[LEDE-DEV] Recent issues w/ sysupgrade

[Philip Prindeville]

> On Sep 5, 2017, at 8:50 PM, Ryan Mounce <ryan at mounce.com.au> wrote:
> 
> On 31 August 2017 at 04:11, Philip Prindeville
> <philipp_subx at redfish-solutions.com> wrote:
>> Looking a little further into the console logging, what I’m seeing after a “dd” directly onto the flash is this:
>> 
>> Press the [f] key and hit [enter] to enter failsafe mode
>> Press the [1], [2], [3] or [4] key and hit [enter] to select the debug level
>> [    6.299598] mount_root: loading kmods from internal overlay
>> [    6.307977] kmodloader: loading kernel modules from //etc/modules-boot.d/*
>> [    6.316014] kmodloader: done loading kernel modules from //etc/modules-boot.d/*
>> [    6.459868] block: attempting to load /etc/config/fstab
>> [    6.472689] block: extroot: not configured
>> [    6.488330] mount_root: rootdisk overlay filesystem has not been formatted yet
>> [    6.517993] blk_update_request: I/O error, dev loop0, sector 0
>> [    6.724898] block: attempting to load /etc/config/fstab
>> [    6.730248] block: extroot: not configured
>> [    6.734560] mount_root: overlay filesystem has not been fully initialized yet
>> [    6.741823] mount_root: switching to f2fs overlay
>> 
>> 
>> if I take the exact same image and try to reinstall it with sysupgrade, I see this:
>> 
>> 
>> Press the [f] key and hit [enter] to enter failsafe mode
>> Press the [1], [2], [3] or [4] key and hit [enter] to select the debug level
>> [    5.319594] mount_root: loading kmods from internal overlay
>> [    5.328062] kmodloader: loading kernel modules from //etc/modules-boot.d/*
>> [    5.336078] kmodloader: done loading kernel modules from //etc/modules-boot.d/*
>> [    5.480712] block: attempting to load /etc/config/fstab
>> [    5.493487] block: extroot: not configured
>> [    5.498237] random: procd: uninitialized urandom read (4 bytes read)
>> [    5.515173] mount_root: rootdisk overlay filesystem has not been formatted yet
>> 
>> 
>> it’s tough to reproduce, which makes me wonder if it’s either a race condition or an uninitialized variable.
> 
> I can consistently reproduce this issue when testing my Turris Omnia
> patch set, hopefully I can help to narrow it down.
> 
> A couple of times interrupting boot into failsafe and then running
> mount_root has fixed things, however more often than not mount_root
> hangs in the same way.
> 
> This appears to have been introduced in the last couple of months so I
> will try to bisect it.


Thanks.

It’s not 100% reproducible which makes it time-consuming to bisect because you have to repeat steps a few times.

But I did conclude decisively that it did NOT present before:

5eb216e mediatek: update to latest kernel patchset from v4.13-rc

So you don’t have to search any earlier than that.

It’s in the last 3 weeks.

-Philip