[PATCH] solution: Fails to close session for DVD+R on some PIONEER drives

[T <mlist4suntong@yahoo.com>]

Hi, 

Sorry to send the patch here because my ISP does not allow me to send email to bug report.

This patch is similar to bug #323602, but fix another problem. It may solved
the problem raised at the end of message of http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=323602

which is:

,-----
| The important one however is the CLOSE SESSION failed one, which looks
| identical to this bug reportee's problem. Perhaps the patch hasn't been
| applied or something?
`-----

The patch comes from RISKO Gergely's message:

http://article.gmane.org/gmane.comp.audio.cd-record/4805

which I believe that the solution is the same as bug #323602, but at
different a place. Here is the full description of the symptom:

,-----
| I got the same problem as Felix and Gadi had in January, but it
| irritated me enough to solve it.  (quick hack only)
| 
| The problem was (just to remember) with different Pioneer drives, my
| drive is a DVR-109 with firmware version 1.40 (the newest one, but the
| problem was occurred with previous versions too).  When I write a
| Philips 8x DVD+R disk with quite a lot data (my experiences shows,
| that a test amount (say <1G) is not enough to reproduce the bug), the
| write finishes OK, but when growisofs wants to close the disk, I got:
| 
| /dev/hdc: flushing cache
| /dev/hdc: closing track
| /dev/hdc: closing disc
| :-[ CLOSE SESSION failed with SK=2h/ASC=04h/ACQ=07h]: Resource temporarily unavailable
| /dev/hdc: reloading tray
| 
| and after this termination, the disk left in an unmountable state
| (linux says, that media not found):
| [...]
| 
| When I searched the specification for the meaning of
| SK=2h/ASC=04h/ACQ=07h, I found, that it means something like 'drive
| not ready', but not an error, just a query from the drive to wait a
| minute and try again.
`-----

Why I'm redoing this, since there is a solution already? 

- The fix is not officially included yet, as of 2005.12.19.

http://packages.debian.org/changelogs/pool/main/d/dvd+rw-tools/dvd%2brw-tools_5.21.4.10.8-4/changelog

- I was not able to apply the RISKO's patch, but have to do it manually.

- Mine use more "native" solution, ie, poll, instead of usleep.

- The wait time is carefully calculated, only about 1/5 of RISKO's patch.

- The compile code from my source is shorter.

Here it is:

 diff -wu growisofs_mmc.cpp.org growisofs_mmc.cpp    
--- growisofs_mmc.cpp.org       2005-12-19 16:13:38.000000000 -0500
+++ growisofs_mmc.cpp   2005-12-19 16:30:51.000000000 -0500
@@ -1484,8 +1484,17 @@
        cmd[1] = 0x01;          // "IMMED"
        cmd[2] = mode;          // "Close session"
        cmd[9] = 0;
-       if ((err=cmd.transport()))
+
+        // it seems, that pioneer is a bit crappy
+        while (err=cmd.transport()) {
+            if (err == 0x20407) {
+                    sperror ("CLOSE SESSION (but try to continue)",err);
+                   poll (NULL,0,1400);
+            } else {
            sperror ("CLOSE SESSION",err);
+                    break;
+            }
+        }
 
        if (wait_for_unit (cmd)) break;
  
Why poll:1400? I first tried with poll 333, and the message showed up 7 times. So 1400 is just enough for the message to show up only once.

FYI, major part of RISKO's patch (so as bug #323602)

+	    if (SK(err)==0x2 && ASC(err)==0x04 && ASCQ(err)==0x07) {
+		    sperror ("CLOSE SESSION (but try to continue)",err);
+		    usleep(10000);

PS. I hope the fix for bug #323602 can be changed to above 'err == 0x20407' and the 'poll' method as well. 

thanks

Antonio Tong Sun