From mboxrd@z Thu Jan 1 00:00:00 1970 From: Scott Wood Subject: Re: [PATCH v6 22/22] of/platform: Defer probes of registered devices Date: Thu, 29 Oct 2015 11:06:18 -0500 Message-ID: <1446134778.701.374.camel@freescale.com> References: <1442844182-27787-1-git-send-email-tomeu.vizoso@collabora.com> <1442844182-27787-23-git-send-email-tomeu.vizoso@collabora.com> <1445406845.701.55.camel@freescale.com> <1445467912.701.90.camel@freescale.com> <1445549230.701.116.camel@freescale.com> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: Sender: devicetree-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: Tomeu Vizoso Cc: Rob Herring , "linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org" , Stephen Warren , Javier Martinez Canillas , Greg Kroah-Hartman , Mark Brown , Thierry Reding , Alan Stern , "Rafael J. Wysocki" , "linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r@public.gmane.org" , Dmitry Torokhov , "devicetree-u79uwXL29TY76Z2rM5mHXA@public.gmane.org" , Linus Walleij , "linux-acpi-u79uwXL29TY76Z2rM5mHXA@public.gmane.org" , Arnd Bergmann , linuxppc-dev , Hu Mingkai-B21284 List-Id: linux-acpi@vger.kernel.org On Wed, 2015-10-28 at 15:40 +0100, Tomeu Vizoso wrote: > On 22 October 2015 at 23:27, Scott Wood wrote: > > On Thu, 2015-10-22 at 15:04 +0200, Tomeu Vizoso wrote: > > > On 22 October 2015 at 00:51, Scott Wood wrote: > > > > On Wed, 2015-10-21 at 08:44 -0500, Rob Herring wrote: > > > > > On Wed, Oct 21, 2015 at 12:54 AM, Scott Wood < > > > > > scottwood-KZfg59tc24xl57MIdRCFDg@public.gmane.org> > > > > > wrote: > > > > > > On Mon, 2015-09-21 at 16:03 +0200, Tomeu Vizoso wrote: > > > > > > > Instead of trying to match and probe platform and AMBA devices > > > > > > > right > > > > > > > after each is registered, delay their probes until > > > > > > > device_initcall_sync. > > > > > > > > > > > > > > This means that devices will start probing once all built-in > > > > > > > drivers > > > > > > > have registered, and after all platform and AMBA devices from > > > > > > > the DT > > > > > > > have been registered already. > > > > > > > > > > > > > > This allows us to prevent deferred probes by probing > > > > > > > dependencies on > > > > > > > demand. > > > > > > > > > > > > > > Signed-off-by: Tomeu Vizoso > > > > > > > --- > > > > > > > > > > > > > > Changes in v4: > > > > > > > - Also defer probes of AMBA devices registered from the DT as > > > > > > > they > > > > > > > can > > > > > > > also request resources. > > > > > > > > > > > > > > drivers/of/platform.c | 11 ++++++++--- > > > > > > > 1 file changed, 8 insertions(+), 3 deletions(-) > > > > > > > > > > > > This breaks arch/powerpc/sysdev/fsl_pci.c. The PCI bus is an OF > > > > > > platform > > > > > > device, and it must be probed before pcibios_init() which is a > > > > > > subsys_initcall(), or else the PCI bus never gets scanned. > > > > > > > > > > Thanks for the report. This is probably getting dropped, but it > > > > > could > > > > > be disabled for PPC. > > > > > > > > I don't think that adding another arbitrary arch difference would be > > > > the > > > > right solution. > > > > > > I think Rob meant temporarily disable it while things get fixed. At > > > least, > > > > So, what is the permanent fix for the swiotlb issue (or more generally, > > the > > inability to have a late_initcall that runs after non-module, non-hotplug > > platform devices have been probed)? > > If the code in pcibios_init() depends on the PCI bus device having > probed, then I would recommend making that dependency explicit by > calling of_device_probe() on the OF node of the PCI controller when > looking it up. "when looking it up"? pcibios_init() doesn't do anything with the OF node or know any details about the particular PCI bus host implementation. > > > I don't see any reason why PPC wouldn't benefit from this > > > series. > > > > It's not clear to me what the benefit of this is at all, much less for > > PPC. > > What is the fundamental problem with deferred probes? In the cover letter > > you say this change saves 2.3 seconds, but where is that time being > > consumed? > > Are the drivers taking too long in their probe function trying to > > initialize > > and then deferring, rather than checking for dependencies up front? Or > > are > > there really so many devices and such a pessimal ordering that most of the > > time is spent iterating through and reordering the list, with each defer > > happening quickly? > > The problem is that a device that defers its probe is currently sent > to the back of the queue, and that's undesired in some use cases in > which there's a device that should be up as soon as possible during > boot (and boot takes a long time). So the goal is to change the order > in which devices with dependencies end up probing. That doesn't answer my question about where the time is being spent. Did you profile? -Scott -- To unsubscribe from this list: send the line "unsubscribe devicetree" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html