On Thu, Jan 07, 2016 at 05:28:10PM +0100, Thierry Reding wrote:

On Thu, Jan 07, 2016 at 03:59:58PM +0000, Mark Brown wrote:

quoted

On Thu, Jan 07, 2016 at 07:41:52AM -0800, kernelci.org bot wrote:

quoted

Tree: next
Branch: local/master
Git Describe: next-20160107
Git Commit: 0f023a298a7d2a8e99c54913f7277b3c08c1f5c9
Git URL: git://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git
Tested: 95 unique boards, 25 SoC families, 28 builds out of 135

Boot Failures Detected: https://kernelci.org/boot/?next-20160107&fail

For the past couple of days -next has been failing to boot on various
Tegra platforms, there are failures involving either no kernel output or
(with tegra_defconfig) backtraces during CMA allocation from USB, for
example:

   http://storage.kernelci.org/next/next-20160107/arm-tegra_defconfig/lab-khilman/boot-tegra30-beaver.html

quoted

    multi_v7_defconfig+CONFIG_THUMB2_KERNEL=y:
        tegra124-jetson-tk1: 2 failed labs
        tegra30-beaver: 1 failed lab

    tegra_defconfig:
        tegra124-jetson-tk1: 2 failed labs
        tegra124-jetson-tk1_rootfs:nfs: 1 failed lab
        tegra124-nyan-big: 1 failed lab
        tegra30-beaver: 1 failed lab

    multi_v7_defconfig+CONFIG_ARM_LPAE=y:
        tegra124-jetson-tk1: 1 failed lab
        tegra124-nyan-big: 1 failed lab

    multi_v7_defconfig:
        tegra124-jetson-tk1: 2 failed labs
        tegra30-beaver: 1 failed lab

    multi_v7_defconfig+CONFIG_LKDTM=y:
        tegra124-jetson-tk1: 2 failed labs
        tegra30-beaver: 1 failed lab

    multi_v7_defconfig+CONFIG_PROVE_LOCKING=y:
        tegra124-jetson-tk1: 2 failed labs
        tegra30-beaver: 1 failed lab

I did bisects yesterday but they were a bit inconclusive, showing a
series of skipped commits in the TTY tree as being the culprits:

afd7f88f157796e586fc99d62da13a54024e0731 serial: 8250: move of_serial code to 8250 directory
4e33870b3bb691996354a8f9e8f69458b4fc34d9 serial: of: CONFIG_SERIAL_8250 is always set
d1b5c87fa8058a3f477ae05555916dd1cea934ad serial: remove NWP serial support

which seems more than a little surprising and like they might be false
positives, especially in the cases that do produce output.  The errors
today look the same.  I've not investigated further than the bisects.

Full bisect logs:

 https://ci.linaro.org/view/people/job/tbaker-boot-bisect-bot/140/console
 https://ci.linaro.org/view/people/job/tbaker-boot-bisect-bot/141/console

These should all be fixed by this patch:

	serial: 8250: of: Fix the driver and actually compile the 8250_of

which can be found in patchwork here:

	https://patchwork.kernel.org/patch/7883541/

I think this is merely waiting to be picked up into the tty tree. I
suspect the reason why there's output in the first place is because
of earlyprintk. I've seen recent linux-next work properly, with the
serial console hanging at some point (fbcon shows login prompt). If
I apply the above patch things are back to normal.

I'll go queue that up later, but how is this suddenly showing up in
-next?  The tty tree has been pretty quiet for a while now..

thanks,

greg k-h