Attachment 948562 Details for Bug 1021327 – Patch that keeps PulseAuidio from aborting on my system

[patch] Patch that keeps PulseAuidio from aborting on my system

pulseaudio-5.0-dont-abort.patch (text/plain), 2.19 KB, created by Jay Fenlason on 2014-10-20 14:13:09 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Jay Fenlason

Created: 2014-10-20 14:13:09 UTC

Size: 2.19 KB

patch

obsolete

>--- pulseaudio-5.0/src/modules/#module-stream-restore.c~	2014-09-24 17:26:26.308870226 -0400
>+++ pulseaudio-5.0/src/modules/module-stream-restore.c	2014-09-24 21:07:32.077596814 -0400
>@@ -1283,8 +1283,10 @@ static void subscribe_callback(pa_core *
>         if ((old = entry_read(u, name))) {
>             entry = entry_copy(old);
>             created_new_entry = false;
>-        } else
>+        } else {
>             entry = entry_new();
>+            created_new_entry = true;
>+	}
> 
>         if (sink_input->save_volume && pa_sink_input_is_volume_readable(sink_input)) {
>             pa_assert(sink_input->volume_writable);
>@@ -1333,8 +1335,10 @@ static void subscribe_callback(pa_core *
>         if ((old = entry_read(u, name))) {
>             entry = entry_copy(old);
>             created_new_entry = false;
>-        } else
>+        } else {
>             entry = entry_new();
>+            created_new_entry = true;
>+	}
> 
>         if (source_output->save_volume && pa_source_output_is_volume_readable(source_output)) {
>             pa_assert(source_output->volume_writable);
>@@ -1391,19 +1395,24 @@ static void subscribe_callback(pa_core *
>         trigger_save(u);
> 
> #ifdef HAVE_DBUS
>-    if (created_new_entry) {
>-        de = dbus_entry_new(u, name);
>-        pa_assert_se(pa_hashmap_put(u->dbus_entries, de->entry_name, de) == 0);
>-        send_new_entry_signal(de);
>-    } else {
>-        pa_assert_se(de = pa_hashmap_get(u->dbus_entries, name));
>-
>+    /* 
>+     * FIXME: this code used to assert that there was no entry already in the
>+     * hashmap for name, but this caused aborts.  Someone with better
>+     * understanding of this code should figure out why created_new_entry is
>+     * set when it is already in the hashmap.
>+     */
>+    de = pa_hashmap_get(u->dbus_entries, name);
>+    if (de) {
>         if (device_updated)
>             send_device_updated_signal(de, entry);
>         if (volume_updated)
>             send_volume_updated_signal(de, entry);
>         if (mute_updated)
>             send_mute_updated_signal(de, entry);
>+    } else {
>+        de = dbus_entry_new(u, name);
>+        pa_assert_se(pa_hashmap_put(u->dbus_entries, de->entry_name, de) == 0);
>+        send_new_entry_signal(de);
>     }
> #endif
>

Actions: View | Diff

Attachments on bug 1021327: 814405 | 814407 | 814846 | 948562