sof-kernel-log-check: fix last_line / begin_line off-by-one error

[marc-hb]

If the last line of kern.log was an error then the next test was failing.

Found thanks to the new fake_kern_error() function submitted separately.

Signed-off-by: Marc Herbert marc.herbert@intel.com

[marc-hb]

All the shellcheck warnings in https://travis-ci.org/github/thesofproject/sof-test/builds/725412899 were already there, this adds none.

[marc-hb]

Interestingly, this makes a lot of tests fail in https://sof-ci.01.org/softestpr/PR366/build161/devicetest/. All of them seem to be because of:
Initramfs unpacking failed: Decoding failed or EXT4-fs (mmcblk0p2): re-mounted. Opts: errors=remount-ro or both. Will investigate.

[xiulipan]

@marc-hb I check the CI result. The logs are from very beginning this boot. It should be filter out by our last_line cmd. I think your PR here may have something wrong.

 Sep 8 23:32:35 sh-bdw-wsb-rt286-02 kernel: [ 0.425729] acpi PNP0A08:00: _OSC failed (AE_ERROR); disabling ASPM
Sep 8 23:32:35 sh-bdw-wsb-rt286-02 kernel: [ 0.704764] Initramfs unpacking failed: Decoding failed
Sep 8 23:32:35 sh-bdw-wsb-rt286-02 kernel: [ 3.368895] EXT4-fs (sda2): re-mounted. Opts: errors=remount-ro

[xiulipan]

Please be aware the if input is 0, we will have fallback.

# confirm begin_line is number, if it is not the number, direct using dmesg
[[ "${begin_line//[0-9]/}" ]] && begin_line=0
[[ "$begin_line" -eq 0 ]] && cmd="dmesg" || cmd="sed -n '$begin_line,\$p' /var/log/kern.log"

[marc-hb]

Will hopefully be made useless by #470. Holding my breath.

Now #509

[marc-hb]

Made obsolete by #509