[22.04 FEAT] Transparent PCI device recovery

[bugproxy]

A Linux on Z admin can make use of PCI-based devices in case of errors without the need to perform manual recovery.

Details:
Use cooperative recovery strategies that allow drivers to recover from error scenarios without complete tear-down + re-init. See approach documented for Linux on Power in Linux/Documentation/PCI/pci-error-recovery.rst.

Business value:
Improved reliability, reduced down-times.

[Frank Heimes (fheimes)]

Changing ticket to Incomplete until (upstream) kernel version and/or commits are known.

[bugproxy (bugproxy)]

------- Comment From <email address hidden> 2022-02-01 11:57 EDT-------
(In reply to comment #6)
> Changing ticket to Incomplete until (upstream) kernel version and/or commits
> are known.

This is upstream in v5.16 finnished with commit:

4cdf2f4e24ff ("s390/pci: implement minimal PCI error recovery")

There are a couple of prerequisite commits so I'll follow up with a list
needed on top of v5.15 once I've done some testing but I believe it should
be rather small.

[Frank Heimes (fheimes)]

Hi Niklas, well, for this specific patch 4cdf2f4e24ff a clean cherry-pick is possible.
But if other (rather independent ones) are needed on top, I'll wait ...

[bugproxy (bugproxy)]

------- Comment From <email address hidden> 2022-03-04 11:03 EDT-------
(In reply to comment #8)
> Hi Niklas, well, for this specific patch 4cdf2f4e24ff a clean cherry-pick is
> possible.
> But if other (rather independent ones) are needed on top, I'll wait ...

Sorry for the late reply. Even though this(In reply to comment #8)
> Hi Niklas, well, for this specific patch 4cdf2f4e24ff a clean cherry-pick is
> possible.
> But if other (rather independent ones) are needed on top, I'll wait ...

Sorry for the late reply. The commits needed for this feature all landed
in v5.16-rc1 and are:

1c8174fdc798489159a79466fca782daa231219a ("s390/pci: tolerate inconsistent handle in recover")
6526a597a2e856df9ae94512f9903caccd5196d6 ("s390/pci: add simpler s390dbf traces for events ")
4fe204977096e900cb91a3298b05c794ac24f540 ("s390/pci: refresh function handle in iomap")
da995d538d3a17610d89fea0f5813cf7921b3c2c ("s390/pci: implement reset_slot for hotplug slot")
dfd5bb23ad75bdabde89ac3166705a450bf16acb ("PCI: Export pci_dev_lock()")
4cdf2f4e24ff0d345fc36ef6d6aec059333a261e ("s390/pci: implement minimal PCI error recovery")

The first one isn't a strict dependency but a trivial bug fix to a commit in v5.15 in the same area.
I must have missed tagging that for stable. If you prefer I can also create a separate BZ for it.

[Patricia Domingues (patriciasd)]

built kernel available and tested: https://people.canonical.com/~patriciasd/kernel-lp1959532/
```
ubuntu@s1lp13:~$ uname -a
Linux s1lp13 5.15.0-23-generic #23 SMP Wed Mar 16 00:21:04 UTC 2022 s390x s390x s390x GNU/Linux
```

[Patricia Domingues (patriciasd)]

kernel with the patches: https://launchpad.net/~fheimes/+archive/ubuntu/lp1959532

[Ubuntu Kernel Bot (ubuntu-kernel-bot)]

This bug is awaiting verification that the linux-hwe-5.15/5.15.0-25.25~20.04.1 kernel in -proposed solves the problem. Please test the kernel and update this bug with the results. If the problem is solved, change the tag 'verification-needed-focal' to 'verification-done-focal'. If the problem still exists, change the tag 'verification-needed-focal' to 'verification-failed-focal'.

If verification is not done by 5 working days from today, this fix will be dropped from the source code, and this bug will be closed.

See https://wiki.ubuntu.com/Testing/EnableProposed for documentation how to enable and use -proposed. Thank you!

[Frank Heimes (fheimes)]

This feature was request for 22.04, hence the hwe kernel is more a fall out.
Hence updating tag to unblock the process.

[Launchpad Janitor (janitor)]

This bug was fixed in the package linux - 5.15.0-25.25

---------------
linux (5.15.0-25.25) jammy; urgency=medium

  * jammy/linux: 5.15.0-25.25 -proposed tracker (LP: #1967146)

  * Miscellaneous Ubuntu changes
    - SAUCE: Revert "scsi: core: Reallocate device's budget map on queue depth
      change"

 -- Paolo Pisati <email address hidden> Wed, 30 Mar 2022 17:28:11 +0200