On 08/03/2018 08:40 AM, Nicolas Chauvet wrote:
Le jeu. 2 août 2018 à 21:24, David R. Piegdon <lkml@xxxxxxxx> a écrit :
Considering that nobody has answered if he is affected by this too, or
not, I have to assume that this only affects my platform. Maybe its a
pinmuxing issue. We will have to investigate further. :-(.
David,
I've experienced this bug on a tegra20 device today (paz00). Is this
the issue you have experienced ?
I confirm that I don't reproduce this on every boot (even using the
same kernel).
Hello Nicolas,
it looks very much like it, yes. I assume that you are running the
serial console with the 8250 driver.
The console would stutter afterwards (polling mode). Also one core runs
at 100% sirq.
But once getty re-opens the device (e.g. initial login or when you log
out and getty gets restarted), the device uses IRQs again and works
properly.
For me, the bug was triggered roughly 3-5% of all reboots before the
first getty pops up.
Can you confirm this?
Then you might want to try the patch and let me know if it helped.
The patch should fix the issue most of the time. But in about 0.05% or
so of all reboots the console will stop receiving characters until it is
reopened. I was not able to fix this so far. But at least there is no
more 100% sirq hogging of one core.
Please let me know about any progress!
David
--
To unsubscribe from this list: send the line "unsubscribe linux-tegra" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
