From c33b11d09f526c156760b24bb88c9100f107a042 Mon Sep 17 00:00:00 2001
From: Fabrice Fontaine <fabrice.fontaine@orange-ftgroup.com>
Date: Tue, 28 Sep 2010 13:27:37 +0200
Subject: [PATCH] Bug fix on burst of GENA notification

When a lot of notifications were generated by a device in a short
period of time then 100% of the CPU was used to reorder those
notifications by pushing back the thread in the job queue. This
mechanism has been modified so now thread sleep 1 ms before being
pushed back into the job queue.

Removing DEFAULT_SCHED_PARAM parameter and use
sched_get_priority_min(DEFAULT_POLICY) instead.
---
 ChangeLog                   | 13 +++++++++++++
 threadutil/inc/ThreadPool.h |  4 ----
 threadutil/src/ThreadPool.c |  2 +-
 upnp/src/gena/gena_device.c | 13 ++++++++++++-
 4 files changed, 26 insertions(+), 6 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index 1c781cf..7d4d4a9 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -2,6 +2,19 @@
 Version 1.6.7
 *******************************************************************************
 
+2010-09-28 Marc Essayan <marc.essayan(at)orange-ftgroup.com>
+
+	Bug fix on burst of GENA notification
+
+	When a lot of notifications were generated by a device in a short
+	period of time then 100% of the CPU was used to reorder those
+	notifications by pushing back the thread in the job queue. This
+	mechanism has been modified so now thread sleep 1 ms before being
+	pushed back into the job queue.
+
+	Removing DEFAULT_SCHED_PARAM parameter and use
+	sched_get_priority_min(DEFAULT_POLICY) instead.
+
 2010-09-22 Fabrice Fontaine <fabrice.fontaine(at)orange-ftgroup.com>
 
 	Bug fix on M-SEARCH response
diff --git a/threadutil/inc/ThreadPool.h b/threadutil/inc/ThreadPool.h
index c33e97c..0b1044f 100644
--- a/threadutil/inc/ThreadPool.h
+++ b/threadutil/inc/ThreadPool.h
@@ -158,10 +158,6 @@ typedef int PolicyType;
 #define DEFAULT_POLICY SCHED_OTHER
 
 
-/*! Default priority */
-#define DEFAULT_SCHED_PARAM 0
-
-
 /****************************************************************************
  * Name: free_routine
  *
diff --git a/threadutil/src/ThreadPool.c b/threadutil/src/ThreadPool.c
index da8a91c..d183c75 100644
--- a/threadutil/src/ThreadPool.c
+++ b/threadutil/src/ThreadPool.c
@@ -246,7 +246,7 @@ static int SetPolicyType(PolicyType in)
 
 	memset(&current, 0, sizeof(current));
 	sched_getparam(0, &current);
-	current.sched_priority = DEFAULT_SCHED_PARAM;
+	current.sched_priority = sched_get_priority_min(DEFAULT_POLICY);
 	sched_result = sched_setscheduler(0, in, &current);
 	retVal = (sched_result != -1 || errno == EPERM) ? 0 : errno;
 #else
diff --git a/upnp/src/gena/gena_device.c b/upnp/src/gena/gena_device.c
index 4d60131..0c867ae 100644
--- a/upnp/src/gena/gena_device.c
+++ b/upnp/src/gena/gena_device.c
@@ -342,7 +342,12 @@ static void genaNotifyThread(
     struct Handle_Info *handle_info;
     ThreadPoolJob job;
 
-    HandleReadLock();
+    /* This should be a HandleLock and not a HandleReadLock otherwise if there
+     * is a lot of notifications, then multiple threads will acquire a read
+     * lock and the thread which sends the notification will be blocked forever
+     * on the HandleLock at the end of this function. */
+    //HandleReadLock();
+    HandleLock();
     //validate context
 
     if( GetHandleInfo( in->device_handle, &handle_info ) != HND_DEVICE ) {
@@ -366,6 +371,12 @@ static void genaNotifyThread(
         TPJobInit( &job, ( start_routine ) genaNotifyThread, input );
         TPJobSetFreeFunction( &job, ( free_function ) free_notify_struct );
         TPJobSetPriority( &job, MED_PRIORITY );
+
+        /* Sleep a little before creating another thread otherwise if there is
+         * a lot of notifications to send, the device will take 100% of the CPU
+         * to create threads and push them back to the job queue. */
+        imillisleep( 1 );
+
         ThreadPoolAdd( &gSendThreadPool, &job, NULL );
 
         freeSubscription( &sub_copy );