From: Joe Perches Subject: Re: [PATCH 04/13] autofs4 - change printks AUTOFS defined prints Date: Wed, 05 Nov 2014 15:20:42 -0800 Message-ID: <1415229642.6634.32.camel@perches.com> References: <20141103081218.6215.85903.stgit@perseus.themaw.net> <20141103081234.6215.82821.stgit@perseus.themaw.net> <1415003136.17743.9.camel@perches.com> <1415006418.6573.6.camel@perseus.themaw.net> <1415025215.17743.12.camel@perches.com> <1415228573.2365.24.camel@perseus.themaw.net> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <1415228573.2365.24.camel@perseus.themaw.net> Sender: autofs-owner@vger.kernel.org List-ID: Content-Type: text/plain; charset="us-ascii" To: Ian Kent Cc: Andrew Morton , linux-fsdevel , autofs mailing list , Kernel Mailing List On Thu, 2014-11-06 at 07:02 +0800, Ian Kent wrote: > On Mon, 2014-11-03 at 06:33 -0800, Joe Perches wrote: > > > > That's fine. I left out the trailing semicolon/space. > > The pr_fmt could be something like: > > #define pr_fmt(fmt) KBUILD_MODNAME ":%d:%s: " fmt, current->pid, __func__ > > or add a "pid:" descriptor prefix if you like too: > > #define pr_fmt(fmt) KBUILD_MODNAME ":pid:%d:%s: " fmt, current->pid, __func__ > > > > > > it's better to use a consistent style for > > > > these logging functions ideally with terminating > > > > newlines so there isn't a mix of code with > > > > and without those newlines. That inconsistency > > > > leads to unintended defects. > > > > > > The idea here was to make the logging consistent throughout. > > > > Mine too. > > > > > I have become used of not using the new-line terminator in logging over > > > the years and tend to favour that myself. You recommend not doing that > > > probably from a kernel wide consistency perspective? Maybe that is > > > better ... > > > > Yes, kernel style consistency is the rationale. > > > > Over time, people come along and add messages > > while not reading the code very closely so using > > the kernel style with newlines can help avoid > > these trivial defects. > > I can see how not including the trailing newline in the macros is a good > thing and I'll forward a couple more patches to Andrew for this and fix > the inconsistencies. OK, great. > But idea of using pr_xxx() and pr_fmt() (actually that's too open to > name clashes so it would need to be named something like autofs_pr_fmt() > anyway) looks like it results in less readable code so I'd really prefer > not to do that. Using pr_info/pr_debug (or any other pr_) is a generic mechanism in the kernel. Adding a #define pr_fmt is also generic thing that works with all the pr_ uses in a specific compilation unit. cheers, Joe