arm64 Ubuntu kernel builds missing CONFIG_NLS_ASCII

[Adam Bell]

The CONFIG_NLS_ASCII was already set to Y in the raspi-5.4 kernel
and bionic kernel by Juerg, now do the same change in the
raspi2-5.3 kernel.

[Impact]
users reported that the raspi ubuntu core has some problem to mount
filessytem, need to enable the NLS_ASCII and make it built-in.

[Fix]
change it from m to y just as the focal and bionic raspi kernel.

[Test Case]
building the kernel and boot the kernel, it could mount the
filesystem successfully.

[Regression Risk]
Low, just changing a kconfig to y from m.

Remounting log mounts in Ubuntu core are failing on arm64 builds due to missing CONFIG_NLS_ASCII in the arm64 kernel build options.
This option is enabled on the other kernel builds but somehow was overlooked on arm64 appears to affect our ability to access logs via journalctl with data from prior to restart because certain mounts are failing.
Other Raspberry Pi kernel builds, such as the armhf build, have the option enabled and so do not encounter the mount error.
The kernel option that needs to be turned on to resolve this is CONFIG_NLS_ASCII. It needs to be set to either 'm' or 'y'. Most of the other chips have it set to 'y', keeping it built-in.

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Oliver Grawert (ogra)]

this is definitely a regression, mounting partitions should work ... marking as critical.

[Loïc Minier (lool)]

ubuntu-focal.git, focal/raspi2 branch:
debian.raspi2/config/config.common.ubuntu:CONFIG_NLS_ASCII=y

ubuntu-bionic.git, bionic/raspi2 branch:
debian.raspi2/config/config.common.ubuntu:CONFIG_NLS_ASCII=y

So not quite sure where this is coming from

[Brent DeGraaf (brentk4)]

Perhaps looking backward from this Raspberry pi kernel will help locate the
issue, which was definitely present in the build below:

5.3.0-1026-raspi2 #28~18.04.1-Ubuntu SMP Thu May 21 14:48:58 UTC 2020
aarch64 aarch64 aarch64 GNU/Linux

The final kernel that was getting built on arm64 did not have the flag set
on that tag. The same flag is notably absent in the arm64 defconfig, which
appears to be what was used judging from the buildlog output I was able to
see searching on that tag. If this was very recently fixed, we haven't
received a snap refresh as of yet.

--

Brent DeGraaf

Principal Engineer

734.395.0852

<email address hidden>

[image: image.png]

www.K4Connect.com <http://www.k4connect.com/> | @K4Connect
<https://twitter.com/K4Connect> | Facebook
<https://www.facebook.com/k4connect>| LinkedIn
<https://www.linkedin.com/company/5187480>

5511 Capital Center Dr. Suite 350

Raleigh, NC 27606

On Tue, Jun 2, 2020 at 11:45 AM Loïc Minier <email address hidden>
wrote:

> ubuntu-focal.git, focal/raspi2 branch:
> debian.raspi2/config/config.common.ubuntu:CONFIG_NLS_ASCII=y
>
> ubuntu-bionic.git, bionic/raspi2 branch:
> debian.raspi2/config/config.common.ubuntu:CONFIG_NLS_ASCII=y
>
> So not quite sure where this is coming from
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1881631
>
> Title:
> arm64 Ubuntu kernel builds missing CONFIG_NLS_ASCII
>
> Status in linux-raspi2 package in Ubuntu:
> Confirmed
>
> Bug description:
> Remounting log mounts in Ubuntu core are failing on arm64 builds due to
> missing CONFIG_NLS_ASCII in the arm64 kernel build options.
> This option is enabled on the other kernel builds but somehow was
> overlooked on arm64 appears to affect our ability to access logs via
> journalctl with data from prior to restart because certain mounts are
> failing.
> Other Raspberry Pi kernel builds, such as the armhf build, have the
> option enabled and so do not encounter the mount error.
> The kernel option that needs to be turned on to resolve this is
> CONFIG_NLS_ASCII. It needs to be set to either 'm' or 'y'. Most of the
> other chips have it set to 'y', keeping it built-in.
>
> To manage notifications about this bug go to:
>
> https://bugs.launchpad.net/ubuntu/+source/linux-raspi2/+bug/1881631/+subscriptions
>

[Francis Ginther (fginther)]

@arbell,

Do you have the specific kernel and version you observed this on? A `uname -a` or `dmesg` log would be sufficient. Based on some of the comments here, the config value may have changed between versions so I want to make sure we have the failing one identified correctly.

