From mboxrd@z Thu Jan 1 00:00:00 1970 From: Rob Herring Subject: Re: [PATCH v2] drivers: base: initcall_debug logs for driver probe times Date: Mon, 25 Jun 2018 09:20:55 -0600 Message-ID: <20180625152055.GA23270@rob-hp-laptop> References: <20180621003556.65578-1-toddpoynor@gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <20180621003556.65578-1-toddpoynor@gmail.com> Sender: linux-kernel-owner@vger.kernel.org To: Todd Poynor Cc: Frank Rowand , devicetree@vger.kernel.org, Greg Kroah-Hartman , linux-kernel@vger.kernel.org, kernel-team@android.com, Todd Poynor List-Id: devicetree@vger.kernel.org On Wed, Jun 20, 2018 at 05:35:56PM -0700, Todd Poynor wrote: > From: Todd Poynor > > Add initcall_debug logs for each driver device probe call, for example: > > probe of a3800000.ramoops returned 1 after 3007 usecs > > This replaces the previous code added to report times for deferred > probes. It also reports OF platform bus device creates that were > formerly lumped together in a single entry for function > of_platform_default_populate_init, as well as helping to annotate other > initcalls that involve device probing. > > Remove restriction on printing probe times only during initcalls, since > initcall_debug now continues to show driver timing info past the boot > phase. > > Signed-off-by: Todd Poynor > --- > drivers/base/dd.c | 50 +++++++++++++++++++++-------------------------- > 1 file changed, 22 insertions(+), 28 deletions(-) > > Changes from v1: > * unsigned long long -> s64 > * use ktime_to_us() instead of ktime_to_ns with shift > * print probe return value > * drop code that stopped output on end of initcall processing Reviewed-by: Rob Herring