oeqa/utils/httpserver: Rework to avoid hangs and improve logging

testimage.bbclass installs a SIGTERM handler which conflicts with the
use of multiprocessing here. This is paritcularly problematic if the http
service is terminated before its started and hence before its had a chance
to reset the default signal handler (as the code was written).

Instead, temporarily remove testimage's handler whilst forking the http process
which means the correct handler is installed and won't deadlock.

Also take the opportunity to add in some log messages about the server start
and shutdown so that future debugging is easier and its clearer what the code
is doing.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

4 files changed, 32 insertions, 11 deletions

diff --git a/meta/lib/oeqa/runtime/cases/apt.py b/meta/lib/oeqa/runtime/cases/apt.py
index 8d4dd35c57..793143f72c 100644
--- a/meta/lib/oeqa/runtime/cases/apt.py
+++ b/meta/lib/oeqa/runtime/cases/apt.py
@@ -18,7 +18,7 @@ class AptRepoTest(AptTest):
     @classmethod
     def setUpClass(cls):
         service_repo = os.path.join(cls.tc.td['DEPLOY_DIR_DEB'], 'all')
-        cls.repo_server = HTTPService(service_repo, cls.tc.target.server_ip)
+        cls.repo_server = HTTPService(service_repo, cls.tc.target.server_ip, logger=cls.tc.logger)
         cls.repo_server.start()
 
     @classmethod
diff --git a/meta/lib/oeqa/runtime/cases/dnf.py b/meta/lib/oeqa/runtime/cases/dnf.py
index e93a1cea06..c1ed39d776 100644
--- a/meta/lib/oeqa/runtime/cases/dnf.py
+++ b/meta/lib/oeqa/runtime/cases/dnf.py
@@ -55,7 +55,7 @@ class DnfRepoTest(DnfTest):
     @classmethod
     def setUpClass(cls):
         cls.repo_server = HTTPService(os.path.join(cls.tc.td['WORKDIR'], 'oe-testimage-repo'),
-                                      cls.tc.target.server_ip)
+                                      cls.tc.target.server_ip, logger=cls.tc.logger)
         cls.repo_server.start()
 
     @classmethod
diff --git a/meta/lib/oeqa/runtime/cases/opkg.py b/meta/lib/oeqa/runtime/cases/opkg.py
index 50de8001c3..29e990269b 100644
--- a/meta/lib/oeqa/runtime/cases/opkg.py
+++ b/meta/lib/oeqa/runtime/cases/opkg.py
@@ -21,7 +21,7 @@ class OpkgRepoTest(OpkgTest):
         if cls.tc.td["MULTILIB_VARIANTS"]:
             allarchfeed = cls.tc.td["TUNE_PKGARCH"]
         service_repo = os.path.join(cls.tc.td['DEPLOY_DIR_IPK'], allarchfeed)
-        cls.repo_server = HTTPService(service_repo, cls.tc.target.server_ip)
+        cls.repo_server = HTTPService(service_repo, cls.tc.target.server_ip, logger=cls.tc.logger)
         cls.repo_server.start()
 
     @classmethod
diff --git a/meta/lib/oeqa/utils/httpserver.py b/meta/lib/oeqa/utils/httpserver.py
index 7d12331453..a48d4994fd 100644
--- a/meta/lib/oeqa/utils/httpserver.py
+++ b/meta/lib/oeqa/utils/httpserver.py
@@ -1,13 +1,13 @@
 import http.server
 import multiprocessing
 import os
+import traceback
+import signal
 from socketserver import ThreadingMixIn
 
 class HTTPServer(ThreadingMixIn, http.server.HTTPServer):
 
-    def server_start(self, root_dir):
-        import signal
-        signal.signal(signal.SIGTERM, signal.SIG_DFL)
+    def server_start(self, root_dir, logger):
         os.chdir(root_dir)
         self.serve_forever()
 
@@ -18,19 +18,40 @@ class HTTPRequestHandler(http.server.SimpleHTTPRequestHandler):
 
 class HTTPService(object):
 
-    def __init__(self, root_dir, host=''):
+    def __init__(self, root_dir, host='', logger=None):
         self.root_dir = root_dir
         self.host = host
         self.port = 0
+        self.logger = logger
 
     def start(self):
+        if not os.path.exists(self.root_dir):
+            self.logger.info("Not starting HTTPService for directory %s which doesn't exist" % (self.root_dir))
+            return
+
         self.server = HTTPServer((self.host, self.port), HTTPRequestHandler)
         if self.port == 0:
             self.port = self.server.server_port
-        self.process = multiprocessing.Process(target=self.server.server_start, args=[self.root_dir])
+        self.process = multiprocessing.Process(target=self.server.server_start, args=[self.root_dir, self.logger])
+
+        # The signal handler from testimage.bbclass can cause deadlocks here
+        # if the HTTPServer is terminated before it can restore the standard 
+        #signal behaviour
+        orig = signal.getsignal(signal.SIGTERM)
+        signal.signal(signal.SIGTERM, signal.SIG_DFL)
         self.process.start()
+        signal.signal(signal.SIGTERM, orig)
+
+        if self.logger:
+            self.logger.info("Started HTTPService on %s:%s" % (self.host, self.port))
+
 
     def stop(self):
-        self.server.server_close()
-        self.process.terminate()
-        self.process.join()
+        if hasattr(self, "server"):
+            self.server.server_close()
+        if hasattr(self, "process"):
+            self.process.terminate()
+            self.process.join()
+        if self.logger:
+            self.logger.info("Stopped HTTPService on %s:%s" % (self.host, self.port))
+