Proxmark3 community

doegox

Contributor

Registered: 2009-03-19

Posts: 11

proxmark3 client exit conditions are strange...

Hi,

I wanted to fix a simple issue, that quitting by CTRL-D caused a segfault when working offline, but the situation is more complex than I thought.

Current code is:
upon "quit" the client exits immediately without any cleaning: exit(0)
upon "CTRL-D" the main_loop() returns and main() continue with:
* uart_close()
* pthread_mutex_destroy(&print_lock);

Both methods (quit & CTRL-D) should lead to the same exit procedure IMHO.
But I don't know how to make "quit" returning from main_loop, replacing exit by return makes the client to be stalled.

Then once both end up into the final code of main(), one should protect uart_close() by this otherwise you get a segfault when working offline:
if (!offline) uart_close(sp);

Still I get the following error with CTRL-D:
GLib-CRITICAL **: g_main_context_pop_thread_default: assertion 'stack != NULL' failed

So to conclude I've two issues I don't know how to solve:
* get "quit" acting as "CTRL-D", so returning to main() instead of brutal exit(0)
* get rid of the g_main_context_pop_thread_default error

Phil

wil

Contributor

Registered: 2010-04-13

Posts: 14

Re: proxmark3 client exit conditions are strange...

Hi Phil,

Here is an ugly quick fix. It was working quite well with the non-CDC version of the client.

Wil

diff --git a/client/proxmark3.c b/client/proxmark3.c
index 4b898d5..1b03fe9 100644
--- a/client/proxmark3.c
+++ b/client/proxmark3.c
@@ -137,9 +137,7 @@ static void *main_loop(void *targ)
         script_file = NULL;
     }

-    ExitGraphics();
-    pthread_exit(NULL);
-    return NULL;
+    exit(0);
 }

 int main(int argc, char **argv)

doegox

Contributor

Registered: 2009-03-19

Posts: 11

Re: proxmark3 client exit conditions are strange...

That's again a kind of brutal ending :-)
If main() is willing to call uart_close(sp); before quitting I guess we should make sure this point is reached and avoid abrupt exits from anywhere...
Same thing for ExitGraphics(), pthread_exit() etc, we should better understand them, fix them if needed, rather than killing blindly the process.
IMHO
Phil

wil

Contributor

Registered: 2010-04-13

Posts: 14

Re: proxmark3 client exit conditions are strange...

Sure

Here is the message you get with a QT lib compiled with debug flags (on Gentoo):
Never mind the use of /dev/null ...

% ./proxmark3 /dev/null
ERROR: invalid serial port
proxmark3> data load /tmp/proxif-17 
loaded 16000 samples          
proxmark3> data plot
proxmark3> 
ASSERT failure in QCoreApplication::sendEvent: "Cannot send events to objects owned by a different thread. Current thread 7f991c0130a0. Receiver '' (of type 'QWidget') was created in thread 1f2efa0", file kernel/qcoreapplication.cpp, line 535
zsh: abort      ./proxmark3 /dev/null
%

Wil

Last edited by wil (2014-04-01 09:03:44)

wil

Contributor

Registered: 2010-04-13

Posts: 14

Re: proxmark3 client exit conditions are strange...

I observe this ASSERT message while quitting with CTRL-D, but not with the "quit" command.

Wil