On Fri, Aug 10, 2018 at 11:44:17AM +0200, Ondřej Jirman wrote:
> Hi,
> 
> On Thu, Aug 09, 2018 at 09:42:21AM -0700, Nuno Gonçalves wrote:
> > Dear Ondřej,
> > 
> > Thank you. I applied this patches to linux and u-boot and the number of 
> > crashes did reduce, but they still happen.
> > 
> > Any other idea? :)
> 
> You may try a stress test and toggle the cpufreq frequency settings manually
> in a fast loop, to see if you can reproduce it faster. Because, waiting 8h
> or more for a crash is not optimal. :)

You may want to try it under various loads, if it doesn't crash when idle.

> If yes, you may try increasing delay between regulator voltage change and cpu
> frequency change. It's a regulator-ramp-delay property in dts, if that changes
> anything.
> 
> regards,
>   o.
> 
> > Thanks,
> > Nuno
> > 
> > On Monday, July 30, 2018 at 7:41:13 PM UTC+2, Ondřej Jirman wrote:
> > >
> > > Hi, 
> > >
> > > there's issue in mainline kernel, where some NKMP CPU clock rate changes 
> > > lockup the ARM cores (or cause some corruption), because of use of 
> > > inappropriate 
> > > clock factor changes. 
> > >
> > > You may try applying these three patches, to see if it helps and makes 
> > > your 
> > > system stable (it's for H3, so you should check that H2 is using the same 
> > > clock 
> > > driver as H3): 
> > >
> > >
> > > https://github.com/megous/linux/commit/8477a566b36aaae77e53a9949f963ce6ebad55fe
> > >  
> > >
> > > https://github.com/megous/linux/commit/f07a20e4ce93e7e0f333f9f6e57f14fd4ab66abd
> > >  
> > >
> > > https://github.com/megous/linux/commit/c313a0ac340bf1131d475527171764dc49901895
> > >  
> > >
> > > And this one to u-boot (otherwise you might get instability on the first 
> > > CPU 
> > > frequency change after boot): 
> > >
> > >
> > > https://github.com/megous/linux/blob/orange-pi-4.18/0001-sunxi-h3-Fix-PLL1-setup-to-never-use-dividers.patch
> > >  
> > >
> > > Let me know if that fixes it. I got the exact same issues on H3. 
> > >
> > > regards, 
> > >   o. 
> > >
> > > On Mon, Jul 30, 2018 at 03:58:56AM -0700, Nuno Gonçalves wrote: 
> > > > Hi, 
> > > > 
> > > > I've been facing a issue since I enabled CPUFreq on 4.18, and I have 
> > > > confirmed it up to today's 4.18-rc7. 
> > > > 
> > > > I'm using the Orange Pi Zero (H2+), currently with 12 devices under 
> > > test, 
> > > > and they crash randomly in average every 8 hours. 
> > > > 
> > > > No temperature abnormality, and they actually seem to not crash under 
> > > > stress(1) full CPU load, but they do under some "medium" userland load. 
> > > > 
> > > > This board have a voltage regulator with 1.1 or 1.3V output, so I also 
> > > > investigated if it was related. It seems not. Even when enabling only 
> > > > 240MHz to 816MHz, always at 1.1V, it crash anyway. 
> > > > 
> > > > This is the output I get on console: 
> > > > 
> > > > [ 5339.047465] INFO: rcu_sched detected stalls on CPUs/tasks: 
> > > > [ 5339.053764] 1-...!: (0 ticks this GP) idle=028/0/0 
> > > softirq=29760/29760 
> > > > fqs=1 
> > > > [ 5339.061215] (detected by 2, t=8407 jiffies, g=45041, c=45040, 
> > > > q=2011) 
> > > > [ 5339.075799] rcu_sched kthread starved for 6302 jiffies! g45041 
> > > > c45040 
> > > > f0x0 RCU_GP_WAIT_FQS(3) ->state=0x402 ->cpu=1 
> > > > [ 5339.086470] RCU grace-period kthread stack dump: 
> > > > [ 5358.496609] INFO: rcu_sched detected stalls on CPUs/tasks: 
> > > > [ 5358.502901] 1-...!: (0 ticks this GP) idle=5e4/0/0 
> > > softirq=29760/29760 
> > > > fqs=0 
> > > > [ 5358.510591] 3-...!: (3 GPs behind) idle=e5c/0/0 softirq=49624/49624 
> > > > fqs=0 
> > > > [ 5358.517771] (detected by 2, t=2102 jiffies, g=45042, c=45041, 
> > > > q=1958) 
> > > > [ 5358.539568] rcu_sched kthread starved for 2106 jiffies! g45042 
> > > > c45041 
> > > > f0x0 RCU_GP_WAIT_FQS(3) ->state=0x402 ->cpu=1 
> > > > [ 5358.550237] RCU grace-period kthread stack dump: 
> > > > [ 5416.697273] INFO: rcu_sched detected stalls on CPUs/tasks: 
> > > > [ 5416.702781] 1-...!: (0 ticks this GP) idle=2f4/0/0 
> > > softirq=29760/29760 
> > > > fqs=1 
> > > > [ 5416.709991] (detected by 2, t=8407 jiffies, g=45042, c=45041, 
> > > > q=2912) 
> > > > [ 5416.717537] rcu_sched kthread starved for 6301 jiffies! g45042 
> > > > c45041 
> > > > f0x0 RCU_GP_WAIT_FQS(3) ->state=0x402 ->cpu=1 
> > > > [ 5416.728043] RCU grace-period kthread stack dump: 
> > > > [ 5436.138107] INFO: rcu_sched detected stalls on CPUs/tasks: 
> > > > [ 5436.144416] 3-...!: (4 GPs behind) idle=e70/0/0 softirq=49624/49624 
> > > > fqs=0 
> > > > [ 5436.151609] (detected by 2, t=2102 jiffies, g=45043, c=45042, 
> > > > q=2920) 
> > > > [ 5436.166152] rcu_sched kthread starved for 2105 jiffies! g45043 
> > > > c45042 
> > > > f0x0 RCU_GP_WAIT_FQS(3) ->state=0x402 ->cpu=1 
> > > > [ 5436.176827] RCU grace-period kthread stack dump: 
> > > > 
> > > > I can't get a getty console anymore when this happens. 
> > > > 
> > > > Looking for suggestions how to pinpoint the issue. 
> > > > 
> > > > Thanks, 
> > > > Nuno 
> > > > 
> > > > -- 
> > > > You received this message because you are subscribed to the Google 
> > > Groups "linux-sunxi" group. 
> > > > To unsubscribe from this group and stop receiving emails from it, send 
> > > an email to linux-sunxi...@googlegroups.com <javascript:>. 
> > > > For more options, visit https://groups.google.com/d/optout. 
> > >
> > >
> > 
> > -- 
> > You received this message because you are subscribed to the Google Groups 
> > "linux-sunxi" group.
> > To unsubscribe from this group and stop receiving emails from it, send an 
> > email to linux-sunxi+unsubscr...@googlegroups.com.
> > For more options, visit https://groups.google.com/d/optout.
> 

-- 
You received this message because you are subscribed to the Google Groups 
"linux-sunxi" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to linux-sunxi+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to