From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jean Delvare Subject: Re: [PATCH] i2c: Add message transfer tracepoints for I2C and SMBUS Date: Fri, 13 Dec 2013 17:53:50 +0100 Message-ID: <20131213175350.0c4a95d5@endymion.delvare> References: <20131213163349.4bc686c0@endymion.delvare> <20131213142627.4014.42860.stgit@warthog.procyon.org.uk> <3708.1386950736@warthog.procyon.org.uk> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <3708.1386950736@warthog.procyon.org.uk> Sender: linux-kernel-owner@vger.kernel.org To: David Howells Cc: linux-i2c@vger.kernel.org, rostedt@goodmis.org, linux-kernel@vger.kernel.org, Wolfram Sang List-Id: linux-i2c@vger.kernel.org On Fri, 13 Dec 2013 16:05:36 +0000, David Howells wrote: > Jean Delvare wrote: > > > One significant difference between both implementations is that the old > > one logs before the actual transfer, while yours logs afterward. While I > > understand this allows you to log the result of the transfer, this also > > means you'll miss the log if the actual transaction locks the system > > (we've seen this before.) Something to think about... > > I could split each into three messages: > > - Write request (has params & data buffer) > - Read request (has params but no data buffer) > - Read reply (has data buffer only) > > It will make the transfer functions more complex, though, and will mean that, > for i2c, you won't get all the replies to the messages in a batch in with the > requests. I can also label the messages with the index number. Mostly I > suspect this won't be a problem. Fine with me, we can leave it as is and revisit if it ever is a problem in practice. -- Jean Delvare