslaan 14, 3001 Leuven, Belgium T: +32 16 311 621 F: +32 16 207 164 d.barow@option.com www.option.com Disclaimer: http://www.option.com/company/disclaimer.shtml