[Brent DeGraaf (brentk4)]

@fginther

Just to make sure that it was not my comment that misled you, I am not aware of any currently fixed version.

[Adam Bell (arbell)]

@fginther

I opened this bug report on behalf of @brentk4 from a support ticket. Please refer to the kernel that referenced in #4.

[Hui Wang (hui.wang)]

ubuntu raspi-5.3 kernel (both armhf and arm64) set this config to m while ubutnu raspi-5.4 kernel set it to y. Maybe the initrd of ubuntu core doesn't contain nls_ascii.ko, so we need to set it to y in the raspi-5.3 too?

[Brent DeGraaf (brentk4)]

So is it now changed to 'y'? Last night I confirmed that the module is missing in initrd.img and that is what was causing the mount error, so this is great news.

[Launchpad Janitor (janitor)]

This bug was fixed in the package linux-raspi2 - 5.3.0-1028.30

---------------
linux-raspi2 (5.3.0-1028.30) eoan; urgency=medium

  * eoan/linux-raspi2: 5.3.0-1028.30 -proposed tracker (LP: #1882700)

  * Eoan update: upstream stable patchset 2020-06-01 (LP: #1881657)
    - [Config] updateconfigs for CC_HAS_WARN_MAYBE_UNINITIALIZED

  * arm64 Ubuntu kernel builds missing CONFIG_NLS_ASCII (LP: #1881631)
    - [Config] raspi2: Set NLS_ASCII=y

  [ Ubuntu: 5.3.0-60.54 ]

  * CVE-2020-0543
    - UBUNTU/SAUCE: x86/speculation/srbds: do not try to turn mitigation off when
      not supported
  * Packaging resync (LP: #1786013)
    - [Packaging] update helper scripts
  * Eoan update: upstream stable patchset 2020-06-05 (LP: #1882303)
    - i2c: dev: Fix the race between the release of i2c_dev and cdev
    - KVM: SVM: Fix potential memory leak in svm_cpu_init()
    - ima: Set file->f_mode instead of file->f_flags in ima_calc_file_hash()
    - evm: Check also if *tfm is an error pointer in init_desc()
    - ima: Fix return value of ima_write_policy()
    - mtd: spinand: Propagate ECC information to the MTD structure
    - fix multiplication overflow in copy_fdtable()
    - ubifs: remove broken lazytime support
    - iommu/amd: Fix over-read of ACPI UID from IVRS table
    - i2c: mux: demux-pinctrl: Fix an error handling path in
      'i2c_demux_pinctrl_probe()'
    - ubi: Fix seq_file usage in detailed_erase_block_info debugfs file
    - gcc-common.h: Update for GCC 10
    - HID: multitouch: add eGalaxTouch P80H84 support
    - HID: alps: Add AUI1657 device ID
    - HID: alps: ALPS_1657 is too specific; use U1_UNICORN_LEGACY instead
    - scsi: qla2xxx: Fix hang when issuing nvme disconnect-all in NPIV
    - scsi: qla2xxx: Delete all sessions before unregister local nvme port
    - configfs: fix config_item refcnt leak in configfs_rmdir()
    - vhost/vsock: fix packet delivery order to monitoring devices
    - aquantia: Fix the media type of AQC100 ethernet controller in the driver
    - component: Silence bind error on -EPROBE_DEFER
    - scsi: ibmvscsi: Fix WARN_ON during event pool release
    - HID: i2c-hid: reset Synaptics SYNA2393 on resume
    - x86/apic: Move TSC deadline timer debug printk
    - gtp: set NLM_F_MULTI flag in gtp_genl_dump_pdp()
    - HID: quirks: Add HID_QUIRK_NO_INIT_REPORTS quirk for Dell K12A keyboard-dock
    - ceph: fix double unlock in handle_cap_export()
    - stmmac: fix pointer check after utilization in stmmac_interrupt
    - USB: core: Fix misleading driver bug report
    - platform/x86: asus-nb-wmi: Do not load on Asus T100TA and T200TA
    - ARM: futex: Address build warning
    - padata: Replace delayed timer with immediate workqueue in padata_reorder
    - padata: initialize pd->cpu with effective cpumask
    - padata: purge get_cpu and reorder_via_wq from padata_do_serial
    - ALSA: iec1712: Initialize STDSP24 properly when using the model=staudio
      option
    - ALSA: pcm: fix incorrect hw_base increase
    - ALSA: hda/realtek - Fix silent output on Gigabyte X570 Aorus Xtreme
    - ALSA: hda/realtek - Add more fixup entries for Clevo machines
    - drm/etnaviv: fix perfmon domain interation
    - apparmor...