[avrdude-dev] [Fwd: Avrdude 5.0 beta problems and patches]

[Bernard Fouché]

Hi.

I just posted the eeprom patch for atmega64/128 about 20 days ago, and 
nobody reacted at this time. I wrote to Brian today and he told me to 
send this on the list again because he is very busy at the moment.

Here is the mail I sent to him, with the patches:

----
Hello Brian.

I just had a look a the CVS depot, and I do not see the patch (or a 
correction for the corresponding bug) that makes stk500v2.c able to 
write the atmega64/128 eeprom. I've posted the patch to the avrdev-list 
but not on the savannah web site since it requires some explanations: 
the patch requires that avrdude.conf must be also modified and have, for 
eeprom fields, "read" replaced by "read_lo" and "write" replaced by 
"write_lo" for the patch to work. Now, I dunno if this is a good choice, 
or if other tokens must be taken into account by stk500v2.c. Note that 
without this problem fixed, one can not write to an atmega64/128 eeprom!

I've also a patch to have device locking management on Linux. This patch 
with be activated if 'HAS_LOCKDEV' is provided after configure(1). Now I 
never used autoconf of such tools, so I don't know where to perform 
modifications to have this constant set to 1 if the 'lockdev' package is 
available on the computer. Also the configure script must set the 
correct value for the path of the binary file "lockdev". One must use 
that program to perform file locking operations, or have the process 
needing file locking to be setuid root, which is of course a bad idea.

BTW if I download the version 5.0 beta from the CVS on a Linux FC3, what 
program should I launch to have the 'configure' file built? I tried 
'autoconf' with no luck...

At last there is a problem that, IMHO, stops the release of version 5.0: 
one can note write/verify a bootloader with the current version: if the 
hex file written to the target does not start at 0x0000, then the verify 
routine fails.

I've not noticed other problems after a few days of use and many 
reflashings on atmega64 and atmega128.

So here are the two incomplete patches...

Regards,

  Bernard

------------------------------------------------------------------------

*** stk500v2.c  2005-07-11 10:34:15.000000000 +0200
--- ../avrdude-5.0-BETA/stk500v2.c      2005-05-11 19:09:22.000000000 +0200
***************
*** 543,551 ****
       return -1;
     }
     avr_set_bits(m->op[AVR_OP_WRITE_LO], cmds);
!     //    commandbuf[6] = cmds[0];
!     commandbuf[5] = cmds[0];
!     commandbuf[6] = 0;
   }
 
   // the read command is common to both methods
--- 543,549 ----
       return -1;
     }
     avr_set_bits(m->op[AVR_OP_WRITE_LO], cmds);
!     commandbuf[6] = cmds[0];
   }
 
   // the read command is common to both methods
***************
*** 626,632 ****
 
   // the read command is common to both methods
   if (m->op[AVR_OP_READ_LO] == NULL) {
!     fprintf(stderr, "%s: stk500v2_paged_load: read instruction not defined for part 
\"%s\"\n",
             progname, p->desc);
     return -1;
   }
--- 624,630 ----
 
   // the read command is common to both methods
   if (m->op[AVR_OP_READ_LO] == NULL) {
!     fprintf(stderr, "%s: stk500v2_paged_write: read instruction not defined for part 
\"%s\"\n",
             progname, p->desc);
     return -1;
   }
***************
*** 651,657 ****
 
     result = stk500v2_command(pgm,buf,4,sizeof(buf));
     if (buf[1] != STATUS_CMD_OK) {
!       fprintf(stderr,"%s: stk500v2_paged_load: read command failed with %d\n",
               progname,buf[1]);
       return -1;
     }
--- 649,655 ----
 
     result = stk500v2_command(pgm,buf,4,sizeof(buf));
     if (buf[1] != STATUS_CMD_OK) {
!       fprintf(stderr,"%s: stk500v2_paged_write: read command failed with 
%d\n",
               progname,buf[1]);
       return -1;
     }

------------------------------------------------------------------------

*** ser_posix.c 2005-07-19 10:48:46.000000000 +0200
--- ../avrdude-5.0-BETA/ser_posix.c     2005-05-14 16:27:24.000000000 +0200
***************
*** 22,27 ****
--- 22,28 ----
 /*
  * Posix serial interface for avrdude.
  */
+ 
 #if !defined(WIN32NATIVE)
 
 
***************
*** 47,57 ****
   speed_t speed;
 };
 
- #ifdef	HAS_LOCKDEV
- #define LOCKDEV_LOCATION "/usr/sbin/lockdev"
- char *serial_port_name;
- #endif
- 
 /* There are a lot more baud rates we could handle, but what's the point? */
 
 static struct baud_mapping baud_lookup_table [] = {
--- 48,53 ----
***************
*** 130,185 ****
   return 0;
 }
 
- #ifdef	HAS_LOCKDEV
- /*
-  * Build and return a pointer to a string like:
-  * /usr/sbin/lockdev -{query} {port}
-  *
-  * Query is 'l' (lock) or 'u' (unlock).
-  *
-  */
- int make_lock_dev_query(char * query, char * port)
- {
-   char *p;
-   int l;
-   int r;
- 
-   l=strlen(LOCKDEV_LOCATION)+
-     2+        /* 1 char for space + 1 char for '-'*/
-     strlen(query)+
-     1+  /* 1 char for space */
-     strlen(port)+
-     1;        /* Ending \0 */
- 
-   p=malloc(l);
-   sprintf(p,"%s -%s %s",
-         LOCKDEV_LOCATION,
-         query,
-         port);
- 
-   r=system(p);
-   free(p);
- 
-   return r;
- }
- #endif
 
 int serial_open(char * port, int baud)
 {
   int rc;
   int fd;
 
- #ifdef	HAS_LOCKDEV
-   /*
-    * Lock the device
-    */
-   serial_port_name=strdup(port);      // To be able to unlock it by name later
-   if(make_lock_dev_query("l",serial_port_name)){
-     fprintf(stderr, "%s: serial_open(): device \"%s\" is already in use.\n",
-             progname, port);
-     exit(1);
-   }
- #endif
   /*
    * open the serial port
    */
--- 126,137 ----
***************
*** 209,224 ****
 {
   /* FIXME: Should really restore the terminal to original state here. */
 
- #ifdef	HAS_LOCKDEV
-   /*
-    * Unlock device
-    */
-   if(make_lock_dev_query("u",serial_port_name)){
-     fprintf(stderr, "%s: serial_open(): can't unlock device \"%s\" ?!.\n",
-             progname, serial_port_name);
-   }
-   free(serial_port_name);
- #endif
   close(fd);
 }
 
--- 161,166 ----


----