Add tracers tests (New)

[ahresse]

This is an attempt to add ptrace and ftrace tests in checkbox.

Description

For a specific project at Canonical we are using checkbox and we need to check for the availability of user-space ptrace and ftrace tools in our images. Thus, we want to add a dedicated test to check this and since it could be relevant to other, this might be into base checkbox test plan.

This, PR:

• Introduces a new dedicated unit for tracers.
• Introduces a specific binary tool: ptrace_test (source code in C) that aims to test the ptrace() system call availability.
• Introduces the associated test with ptrace_test.
• Introduces a new test to check ftrace availability (based on the availability of tracefs).

A potential alternative to the usage of specific ptrace_test binary: parsing /proc/kallsyms got identified but not developed.

Resolved issues

Associated Jira ticket

Documentation

Tests

[codecov]

Codecov Report

Merging #735 (0ef5bdc) into main (c3c3f45) will not change coverage.
Report is 1 commits behind head on main.
The diff coverage is n/a.

@@           Coverage Diff           @@
##             main     #735   +/-   ##
=======================================
  Coverage   34.83%   34.83%           
=======================================
  Files         302      302           
  Lines       34165    34165           
  Branches     5909     5909           
=======================================
  Hits        11903    11903           
  Misses      21697    21697           
  Partials      565      565           

Flag	Coverage Δ
provider-base	3.13% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

[ahresse]

Can't assign reviewer but @zongminl and @LaiderLai should be in it.

[yphus]

What are the reasons why ptrace/ftrace would not be available on the base image? is it via kernel config?

I'd suggest turning the two proposed jobs into a resource job that would output something like that:

ptrace: supported
ftrace: unsupported

More tracers could be added in the future to this resource.
and have a new job that checks their availability (using the fail-on-resource flag):

id: tracers/check
category_id: tracer
_summary: ptrace and ftrace availability
_description:
  Check availability of ptrace syscall and ftrace mount (tracefs)
flags: simple fail-on-resource
command: echo "ptrace and ftrace found in the image"
requires:
  tracers.ptrace == 'supported'
  tracers.ftrace == 'supported'

[kissiel]

What are the reasons why ptrace/ftrace would not be available on the base image? is it via kernel config?

Yep, https://github.com/torvalds/linux/blob/master/kernel/trace/Kconfig#L157

[kissiel]

Looks great. One comment about descriptions below.

[kissiel]

Unless I'm missing something, this will not work as intended, as the resource job kernel_config_file has to be included in the test plan in the "bootstrapping_include" section, otherwise the ftrace will never run.

I've also placed some comments below.

[kissiel]

when any of those tests fail the error code is not propagated, so in the end this may happen:

=========[ Running job 2 / 3. Estimated time left (at least): 0:00:00 ]=========
--------------------[ Tests the ptrace tracer availability ]--------------------
ID: com.canonical.certification::tracers/ptrace                                 
Category: com.canonical.plainbox::uncategorised                                 
... 8< -------------------------------------------------------------------------
ptrace: unsupported                                                                            
------------------------------------------------------------------------- >8 ---
Outcome: job passed    
 

[ahresse]

/canonical/self-hosted-runners/run-workflows ca01039

[kissiel]

/canonical/self-hosted-runners/run-workflows ca01039

[kissiel]

/canonical/self-hosted-runners/run-workflows 0ef5bdc

[kissiel]

Fantastic work, TYVM