Fix image filters parsing

[umohnani8]

Fix the image filter parsing in the common libraries
to follow an AND logic for all filters passed in ensuring
compatibility with Docker behavior.
Also fix the filter parsing on the tunnel side so that we grab
all the filters given by the user and not only the last filter
in the list.
Add tests for the fixes.

Fixes #18412

Does this PR introduce a user-facing change?

Fix image filter parsing to follow AND logic.
Fix image filter parsing on the tunnel endpoint to pick up all the filters given by the user and not only the last one in the list.

[openshift-ci]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: umohnani8

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [umohnani8]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[umohnani8]

Need to wait on containers/common#1800 before merging

[mtrmac]

This is an improvement, but I can’t see how it could fix https://github.com/containers/podman/pull/21359/checks?check_run_id=20847280131 . In there we have a spurious match, not a spurious mismatch.

[mtrmac]

This would also allow a match against the repo name part of RepoDigests. That’s not quite clean, but it might be sufficient given the test description.

Either way, let’s see if this version works :)

[umohnani8]

@mheon @ashley-cui tests are happy now with the c/common bump.

We just need bloat approval, PTAL

[ashley-cui]

Excellent, thanks! LGTM

[mheon]

Adding bloat approved.

[TomSweeneyRedHat]

Changes LGTM
but a couple tests still seem to be causing hiccups

[TomSweeneyRedHat]

@umohnani8 ^^

[umohnani8]

I think the machine test failures are expected. All other tests are green now.

[mheon]

WSL and HyperV should be passing

[rhatdan]

/lgtm
/hold

[ashley-cui]

I can reproduce the win failure locally, taking a look....

[ashley-cui]

Looks like the error comes from this pr, on line 297. This is called by the windows client while reading containers.conf

@giuseppe, should getRootlessDirInfo() be reinstated, or is there something else we can do?

[rhatdan]

/unhold

[ashley-cui]

I don't think this should have merged, the windows failure is real. c/common is broken right now because of c/storage.

[giuseppe]

@giuseppe, should getRootlessDirInfo() be reinstated, or is there something else we can do?

can we fix it in c/storage? What is the expected result on windows?

[ashley-cui]

We need to be able to get the rootlessruntimedir in so that we can read containers.conf on the client side for windows. homedir.GetDataHome() is unsupported on windows, it was working before since getRootlessDirInfo() had a fallthrough, where it called getdatahome(), and if that errored, it used the homedir instead. I think putting that fallback logic back should fix the issue.

[ashley-cui]

We either fix it in c/storage there, or c/common here, and not use the function from c/storage. containers/common@ff1039a

[mheon]

Fixing in c/storage seems safer, we don't know what other places will try and call it

[mtrmac]

(OTOH the goal of that effort was to consolidate the heuristics/fallbacks to be the same in all parts of the binary. That’s why it was moved all the way down to c/storage. I didn’t look into details, just to make sure this is considered.)

[mtrmac]

Looks like the error comes from this pr, on line 297. This is called by the windows client while reading containers.conf

Do I undestand correctly that we are trying to compute c/storage options when setting up a podman-remote client on Windows? That seems conceptually completely unnecessary, can all of that be avoided? (And, in the extreme, compiled out?) I guess there is an abstraction boundary somewhere so that we don’t know/care that this is a remote client only… maybe worth at least filing a Podman tracking issue?