Send unit_attention on cd-rom not_ready to ready transition

Patch sends a UNIT_ATTENTION(6), MEDIUM_MAY_HAVE_CHANGED(0x28) sense when cdrom transitions from not ready to ready.

I used the following documentation:
ATA Packet interface for CD-ROMS, SFF8020i.pdf. See state diagram Figure 12, page 82 and Table 44 -recommended Sense Key, ASC

With patch in place HVM win2008 server guest sees the CD/DVD contents have changed when the media is switched.

Patch applies to SVN revision 4456.

Index: hw/ide.c
===================================================================
@@ -351,6 +351,7 @@
 #define ASC_ILLEGAL_OPCODE                   0x20
 #define ASC_LOGICAL_BLOCK_OOR                0x21
 #define ASC_INV_FIELD_IN_CMD_PACKET          0x24
+#define ASC_MEDIUM_MAY_HAVE_CHANGED          0x28
 #define ASC_MEDIUM_NOT_PRESENT               0x3a
 #define ASC_SAVING_PARAMETERS_NOT_SUPPORTED  0x39
 

@@ -1385,6 +1386,11 @@
     switch(s->io_buffer[0]) {
     case GPCMD_TEST_UNIT_READY:
         if (bdrv_is_inserted(s->bs)) {
+            if (s->is_cdrom && s->sense_key == SENSE_NOT_READY) {
+                ide_atapi_cmd_error(s, SENSE_UNIT_ATTENTION,
+                                    ASC_MEDIUM_MAY_HAVE_CHANGED);
+                break;
+            }
             ide_atapi_cmd_ok(s);
         } else {
             ide_atapi_cmd_error(s, SENSE_NOT_READY,

Pat Campbell writes ("[Qemu-devel][PATCH] Send unit_attention on cd-rom not_ready to ready transition"):

Was there something wrong with Pat's patch, or has it just been overlooked ?

Ian.

Pat Campbell writes ("[Qemu-devel][PATCH] Send unit_attention on cd-rom not_ready to ready transition"):

...

I have checked that this code compiles, and it seems to make sense (although I haven't double-checked the SCSI specification to check whether this is the right place to send that UNIT ATTENTION message).

I've also confirmed with Pat by email that it should be considered signed off, so:

Full disclosure: I have an interest in this in as much as I encouraged Pat Campbell to submit this patch here as well as to xen-unstable.

Is this expected to fix the following?

Running Windows XP guest, in the QEMU monitor I do "change ide1-cd0 none", then a little while later, "change ide1-cd0 newfile.iso".

Interestingly, Windows XP _does_ show the CD (drive D:) change from having contents to being not present (after the first change command).

But after the second change command, Windows XP shows it changes to having contents again - but the _old_ contents of the drive!

Using the "eject" QEMU monitor command doesn't help.

However, using the eject menu option from within Windows XP itself, then changing the CD image QEMU's monitor - that does work and it shows the new contents.

I'm thinking perhaps XP caches the disk contents, and even though it sees the CD transition from present to not-present and back, it does not flush it's cache for some (bizarre) reason.

Is this patch expected to fix this, so issuing the "change" command in QEMU's monitor will cause XP to show the new contents? Does that imply _all_ CD drives issue the error shown in this patch?

Thanks,
-- Jamie