From mboxrd@z Thu Jan 1 00:00:00 1970 From: Stefan Wahren Subject: Re: serial/mxs-auart: Unable to get interrupt counter from user space Date: Sun, 24 Jun 2012 11:14:49 +0200 Message-ID: <4FE6DA89.50503@lategoodbye.de> References: <4FE616CC.9000709@lategoodbye.de> <20120624022547.GA7307@kroah.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from www.augenpunkt.de ([213.239.207.9]:36015 "EHLO www.augenpunkt.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754412Ab2FXJL1 (ORCPT ); Sun, 24 Jun 2012 05:11:27 -0400 In-Reply-To: <20120624022547.GA7307@kroah.com> Sender: linux-serial-owner@vger.kernel.org List-Id: linux-serial@vger.kernel.org To: Greg KH Cc: linux-serial@vger.kernel.org Hi, Am 24.06.2012 04:25, schrieb Greg KH: > On Sat, Jun 23, 2012 at 09:19:40PM +0200, Stefan Wahren wrote: >> Hello, >> >> i'm using Linux 2.6.35 at ARM and i experience the following issue >> with the mxs-auart (App UART). If i transmit and receive data over >> the serial port (using mxs-auart), all the interrupt counter (RX, >> TX, frame, parity, break) accessed from user space (ioctl with >> TIOCGICOUNT or /proc/tty/driver/auart) are zero. >> >> Here is the scenario to reproduce the problem: >> >> 1. echoing some characters to the uart device >> 2. fetch the statistics from proc-Filesystem >> >> Expected results: transmit counter greater than zero >> >> Observed results: transmit counter is always zero >> >> The same szenario on the mxs-duart (Debug UART) doesn't show this problem. >> >> Is this a bug in mxs-auart or just not implemented? > > We disabled the transmit counter from being displayed a while ago for > security reasons, perhaps that is what you are seeing? Does this security fix affect serial_core or only mxs-auart? Prevent this security fix from reading all counter (TX, RX, frame erros, parity errors, breaks)? Do you mean this security issue ( https://bugzilla.redhat.com/show_bug.cgi?id=648660 ) ? > > Also, you do realize just how old 2.6.35 really is, right? Unfortunaley yes, Freescale the vendor delivers the Board Support Package only with 2.6.35. > > thanks, > > greg k-h > Thanks Stefan