[Hyper-V] VM console orphan newline input during login

Bug #1813026 reported by Chris Valean
12
This bug affects 1 person
Affects Status Importance Assigned to Milestone
linux-azure (Ubuntu)
Confirmed
Undecided
Unassigned

Bug Description

In Bionic azure-edge proposed kernel 4.18.0-1008 we're observing an issue that got carried from upstream.
This is the first version we checked specifically for this issue, we don't have yet a good/bad proposed comparison set of versions.

linux-next-5.0.0-9673b4a-20012019 does not have this issue <- daily upstream linux-next
linux-next-4.20.0-6a1d293-27122018 has the issue <- all 4.1x/4.2x series from upstream have this problem.

This behavior can be seen only on Hyper-V, as it must use the VM console UI:
- boot the above kernels.
- try to login with username and password.
- After you input the username and press Enter in the Hyper-V console, another Enter (newline) is automatically received by the console, making it impossible to input the password.

Afterward, the console behaves like a series of newlines are received and it resets after a while, presenting again the linux user login.

Repro rates:
- 100% on WS2019
- on WS2016 depending on situation it might not be visible 100%.

Problem does not exist for Azure, it's specific to Hyper-V VM console only.

Tags: hyper-v
Revision history for this message
Joshua R. Poulson (jrp) wrote :

While I do not see the commit in the 4.18.0-1008 changelog ( https://launchpad.net/ubuntu/+source/linux-azure-edge/4.18.0-1008.8~18.04.1 ) I would plant the upstream bisect near "Input: hyper-v - fix wakeup from suspend-to-idle" which was merged upstream in December:

commit adac0753c25217a2365b132c87cb2540b51fa89b
Merge: 0072a0c14d5b 6c3516fed7b6
Author: Linus Torvalds <email address hidden>
Date: Tue Dec 4 08:47:04 2018 -0800

    Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input

    Pull input updates from Dmitry Torokhov:
     "Mostly new IDs for Elan/Synaptics touchpads, plus a few small fixups"

    * 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
      Input: omap-keypad - fix keyboard debounce configuration
      Input: xpad - quirk all PDP Xbox One gamepads
      Input: synaptics - enable SMBus for HP 15-ay000
      Input: synaptics - add PNP ID for ThinkPad P50 to SMBus
      Input: elan_i2c - add ACPI ID for Lenovo IdeaPad 330-15ARR
      Input: elan_i2c - add support for ELAN0621 touchpad
      Input: hyper-v - fix wakeup from suspend-to-idle
      Input: atkbd - clean up indentation issue
      Input: st1232 - convert to SPDX identifiers
      Input: migor_ts - convert to SPDX identifiers
      Input: dt-bindings - fix a typo in file input-reset.txt
      Input: cros_ec_keyb - fix button/switch capability reports
      Input: elan_i2c - add ELAN0620 to the ACPI table
      Input: matrix_keypad - check for errors from of_get_named_gpio()

commit 10f91c73cc41ceead210a905dbd196398e99c7d2
Author: Vitaly Kuznetsov <email address hidden>
Date: Thu Nov 15 11:05:10 2018 -0800

    Input: hyper-v - fix wakeup from suspend-to-idle

    It makes little sense but still possible to put Hyper-V guests into
    suspend-to-idle state. To wake them up two wakeup sources were registered
    in the past: hyperv-keyboard and hid-hyperv. However, since
    commit eed4d47efe95 ("ACPI / sleep: Ignore spurious SCI wakeups from
    suspend-to-idle") pm_wakeup_event() from these devices is ignored. Switch
    to pm_wakeup_hard_event() API as these devices are actually the only
    possible way to wakeup Hyper-V guests.

    Fixes: eed4d47efe95 (ACPI / sleep: Ignore spurious SCI wakeups from suspend-to-idle)
    Reviewed-by: Rafael J. Wysocki <email address hidden>
    Acked-by: K. Y. Srinivasan <email address hidden>
    Acked-by: Jiri Kosina <email address hidden>
    Signed-off-by: Vitaly Kuznetsov <email address hidden>
    Signed-off-by: Dmitry Torokhov <email address hidden>

Revision history for this message
Joseph Salisbury (jsalisbury) wrote :

This bug is fixed by the following upstream commit:
d3736d82e816 (“tty: Don't hold ldisc lock in tty_reopen() if ldisc present”)

That commit is in mainline as of v5.0-rc2 and has been cc'd to upstream stable. The commit has already been applied to linux-azure master-next.

Changed in linux-azure (Ubuntu):
status: New → Confirmed
tags: added: hyper-v
removed: hyperv
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.