Progress of sorts. Made the modification to volk_malloc.c and rebuilt
gnuradio. Ctest still returns some errors (LastTest log attached), but
using a (de)modulator no longer crashes with the volk memory error…so
that’s a plus. Oddly enough the FFT display still doesn’t show axis
----- Original Message -----
From: “Tom R.” [email protected]
To: “Robert Fisher” [email protected]
Cc: “GNURadio D.ion List” [email protected]
Sent: Wednesday, March 19, 2014 11:29:32 PM
Subject: Re: [Discuss-gnuradio] (de)modulation problems gnuradio
3.7.3…also GRC FFT missing axis labels
On Wed, Mar 19, 2014 at 10:25 PM, [email protected] wrote:
Since the LastTests.log file was so large (150K) I extracted a few error
blocks from the file. I’m attempting to compile this on a PowerBook G4
(PowerPC architecture) running Ubuntu 12.04.
Wow. Haven’t had anyone report using a PowerPC with GNU Radio for a
while. But I’m not surprised that there are VOLK related issues; a
fairly untested platform for us these days.
This is failing using volk_malloc, which looks to see if your system
is POSIX compliant and if so, use posix_malloc; otherwise, use an
internally developed way of allocating aligned memory. In your case,
posix_memalign is being called and throwing that warning, which is
If you can apply this diff (a one-line change) and rebuild, it will
disable using posix_memalign. It’ll be interesting to see if the
backup method will work for you.
diff --git a/volk/lib/volk_malloc.c b/volk/lib/volk_malloc.c
index b2b77dd…e913ffc 100644
@@ -54,7 +54,8 @@
// Otherwise, test if we are a POSIX or X/Open system
// This only has a restriction that alignment be a power of 2.
-#if _POSIX_C_SOURCE >= 200112L || _XOPEN_SOURCE >= 600
+//#if _POSIX_C_SOURCE >= 200112L || _XOPEN_SOURCE >= 600
void *volk_malloc(size_t size, size_t alignment)