Re: linux-next: Tree for Aug 1

From: Guenter Roeck
Date: Thu Aug 02 2018 - 12:04:08 EST


On 08/01/2018 10:04 PM, Bart Van Assche wrote:
On Wed, 2018-08-01 at 21:58 -0700, Guenter Roeck wrote:
I am running out of ideas. Any thoughts on how to track this down further ?

Is a shell available when the hang occurs? If so, it would be helpful if you
could provide a dump of the information in /sys/kernel/debug/block. There is
namely detailed information in that directory about pending commands.


No, it hangs hard early in the boot process. See various logs at
http://kerneltests.org/builders/, in the 'next' column.

Here is some interesting information from the x86_64 boot tests.

Building x86_64:q35:Broadwell-noTSX:defconfig:smp:sata:rootfs ... running .................................. failed (timeout)
Building x86_64:q35:IvyBridge:defconfig:smp:nvme:rootfs ... running .................................. failed (timeout)
Building x86_64:q35:SandyBridge:defconfig:smp:usb:rootfs ... running .................................. failed (timeout)
Building x86_64:q35:Haswell:defconfig:smp:usb-uas:rootfs ... running ...... passed
Building x86_64:q35:Skylake-Client:defconfig:smp:mmc:rootfs ... running .................................. failed (timeout)
Building x86_64:q35:Conroe:defconfig:smp:scsi[DC395]:rootfs ... running ........ passed
Building x86_64:q35:Nehalem:defconfig:smp:scsi[AM53C974]:rootfs ... running ...... passed
Building x86_64:q35:Westmere-IBRS:defconfig:smp:scsi[53C810]:rootfs ... running ....... passed
Building x86_64:q35:Skylake-Server:defconfig:smp:scsi[53C895A]:rootfs ... running ....... passed
Building x86_64:pc:EPYC:defconfig:smp:scsi[MEGASAS]:rootfs ... running ...... passed
Building x86_64:q35:EPYC-IBPB:defconfig:smp:scsi[MEGASAS2]:rootfs ... running ....... passed
Building x86_64:q35:Opteron_G5:defconfig:smp:scsi[FUSION]:rootfs ... running ....... passed
Building x86_64:pc:phenom:defconfig:smp:initrd ... running .................................. failed (timeout)
Building x86_64:q35:Opteron_G1:defconfig:smp:initrd ... running .................................. failed (timeout)
Building x86_64:pc:Opteron_G2:defconfig:smp:sata:rootfs ... running .................................. failed (timeout)
Building x86_64:q35:core2duo:defconfig:smp:usb:rootfs ... running .................................. failed (timeout)
Building x86_64:pc:Opteron_G3:defconfig:nosmp:usb:rootfs ... running .................................. failed (timeout)
Building x86_64:q35:Opteron_G4:defconfig:nosmp:sata:rootfs ... running .................................. failed (timeout)

This is consistent across multiple test runs. In summary,

- Boot from initrd fails
- Boot from SATA drive fails (this is with CONFIG_ATA)
- Boot from NVME fails
- Boot from USB drive fails
- Boot from MMC (SD) fails
- Boot from USB UAS drive passes
- Boot from various real SCSI drives passes

Platform (pc,q35), CPU type, or SMP/NOSMP does not seem to make a difference.

Guenter