Date: Thu, 15 Sep 2005 00:05:21 EST-10EDT,10,-1,0,7200,3,-1,0,7200,3600
Subject: [UnixOS2_Archive] No. 612

**************************************************
Wednesday 14 September 2005
 Number  612
**************************************************

Subjects for today
 
1   KbdGetStatus() always returns ERROR_KBD_INVALID_LENGTH : Tobias Huerlimann <tobias at tobiashuerlimann.de>
2  Re:  KbdGetStatus() always returns ERROR_KBD_INVALID_LENGTH : lsunley at mb.sympatico.ca
3  Re:  KbdGetStatus() always returns ERROR_KBD_INVALID_LENGTH : Tobias Huerlimann <tobias at tobiashuerlimann.de>
4  Re:  KbdGetStatus() always returns ERROR_KBD_INVALID_LENGTH : Dave Saville" <dave at deezee.org>


**= Email   1 ==========================**

Date:  Wed, 14 Sep 2005 01:36:51 +0200
From:  Tobias Huerlimann <tobias at tobiashuerlimann.de>
Subject:   KbdGetStatus() always returns ERROR_KBD_INVALID_LENGTH

Hi list:

I'm trying to create a release of GnuPG 1.4.2 for OS/2. After applying 
my old diffs to the 1.4.2 source I ran into the following problem: 
KbdGetStatus() always returns error code 376 = ERROR_KBD_INVALID_LENGTH. 
But this is the same code that has worked flawless in older releases.

To examine this problem, I've written a small test program:
==== snip ============================================================
#define INCL_ERRORS
#define INCL_KBD
#include <os2.h>
#include <stdio.h>

int main(int argc, char** argv) {
	APIRET rc;
	KBDINFO terminfo;

	rc = KbdGetStatus(&terminfo, 0);
	printf("KbdGetStatus() returned rc = %u\n", rc);

	return 0;
}
==== snip ============================================================
This always gives me rc = 376. Has anyone any idea why these few lines 
don't work? As far as I can remember, they somehow did in the past...

My GCC version (gcc --version) is "gcc (GCC) 3.3.5 (Innotek Build 
2005-07-18 03:46)", but I've also tried VAC 3.6.5 (same result).

What am I missing? Thanks for any hints!

Tobias Huerlimann


**= Email   2 ==========================**

Date:  Tue, 13 Sep 2005 21:17:19 -0500
From:  lsunley at mb.sympatico.ca
Subject:  Re:  KbdGetStatus() always returns ERROR_KBD_INVALID_LENGTH

Hi

You did not set the KBDINFO length before the call

Note the terminfo.cb = sizeof(KBDINFO) line

=========

#define INCL_ERRORS
#define INCL_KBD
#include <os2.h>
#include <stdio.h>

int main(int argc, char** argv) {
	APIRET rc;
	KBDINFO terminfo;

        terminfo.cb = sizeof(KBDINFO);
	rc = KbdGetStatus(&terminfo, 0);
	printf("KbdGetStatus() returned rc = %u\n", rc);

	return 0;
}

=========

In <43276293.6050906 at tobiashuerlimann.de>, on 09/14/05 
   at 01:36 AM, Tobias Huerlimann <tobias at tobiashuerlimann.de> said:

>#define INCL_ERRORS
>#define INCL_KBD
>#include <os2.h>
>#include <stdio.h>

>int main(int argc, char** argv) {
>	APIRET rc;
>	KBDINFO terminfo;

>	rc = KbdGetStatus(&terminfo, 0);
>	printf("KbdGetStatus() returned rc = %u\n", rc);

>	return 0;
>}

-- 
-----------------------------------------------------------
lsunley at mb.sympatico.ca
-----------------------------------------------------------



**= Email   3 ==========================**

Date:  Wed, 14 Sep 2005 13:29:20 +0200
From:  Tobias Huerlimann <tobias at tobiashuerlimann.de>
Subject:  Re:  KbdGetStatus() always returns ERROR_KBD_INVALID_LENGTH

lsunley at mb.sympatico.ca wrote:

> You did not set the KBDINFO length before the call

Thanks, that indeed seemed to be the problem. I guessed that 
KbdGetStatus() would set all values in TERMINFO itself. Strange that it 
has worked without that line in my previous ports... must have been pure 
coincidence.

Thanks,

Tobias Huerlimann


**= Email   4 ==========================**

Date:  Wed, 14 Sep 2005 14:50:23 +0100 (BST)
From:  "Dave Saville" <dave at deezee.org>
Subject:  Re:  KbdGetStatus() always returns ERROR_KBD_INVALID_LENGTH

On Wed, 14 Sep 2005 13:29:20 +0200, Tobias Huerlimann wrote:

>lsunley at mb.sympatico.ca wrote:
>
>> You did not set the KBDINFO length before the call
>
>Thanks, that indeed seemed to be the problem. I guessed that 
>KbdGetStatus() would set all values in TERMINFO itself. Strange that it 
>has worked without that line in my previous ports... must have been pure 
>coincidence.

More likely is random compiler storage mapping. Before, you happened to get the
relevant bit of storage with a reasonable value in it. Change the code a bit
and the value changes. This time it got a zero and barfed.

-- 
Regards

Dave Saville