[Discuss-gnuradio] Freeing memory problem in GNURadio

[Gorur, Anisha - 0662 - MITLL]

Hello,
I've been having a strange problem with the gr-fec block, and I think it may be 
connected to other issues I'm having. I'm running a flowgraph on the E310 using 
the LDPC Encoder Definition (via Generator), LDPC Generator Matrix, and FEC 
Async Encoder blocks and it has been running fine for a few months. For another 
block, I needed to update the Volk library because I wanted a new kernel, but 
when I did so a couple of things started to break. 

I thought this was due to my update of GNURadio/Volk, but I have reinstalled 
the SDK as well as UHD and GNUradio from source (using these instructions: 
https://files.ettus.com/manual/page_usrp_e3x0.html) and I still get the same 
problems. I am using UHD version 3.9.2 (I need to for other compatibility 
issues, this has not given me trouble in the past) and GNURadio 3.7.11.1 (The 
version I was using when everything worked). My cross compiling also seems to 
be working fine, I get no errors in my cmake or make process. I am using the 
fpga files and SDK from here: 
http://files.ettus.com/e3xx_images/e3xx-release-4/ 

In the following gdb backtraces, I've built both GNURadio and my OOT module 
with -DCMAKE_BUILD_TYPE=Debug.  

Error 1:
When I kill a flowgraph with the LDPC Encoder in it, I get either a  segfault 
or an invalid pointer error when the program tries to call the  destructor of 
ldpc_gen_mtrx_encoder_impl.cc. (full backtrace in 
message_port_in_not_working.txt)

```
-- Loading FPGA image: 
/home/root/emb/usr/share/uhd/images/usrp_e3xx_fpga_idle_sg3.bit...[Thread 
0xa95ff460 (LWP 1241) exited]
 done
[Thread 0xb1546460 (LWP 1227) exited]
*** Error in `/usr/bin/python': munmap_chunk(): invalid pointer: 0xb6fa7d10 ***

Program received signal SIGABRT, Aborted.
0xb6d38894 in __GI_raise (address@hidden) at 
../sysdeps/unix/sysv/linux/raise.c:55
55        return INLINE_SYSCALL (tgkill, 3, pid, selftid, sig);
(gdb) bt
#0  0xb6d38894 in __GI_raise (address@hidden) at 
../sysdeps/unix/sysv/linux/raise.c:55
#1  0xb6d3c394 in __GI_abort () at abort.c:89
#2  0xb6d707b8 in __libc_message (do_abort=<optimized out>, fmt=0xb6e28a00 "*** 
Error in `%s': %s: 0x%s ***\n") at ../sysdeps/posix/libc_fatal.c:175
#3  0xb6d7aa74 in malloc_printerr (action=3, str=0xb6e28a24 "munmap_chunk(): 
invalid pointer", ptr=<optimized out>) at malloc.c:4960
#4  0xb6d7ab94 in munmap_chunk (p=<optimized out>) at malloc.c:2817
#5  0xb3df81d8 in gsl_block_free (b=0xb6f9fda8 <PyInt_Type>) at 
/usr/src/debug/gsl/1.15-r0/gsl-1.15/block/init_source.c:78
#6  0xb3e47b40 in gsl_matrix_free (m=0x1d754) at 
/usr/src/debug/gsl/1.15-r0/gsl-1.15/matrix/init_source.c:189
#7  0xb3fbedc4 in gr::fec::code::ldpc_G_matrix_impl::~ldpc_G_matrix_impl 
(address@hidden, __in_chrg=<optimized out>, __vtt_parm=<optimized out>) at 
/home/agorur/projects/emb/src/gnuradio/gr-fec/lib/ldpc_G_matrix_impl.cc:294
#8  0xb3fc0664 in ~ldpc_G_matrix_impl (this=0x7af6c0, __in_chrg=<optimized 
out>, __vtt_parm=<optimized out>) at 
/home/agorur/projects/emb/src/gnuradio/gr-fec/lib/ldpc_G_matrix_impl.cc:295
#9  checked_delete<gr::fec::code::ldpc_G_matrix_impl> (x=0x7af6c0) at 
/home/agorur/projects/emb/sysroots/armv7ahf-vfp-neon-oe-linux-gnueabi/usr/include/boost/core/checked_delete.hpp:34
```

Error 2: I have my own message block, which has suddenly stopped working, the 
flowgraph does not recognize the input port (called pdus_in) and the backtrace 
traces back to a boost header file. (full backtrace in 
ldpc_destructor_free_error.txt)
```
-- Loading FPGA image: 
/home/root/emb/usr/share/uhd/images/usrp_e310_fpga_sg3.bit... done
-- Input frequency reference: 40 MHz
-- Detecting internal GPSDO [New Thread 0xb1546460 (LWP 1249)]
.... found
-- Initializing core control...
-- Performing register loopback test... pass
-- Performing register loopback test... pass
-- Performing register loopback test... pass
-- Performing CODEC loopback test... pass
-- Performing CODEC loopback test... pass
-- Setting time source to internal
-- Asking for clock rate 10.24 MHz
-- Actually got clock rate 10.24 MHz
-- Performing timer loopback test... pass
-- Performing timer loopback test... pass
-- Setting time source to gpsdo
Could not find port: pdus_in in:
system
pdus_in
Program received signal SIGBUS, Bus error.
0xb64d24d8 in fetch_add (order=boost::memory_order_relaxed, v=1, 
address@hidden: <error reading variable>) at 
/home/agorur/projects/emb/sysroots/armv7ahf-vfp-neon-oe-linux-gnueabi/usr/include/boost/atomic/detail/ops_gcc_atomic.hpp:100
100     
/home/agorur/projects/emb/sysroots/armv7ahf-vfp-neon-oe-linux-gnueabi/usr/include/boost/atomic/detail/ops_gcc_atomic.hpp:
 No such file or directory.
(gdb) bt
#0  0xb64d24d8 in fetch_add (order=boost::memory_order_relaxed, v=1, 
address@hidden: <error reading variable>) at 
/home/agorur/projects/emb/sysroots/armv7ahf-vfp-neon-oe-linux-gnueabi/usr/include/boost/atomic/detail/ops_gcc_atomic.hpp:100
#1  fetch_add (order=boost::memory_order_relaxed, v=1, this=0x6) at 
/home/agorur/projects/emb/sysroots/armv7ahf-vfp-neon-oe-linux-gnueabi/usr/include/boost/atomic/detail/atomic_template.hpp:115
#2  pmt::intrusive_ptr_add_ref (p=0x2) at 
/home/agorur/projects/emb/src/gnuradio/gnuradio-runtime/lib/pmt/pmt.cc:69
#3  0xb655a5f8 in intrusive_ptr (rhs=..., this=0xbeffec54) at 
/home/agorur/projects/emb/sysroots/armv7ahf-vfp-neon-oe-linux-gnueabi/usr/include/boost/smart_ptr/intrusive_ptr.hpp:92
```
I've rebuilt the SDK multiple times, I'm really not sure what could be giving 
me these errors.
Thank you,

Anisha Gorur

ldpc_destructor_free_error.txt
Description: ldpc_destructor_free_error.txt

pdus_in_port_not_working.txt
Description: pdus_in_port_not_working.txt