From mboxrd@z Thu Jan 1 00:00:00 1970 From: sboyd@codeaurora.org (Stephen Boyd) Date: Mon, 11 Mar 2013 15:17:40 -0700 Subject: [PATCH 03/10] ARM: smp_twd: Divorce smp_twd from local timer API In-Reply-To: <20130311172847.GP26093@atomide.com> References: <20130308004011.GD11806@atomide.com> <51393B59.3070200@codeaurora.org> <20130308180730.GE11806@atomide.com> <513A2B20.30805@codeaurora.org> <20130308194802.GG11806@atomide.com> <513A40EF.8050403@codeaurora.org> <20130308200811.GJ11806@atomide.com> <513A4774.3040605@codeaurora.org> <20130308202947.GK11806@atomide.com> <513A8B22.5030008@codeaurora.org> <20130311172847.GP26093@atomide.com> Message-ID: <513E5804.3050104@codeaurora.org> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On 03/11/13 10:28, Tony Lindgren wrote: > * Stephen Boyd [130308 17:10]: >> On 03/08/13 12:29, Tony Lindgren wrote: >>> Applying that does not seem to help, but you might want to get vexpress >>> running anyways for some multiplatform sanity checks. >>> >>> I just built and installed qemu-linaro from their git, then ran the >>> command above. Looks like stock qemu does not work for vexpress for some >>> reason. You can probably use a dummy initrd and rootfs to debug this >>> though :) >> So my patchset didn't break qemu? > Sorry if that was unclear: Yes your patchset breaks booting vexpress > in qemu. > > Ok. qemu works for me before applying my patches. I've added this into the series before this patch to fix the boot issue. -----8<----- Subject: [PATCH] ARM: vexpress: Move smp_twd setup to time_late_init hook On devices without a clock describing the rate of the twd we detect the rate at runtime via twd_calibrate_rate(). Currently this happens when smp_prepare_cpus() calls the localtimer start callback, after the sp804 timer has been registered and when jiffies have started incrementing. In the next patch we're going to make twd_local_timer_register() register the twd clockevent on the boot CPU. In the case of vexpress, this will hang the system at boot because jiffies haven't started incrementing when twd_calibrate_rate() is called. Register the smp_twd during time_late_init() on vexpress to avoid this problem. Based on a patch by Marc Zyngier . Reported-by: Tony Lindgren Cc: Pawel Moll Signed-off-by: Stephen Boyd --- arch/arm/mach-vexpress/ct-ca9x4.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/arm/mach-vexpress/ct-ca9x4.c b/arch/arm/mach-vexpress/ct-ca9x4.c index 6f34497..a9f39bf 100644 --- a/arch/arm/mach-vexpress/ct-ca9x4.c +++ b/arch/arm/mach-vexpress/ct-ca9x4.c @@ -62,7 +62,7 @@ static void __init ct_ca9x4_init_irq(void) { gic_init(0, 29, ioremap(A9_MPCORE_GIC_DIST, SZ_4K), ioremap(A9_MPCORE_GIC_CPU, SZ_256)); - ca9x4_twd_init(); + late_time_init = ca9x4_twd_init; } static int ct_ca9x4_clcd_setup(struct clcd_fb *fb) -- Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation