volume/local: Make host resolution backwards compatible

[vvoland]

• fixes: container with cifs volume and "addr"-option does not start with version 25 #47184
• relates to Running a container with a Samba volume specified by a hostname seems to have stop working. docker/cli#4816

Commit 8ae94ca added a DNS resolution of the device part of the volume option.

The previous way to resolve the passed hostname was to use addr option, which was handled by the same code path as the nfs mount type.

The issue is that addr is also an SMB module option handled by kernel and passing a hostname as addr produces an invalid argument error.

To fix that, restore the old behavior to handle addr the same way as before, and only perform the new DNS resolution of device if there is no addr passed.

NOTE: While kernel docs don't mention the addr option (https://www.kernel.org/doc/html/latest/admin-guide/cifs/usage.html#use-instructions), it's actually present in the source: https://github.com/torvalds/linux/blob/7ed2632ec7d72e926b9e8bcc9ad1bb0cd37274bf/fs/smb/client/fs_context.c#L166

- What I did

- How I did it

- How to verify it

- Description for the changelog

- Fix CIFS volume mount error when passing an `addr` mount option

- A picture of a cute animal (not mandatory but encouraged)

[thaJeztah]

LGTM

Good call from @robmry; let's look at that

[robmry]

LGTM - although duplicating the code seems like a bit of a shame... could factor it out into a closure that returns a bool to say whether it did anything?

[vvoland]

Removed the duplication, not using a closure though it would make it less readable and more complex than plain duplication IMO.

[thaJeztah]

LGTM

[robmry]

LGTM

[thaJeztah]

Seeing failures like this quite frequently in the BuildKit integration tests. Not specifically this address, but seems to be general lookup / "not found" failures (?)

=== FAIL: client TestIntegration (2.05s)
time="2024-01-23T14:01:29Z" level=info msg="trying next host - response was http.StatusNotFound" host="localhost:42291"
time="2024-01-23T14:01:29Z" level=info msg=request digest="sha256:be691b1535726014cdf3b715ff39361b19e121ca34498a9ceea61ad776b9c215" mediatype=application/vnd.docker.image.rootfs.foreign.diff.tar size=10240 url="https://gist.github.com/cpuguy83/fcf3041e5d8fb1bb5c340915aabeebe0/raw/eebc59ee14939fc38067aa8b8dfbfba053d0af94/base.tar"
    run.go:266: copied docker.io/cpuguy83/buildkit-foreign:latest to local mirror localhost:42291/cpuguy83/buildkit-foreign:latest
time="2024-01-23T14:01:29Z" level=info msg="trying next host - response was http.StatusNotFound" host="localhost:42291"
    run.go:266: copied docker.io/amd64/alpine:latest@sha256:25fad2a32ad1f6f510e528448ae1ec69a28ef81916a004d3629874104f8a7f70 to local mirror localhost:42291/library/alpine:latest
time="2024-01-23T14:01:30Z" level=info msg="trying next host - response was http.StatusNotFound" host="localhost:42291"
    run.go:266: copied docker.io/tonistiigi/test:nolayers to local mirror localhost:42291/tonistiigi/test:nolayers

Either way; it's not related to this PR, so kicked CI again.

Hoping one day we have stats / metrics from CI, so that we can identify flaky tests.