Port-arm archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: assertion failure on netbsd-8 RPI3



Nick Hudson <nick.hudson%gmx.co.uk@localhost> writes:

> On 09/11/2017 23:33, Greg Troxel wrote:
>> I am running a RPI3, built from netbsd-8 on 10/25ish, with evbearmv6hf.
>
> Why evbearmv6 and not evbearmv7?

Because I have a RPI1 (B 1.2) and a RPI3 and wanted to be able to move a
uSD between them and use the same image.  So I did a build of netbsd-8
with "-m evbearmv6hf-el".  I believe that the RPI and RPI2 kernels are
built with device-appropriate CPU tuning, so this only affects the
userland (and packages built by the compiler that is installed).

The text I put in the wiki says this build will work on RPI and RPI3 :-)
I get it that the userspace code will use a lower-spec instruction set
than is possible on RPI3, and that it may be a tiny bit slower.

If this is indeed a really bad idea, please explain briefly and I'll
adjust the wiki.   Specific questions:

  Are there any reasons why using a evbearmv6hf-el build on RPI3 will
  fail to work correctly?

  How much of a speed hit will there be?

>> It is mostly ok, but I found this in the logs:
>>
>> hostname /netbsd: rnel diagnostic assertion "delta > sc->sc_autoinc
>> / 100" failed: file
>> "/u0/n0/gdt/NetBSD-8/src/sys/arch/arm/cortex/gtmr.c", line 316 cpu2:
>> interrupting too quickly (delta=37) autoinc=192000
>>
>> Yes, it really does say rnel in /var/Log/messages.  In dmesg:
>>
>
> http://releng.netbsd.org/cgi-bin/req-8.cgi?show=351
>
> Seems to fix it - not sure what changed to make this a problem though.

Thanks.  I will be updating to newer netbsd-8 regularly anyway, so I'll
see if this recurs after that is merged and I update.

Attachment: signature.asc
Description: PGP signature



Home | Main Index | Thread Index | Old Index