8 Apr
2012
8 Apr
'12
6:36 p.m.
Pardon the newb question, but why does:
a) this work
---------------------------------------------
#include <jack/jack.h>
jack_position_t pos;
jack_transport_state_t state;
jack_nframes_t frame;
int process (jack_nframes_t nframes, void *arg) {
state = jack_transport_query (client, &pos);
frame = pos.frame;
}
---------------------------------------------
b) this segfaults
---------------------------------------------
#include <jack/jack.h>
jack_position_t *pos;
jack_transport_state_t state;
jack_nframes_t frame;
int process (jack_nframes_t nframes, void *arg) {
state = jack_transport_query (client, pos);
frame = pos->frame;
}
Thanks
8 Apr
8 Apr
6:49 p.m.
Pardon the newb question, but why does: ...
This
is a fairly simple C mistake.
In A, &pos refers to valid chunk of memory, thus when
jack_transport_query operates on it, it has a valid piece of memory to
work with.
In B, pos (IIRC) would be initialized to NULL as it is in the file scope.
So when you call jack_transport_query, you have provided a pointer to
NULL, where any memory access will be invalid.
If you set pos in B to be a pointer to a valid object (chunk of
memory) then it should work fine (ie refer to a global object or an
object allocated on the heap with malloc or friends).
--Mark
7:01 p.m.
Hi Mark,
In A, &pos refers to valid chunk of memory, thus
when
jack_transport_query operates on it, it has a valid piece of memory to
work with.
Got it, thanks!
In B, pos (IIRC) would be initialized to NULL as it is
in the file scope.
So when you call jack_transport_query, you have provided a pointer to
NULL, where any memory access will be invalid.
But then why does jack_transport_query not give any error when it is
passed a NULL pointer in the form of pos?
If I remove the line
frame = pos->frame;
from B, the code runs, state is set properly and jackd doesn't report problems.
Thanks for you help.
7:21 p.m.
Hi,
On Sun, 8 Apr 2012 13:01:50 -0400
Egor Sanin <egor.sanin(a)gmail.com> wrote:
But then why does jack_transport_query not give any
error when it is
passed a NULL pointer in the form of pos?
If I remove the line
frame = pos->frame;
from B, the code runs, state is set properly and jackd doesn't report problems.
I didn't look at the jack api, but calling jack_transport_query with a
null-pointer might be valid. In that case jack assumes you're not
interested in position data. iow, the jack authors didn't forget to
check for null-pointers.
Dennis
--
Dennis Schulmeister - Schifferstr. 1 - 76189 Karlsruhe - Germany
Tel: +49 721/5978883 - Fax: +49 721/5978882 - Mob: +49 152/01994400
eMail: dennis(a)windows3.de - web: there-is.no-ip.org
7:43 p.m.
On 04/08/2012 12:01 PM, Egor Sanin wrote:
Because it's valid to do this according to the docs:
http://jackaudio.org/files/docs/html/group__TransportControl.html#ga5f08eb7…
If pos is NULL it just ignores it. It's not an error at all.
OTOH, if pos is uninitialized, then jack will try to write the data to
whatever random address in in pos... and that will usually segfault.
In B, pos
(IIRC) would be initialized to NULL as it is in the file scope.
So when you call jack_transport_query, you have provided a pointer to
NULL, where any memory access will be invalid.
But then why does jack_transport_query not give any error when it is
passed a NULL pointer in the form of pos? If I remove the line
frame = pos->frame;
from B, the code runs, state is set properly and jackd doesn't report problems.
This suggests that pos is indeed NULL, and you tried to dereference a
null pointer... which will always segfault.
-gabriel
9 Apr
9 Apr
12:49 a.m.
http://jackaudio.org/files/docs/html/group__TransportControl.html#ga5f08eb7…
If pos is NULL it just ignores it. It's not an error at all.
Aha! I missed that, thanks for clarifying.
Thanks guys.
4608
days inactive
4608
days old
linux-audio-dev@lists.linuxaudio.org
5 comments
4 participants
participants (4)
-
Dennis Schulmeister -
Egor Sanin -
Gabriel M. Beddingfield -
Mark McCurry