Re: panic on boot with Raspberry Pi 3 Model B v1.2 and Model B+

[Jason Thorpe <thorpej%me.com@localhost>]

> On Mar 1, 2019, at 6:23 PM, Rob Newberry <robthedude%mac.com@localhost> wrote:
> 
>>> On Mar 1, 2019, at 4:10 PM, Rob Newberry <robthedude%mac.com@localhost> wrote:
>>> 
>>> Decided to give aarch64 a try (I’d been using the earmv7hf 32-bit version).  Tried on both my Pi 3s (a B/1.2 and a B+).  Don’t have any other Pi’s available right now.
>> 
>> You're hitting a popular and recent bug that I think may be fixed?  
> 
> Do you know when?  I think I updated my code yesterday (but from GitHub, so maybe a day's delay…)

bcm2835_intr.c:

----------------------------
revision 1.19
date: 2019-03-01 14:53:12 +0000;  author: skrll;  state: Exp;  lines: +126 -23; 
 commitid: TKkFJZyJUKjCLFdB;
Handle processor private interrupts better and add support for the PMU
interrupt.

Thanks to jmcneill for review and suggestions.
----------------------------

...I think is probably what takes care of that particular problem.  Nick can probably confirm, but in the diff the assertion that you tripped over has been removed :-)

> Assuming the fix is in the “dts” files, I took a peek at cvsweb, and they didn’t seem updated since then…but maybe the fix is elsewhere.

The fix was in code... the problem was triggered by a change in the device tree, I think.

>> In any case, you can avoid the problem by getting rid of the pmu entry in the device tree, I believe…
> 
> I’ve got to brush up on this aspect of NetBSD — it’s new from when I was using it several years ago.  I took a quick peek at the dts files and it was alishglt more effort than I was ready for right now.

Fair enough :-). In any case, if you update your tree to include the fixed interrupt handler I mention above, I think you'll be OK.

-- thorpej