[process,PackageManager] Create a mapping of processes to packages

Adds a new manual collection to the `process` plugin, that tries to
compile a mapping of running processes' binaries to an owning package
via the package manager. As such, package managers now have a new
`pkg_by_path()` method that serves this purpose.

Closes: #1350

Signed-off-by: Jake Hunsaker <jhunsake@redhat.com>

6 files changed, 52 insertions, 5 deletions

diff --git a/sos/policies/package_managers/__init__.py b/sos/policies/package_managers/__init__.py
index 8a5c4ee5..f16a57d2 100644
--- a/sos/policies/package_managers/__init__.py
+++ b/sos/policies/package_managers/__init__.py
@@ -50,12 +50,13 @@ class PackageManager():
     verify_command = None
     verify_filter = None
     files_command = None
+    query_path_command = None
     chroot = None
     files = None
 
-    def __init__(self, chroot=None, query_command=None,
-                 verify_command=None, verify_filter=None,
-                 files_command=None, remote_exec=None):
+    def __init__(self, chroot=None, query_command=None, verify_command=None,
+                 verify_filter=None, files_command=None,
+                 query_path_command=None, remote_exec=None):
         self._packages = {}
         self.files = []
 
@@ -63,6 +64,7 @@ class PackageManager():
         self.verify_command = verify_command or self.verify_command
         self.verify_filter = verify_filter or self.verify_filter
         self.files_command = files_command or self.files_command
+        self.query_path_command = query_path_command or self.query_path_command
 
         # if needed, append the remote command to these so that this returns
         # the remote package details, not local
@@ -192,6 +194,24 @@ class PackageManager():
             self.files = files.splitlines()
         return self.files
 
+    def pkg_by_path(self, path):
+        """Given a path, return the package that owns that path.
+
+        :param path:    The filepath to check for package ownership
+        :type path:     ``str``
+
+        :returns:       The package name or 'unknown'
+        :rtype:         ``str``
+        """
+        if not self.query_path_command:
+            return 'unknown'
+        try:
+            cmd = f"{self.query_path_command} {path}"
+            pkg = shell_out(cmd, timeout=5, chroot=self.chroot)
+            return pkg.splitlines() or 'unknown'
+        except Exception:
+            return 'unknown'
+
     def build_verify_command(self, packages):
         """build_verify_command(self, packages) -> str
             Generate a command to verify the list of packages given
diff --git a/sos/policies/package_managers/dpkg.py b/sos/policies/package_managers/dpkg.py
index 7619e929..684c6f60 100644
--- a/sos/policies/package_managers/dpkg.py
+++ b/sos/policies/package_managers/dpkg.py
@@ -16,6 +16,7 @@ class DpkgPackageManager(PackageManager):
     """
 
     query_command = "dpkg-query -W -f='${Package}|${Version}\\n'"
+    query_path_command = "dpkg -S"
     verify_command = "dpkg --verify"
     verify_filter = ""
 
diff --git a/sos/policies/package_managers/rpm.py b/sos/policies/package_managers/rpm.py
index 5150e1e0..0d16a1e3 100644
--- a/sos/policies/package_managers/rpm.py
+++ b/sos/policies/package_managers/rpm.py
@@ -16,6 +16,7 @@ class RpmPackageManager(PackageManager):
     """
 
     query_command = 'rpm -qa --queryformat "%{NAME}|%{VERSION}|%{RELEASE}\\n"'
+    query_path_command = 'rpm -qf'
     files_command = 'rpm -qal'
     verify_command = 'rpm -V'
     verify_filter = ["debuginfo", "-devel"]
diff --git a/sos/report/plugins/__init__.py b/sos/report/plugins/__init__.py
index 42032d50..0fb5dc20 100644
--- a/sos/report/plugins/__init__.py
+++ b/sos/report/plugins/__init__.py
@@ -3164,7 +3164,7 @@ class Plugin():
             self._log_info(f"manual collection '{fname}' finished in {run}")
             if isinstance(tags, str):
                 tags = [tags]
-            self.manifest.collections[fname.split('.')[0]] = {
+            self.manifest.collections[fname] = {
                 'filepath': _pfname,
                 'tags': tags
             }
diff --git a/sos/report/plugins/process.py b/sos/report/plugins/process.py
index 62705c4f..50279502 100644
--- a/sos/report/plugins/process.py
+++ b/sos/report/plugins/process.py
@@ -6,6 +6,7 @@
 #
 # See the LICENSE file in the source distribution for further information.
 
+import json
 import re
 
 from sos.report.plugins import Plugin, IndependentPlugin, PluginOpt
@@ -88,4 +89,28 @@ class Process(Plugin, IndependentPlugin):
             "pidstat -p ALL -rudvwsRU --human -h",
             "pidstat -tl"
         ])
+
+    def collect(self):
+        with self.collection_file('pids_to_packages.json') as pfile:
+            if not self.policy.package_manager.query_path_command:
+                pfile.write('Package manager not configured for path queries')
+                return
+            _ps = self.exec_cmd('ps --no-headers aex')
+            pidpkg = {}
+            paths = {}
+            if not _ps['status'] == 0:
+                pfile.write(f"Unable to get process list: {_ps['output']}")
+                return
+            for proc in _ps['output'].splitlines():
+                proc = proc.strip().split()
+                pid = proc[0]
+                path = proc[4]
+                if not self.path_exists(path):
+                    continue
+                if path not in paths:
+                    paths[path] = self.policy.package_manager.pkg_by_path(path)
+                pidpkg[pid] = {'path': path, 'package': paths[path]}
+
+            pfile.write(json.dumps(pidpkg, indent=4))
+
 # vim: set et ts=4 sw=4 :
diff --git a/tests/report_tests/plugin_tests/collet_manual_tests.py b/tests/report_tests/plugin_tests/collect_manual_tests.py
index fdcda526..7000d5e0 100644
--- a/tests/report_tests/plugin_tests/collet_manual_tests.py
+++ b/tests/report_tests/plugin_tests/collect_manual_tests.py
@@ -34,4 +34,4 @@ class CollectManualTest(StageOneReportTest):
         pkgman = self.get_plugin_manifest('unpackaged')
         self.assertTrue(pkgman['collections']['unpackaged'])
         pyman = self.get_plugin_manifest('python')
-        self.assertTrue(pyman['collections']['digests'])
+        self.assertTrue(pyman['collections']['digests.json'])