pmwg/integ bisection: baseline.login on rk3328-rock64

[Arnd Bergmann]

On Tue, Jan 12, 2021 at 2:46 PM Vincent Guittot
<vincent.guittot at linaro.org> wrote:
> On Tue, 12 Jan 2021 at 12:25, Guillaume Tucker <guillaume.tucker at collabora.com> wrote:
> >
> > Some details can be found here:
> >
> >   https://kernelci.org/test/case/id/5ffb978de38e717501c94cd8/
> >
> > The bisection was run with CONFIG_RANDOMIZE_BASE=y enabled, but
> > the same issue occurs with a plain defconfig from that branch.
> > Results with other configs and platforms can be compared here:
> >
> >   https://kernelci.org/test/job/pmwg/branch/integ/kernel/v5.11-rc3-13-gcea05edf93998/plan/baseline/
> >
> > Please let us know if you need some help to test a fix or debug
> > the issue.
>
> I don't have any rockchip board on which I can run tests but I can
> push an updated version of the patch without trace events if this can
> then be then tested on kernelci rockchip board.

It's odd that there is no output from the kernel at all, which
rules out most of the options that change, as they should only
have an effect later during boot.

One thing I noticed is that the last successful boot has a kernel
image size just under 32MB for the vmlinux, while the failing
configurations have more than 34MB.

I looked at the loadaddress as well:

11:43:57.119234  Filename '536723/tftp-deploy-j462agcw/kernel/Image'.
11:43:57.122622  Load address: 0x2000000
11:43:59.708404  Bytes transferred = 34361856 (20c5200 hex)
...
11:43:59.814346  Filename
'536723/tftp-deploy-j462agcw/ramdisk/ramdisk.cpio.gz.uboot'.
11:43:59.814638  Load address: 0x4000000
11:44:00.447924  Bytes transferred = 8517374 (81f6fe hex)

So the ramdisk and the kernel overlap when the kernel is larger than 32MB!

     Arnd