View Issue Details

IDProjectCategoryView StatusLast Update
0002499ardourbugspublic2010-04-24 10:32
Reporterfullgo Assigned To 
PrioritynormalSeveritycrashReproducibilityalways
Status confirmedResolutionopen 
Product Version2.7.1 
Summary0002499: Bounce with insert crashes program
DescriptionBouncing a track with an active insert reproducibly crashes Ardour. Bounce works fine when insert is inactive. This happens when insert is an outboard program or a buss within Ardour.
TagsNo tags attached.

Activities

cth103

2008-12-21 13:46

administrator   ~0005541

Hi,

Thanks for the report. I can't reproduce here. Can you upload a session which crashes on export?

Thanks

Carl

fullgo

2008-12-21 17:10

reporter   ~0005542

Carl,

I have uploaded a simple session that reproduces this crash. Here are details. This session has one track with audio. There is one buss as well. Track 1 includes a prefader insert connected to the buss. Note that I have not included any plugins in this example. If you right click the audio track in the editor window, you can select option 'bounce' which correctly creates a bounce region (in this case a duplicate of the track. Output from ardour is...

>Result of bounce of martin right-2 len = 30656512 was martin right-2-0-bounce-1 len = 30656512

Now do the same thing, except make the insert active. This should crash with the following output:

>ardour-2.7.1: ../common/JackGraphManager.cpp:44: void Jack::AssertBufferSize(jack_nframes_t): >Assertion `buffer_size <= 8192' failed.

Note that I am using the most up to date jack (jackdmp 1.9.1) on a fedora 8 system. Kernel is from CCRMA (2.6.24.7-1.rt21.1.fc8.ccrmart x86_64).

Thanks. Hope this helps.

-Joe D

2008-12-21 17:38

 

test.tar.gz (3,072,301 bytes)

cth103

2008-12-22 20:55

administrator   ~0005543

Hi Joe,

Thanks for that, confirmed here. I think this may caused by a bug in JACK. The attached patch for ardour works around it, and fixes the problem for me. But it's not a "proper" fix. I'm trying to work out whether ardour or JACK are in the wrong :)

Many thanks for your help with this.

Carl

2008-12-22 20:56

 

export-crash-workaround.patch (437 bytes)   
diff -r b79a5ee173c4 libs/ardour/session.cc
--- a/libs/ardour/session.cc	Mon Dec 22 20:06:21 2008 +0000
+++ b/libs/ardour/session.cc	Mon Dec 22 20:54:34 2008 +0000
@@ -4018,8 +4018,7 @@
 		return result;
 	}
 
-	// any bigger than this seems to cause stack overflows in called functions
-	const nframes_t chunk_size = (128 * 1024)/4;
+	const nframes_t chunk_size = get_block_size ();
 
 	g_atomic_int_set (&processing_prohibited, 1);
 	

Issue History

Date Modified Username Field Change
2008-12-21 05:02 fullgo New Issue
2008-12-21 13:46 cth103 Note Added: 0005541
2008-12-21 13:46 cth103 Status new => feedback
2008-12-21 17:10 fullgo Note Added: 0005542
2008-12-21 17:38 fullgo File Added: test.tar.gz
2008-12-22 20:55 cth103 Note Added: 0005543
2008-12-22 20:55 cth103 Status feedback => confirmed
2008-12-22 20:56 cth103 File Added: export-crash-workaround.patch
2010-04-24 10:28 cth103 Category bugs => bugs2
2010-04-24 10:32 cth103 Category bugs2 => bugs