From: tony@atomide.com (Tony Lindgren) Date: Mon, 8 Jul 2013 04:25:54 -0700 Subject: Boot hang regression 3.10.0-rc4 -> 3.10.0 In-Reply-To: References: <51D577E6.5010507@newflow.co.uk> <51D59146.3070002@newflow.co.uk> <51D59C0E.8080003@newflow.co.uk> <20130705115959.GQ5523@atomide.com> Message-ID: <20130708112553.GU5523@atomide.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org * Bedia, Vaibhav [130705 06:37]: > On Fri, Jul 05, 2013 at 18:50:10, Bedia, Vaibhav wrote: > > Hi Tony, > > > > On Fri, Jul 05, 2013 at 17:29:59, Tony Lindgren wrote: > > > * Bedia, Vaibhav [130705 01:17]: > > > > > > > > I just checked the behavior on my AM335x-EVM. Current mainline boots fine > > > > provided I don't use earlyprintk. The offending patch [1] in this case is the one > > > > that tries to get rid of omap_serial_early_init() for DT boot. This change inadvertently > > > > also results in the console UART getting reset and idled during bootup and that's where > > > > the boot stops for you. I think if you skip earlyprintk from the bootargs you should see > > > > the system booting fine. > > > > > > > > I guess we need to retain the NO_IDLE and NO_RESET aspect for the console UART in > > > > omap_serial_early_init() to get earlyprintk working again. > > > > > > Hmm nothing should get idled while earlyprintk is running, and then when the > > > serial driver kicks in it should not idle anything by default. And for DT based > > > booting we should not have mach-omap2/serial.c initialize anything. > > > > > > > If I add in the HWMOD flags without any reverts I get to the point where the serial driver > > comes up but the boot eventually stops [1]. Without the flags the boot stops much earlier [2] > > just like Mark reported. > > > > Err.. the log with HWMOD flags added is [2] and without flags is [1]. Sorry for the confusion. It sounds like something needs to be fixed for am33xx as omap3 and omap4 won't hang with earlyprintk. Almost certainly mach-omap2/serial.c should not be needed at all for am33xx, and the bug is somewhere else. > > > I wonder if this is because the timeouts get now initialized to 0 instead > > > of -1 for the serial driver? > > > > > > > You meant initialized to -1, right? There's an additional check for timeout being 0. Unless i > > am missing something DT-boot will start off with timeout set to 0 and then get forced to -1. OK Regards, Tony