Hello,

I get a segfault with this plugin opcode, and I don't understand why... The
compilation is OK, and the opcode is recognized by  Csound.

------------------------------------SCONStruct--------------------------
"Opcodes/testread.c" added

-------------------------------------------
*.csd-----------------------------------------------

k1 testRead

--------------------------------------gdb
backtrace-----------------------------------

Program received signal SIGSEGV, Segmentation fault.
0x000000000040beae in kperf (csound=0x8daf20) at Top/csound.c:1621
1621	              (*csound->pds->opadr)(csound, csound->pds); /* run each
opcode */
(gdb) bt
#0  0x000000000040beae in kperf (csound=0x8daf20) at Top/csound.c:1621
#1  0x000000000040c2dc in csoundPerform (csound=0x8daf20) at
Top/csound.c:1731
#2  0x0000000000405c20 in main (argc=2, argv=0x7fffffffe598)
    at frontends/csound/csound_main.c:139

-----------------------------------------------testread.c----------------------------------------------
#include "csdl.h"

// --> "Opcodes/testread.c"  in SCONStruct

typedef struct {
    OPDS  h;
    MYFLT *k1;
} TESTREAD;

int testRead(CSOUND *csound, TESTREAD *p)
{
     MYFLT *k1;
     double val = 50;
     *p->k1 = val;

    return OK;
}

static OENTRY localops[] = {
	{ "testRead", sizeof(testRead), 2, "k", "",
		(SUBR)NULL, (SUBR)testRead, (SUBR)NULL }
};
LINKAGE



-----

Olivier Baudouin, PhD 
http://olivierbaudouin.com 
MINT-OMF 
Paris-Sorbonne

--
View this message in context: http://csound.1045644.n5.nabble.com/segmentation-fault-when-creating-very-simple-plugin-opcode-tp5715349.html
Sent from the Csound - Dev mailing list archive at Nabble.com.

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

On 01/09/2012 10:59, obaudouin wrote:
> Hello,
>
> I get a segfault with this plugin opcode, and I don't understand why... The
> compilation is OK, and the opcode is recognized by  Csound.
>
..

Two problems:

Firstly:
static OENTRY localops[] = {
	{ "testRead", sizeof(testRead), 2, "k", "",
		(SUBR)NULL, (SUBR)testRead, (SUBR)NULL }
};


The second field should be sizeof(TESTREAD). This is bound to elicit an 
access violation, and is very likely the one that is actually hitting, 
when Csound tries to fit all of your struct into an impossibly small 
space, and then access various bits of it.

Secondly, in that struct:

you have declared a ~pointer~ to MYFLT, "k1":

> typedef struct {
>      OPDS  h;
>      MYFLT *k1;
> } TESTREAD;
>

But it is not initialised to point to anything (no init function provided).

So accessing it below (when execution gets that far) will also cause an 
access violation:

>       *p->k1 = val;


So there is the classic question - what are you trying to do? On the 
face of it, k1 should be a plain MYFLT, not a pointer. If it is meant to 
point to some external block, that block needs to be known to exist, 
e.g. allocated via an init function.

Richard Dobson






------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

Richard's first remark is certainly true.  I think the second is not the
case  as the variable is allocated in the instrument setup
==John ff

> On 01/09/2012 10:59, obaudouin wrote:
>> Hello,
>>
>> I get a segfault with this plugin opcode, and I don't understand why...
>> The
>> compilation is OK, and the opcode is recognized by  Csound.
>>
> ..
>
> Two problems:
>
> Firstly:
> static OENTRY localops[] = {
> 	{ "testRead", sizeof(testRead), 2, "k", "",
> 		(SUBR)NULL, (SUBR)testRead, (SUBR)NULL }
> };
>
>
> The second field should be sizeof(TESTREAD). This is bound to elicit an
> access violation, and is very likely the one that is actually hitting,
> when Csound tries to fit all of your struct into an impossibly small
> space, and then access various bits of it.
>
> Secondly, in that struct:
>
> you have declared a ~pointer~ to MYFLT, "k1":
>
>> typedef struct {
>>      OPDS  h;
>>      MYFLT *k1;
>> } TESTREAD;
>>
>
> But it is not initialised to point to anything (no init function
> provided).
>
> So accessing it below (when execution gets that far) will also cause an
> access violation:
>
>>       *p->k1 = val;
>
>
> So there is the classic question - what are you trying to do? On the
> face of it, k1 should be a plain MYFLT, not a pointer. If it is meant to
> point to some external block, that block needs to be known to exist,
> e.g. allocated via an init function.
>
> Richard Dobson
>
>
>
>
>
>
> ------------------------------------------------------------------------------
> Live Security Virtual Conference
> Exclusive live event will cover all the ways today's security and
> threat landscape has changed and how IT managers can respond. Discussions
> will include endpoint security, mobile security and the latest in malware
> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
> _______________________________________________
> Csound-devel mailing list
> Csound-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/csound-devel
>
>
>



------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

Looks like I need to relearn Csound internals. Where/how would this be 
set, in the absence of a user-supplied init func?

Richard Dobson


On 01/09/2012 13:19, jpff@cs.bath.ac.uk wrote:
> Richard's first remark is certainly true.  I think the second is not the
> case  as the variable is allocated in the instrument setup
> ==John ff
>


------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

Hi Richard,

The values after OPDS are the values corresponding to the out- and
in-args to the opcode.  They are set when an instance of the
instrument is loaded. The instance function allocates the single block
of memory that was pre-calculated by otran for the instance (sum of
k-vars, a-vars, f-vars, etc. used in the instrument), and that block
is indexed into and parts assigned to the opcode's memblock instance.
In CS5, the indexes for that that block of memory are pre-calculated
and the variable names discarded.  In the CS6 typesystem branch, the
block of memory is more dynamic, in that the variables retain their
names, so as to support introspection of the instrument's variables as
well as dynamic modification of the instrument's definition.

Hope that helps!
steven


On Sat, Sep 1, 2012 at 3:46 PM, Richard Dobson
 wrote:
> Looks like I need to relearn Csound internals. Where/how would this be
> set, in the absence of a user-supplied init func?
>
> Richard Dobson
>
>
> On 01/09/2012 13:19, jpff@cs.bath.ac.uk wrote:
>> Richard's first remark is certainly true.  I think the second is not the
>> case  as the variable is allocated in the instrument setup
>> ==John ff
>>
>
>
> ------------------------------------------------------------------------------
> Live Security Virtual Conference
> Exclusive live event will cover all the ways today's security and
> threat landscape has changed and how IT managers can respond. Discussions
> will include endpoint security, mobile security and the latest in malware
> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
> _______________________________________________
> Csound-devel mailing list
> Csound-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/csound-devel

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

Thanks - momentary mental aberration. Just been such a long time since I 
looked at the sources...

Richard Dobson

On 01/09/2012 16:13, Steven Yi wrote:
> Hi Richard,
>
> The values after OPDS are the values corresponding to the out- and
> in-args to the opcode.
...


------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

I have modified  OENTRY and now it runs well!



Thanks a lot for the tip.

Olivier.




-----

Olivier Baudouin, PhD 
http://olivierbaudouin.com 
MINT-OMF 
Paris-Sorbonne

--
View this message in context: http://csound.1045644.n5.nabble.com/segmentation-fault-when-creating-very-simple-plugin-opcode-tp5715349p5715373.html
Sent from the Csound - Dev mailing list archive at Nabble.com.

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net