> It's not about mechanical movement of the tray, but about tricking > /usr/sbin/vold to reread and remount the media. In more details. First > note that once vold is running you can't open /dev/rdsk/cXtYdZs2 (Oh! I > failed to mention that "volumized" cdrecord does not expect /dev/scg* to > be present). So patched cdrecord finds its way to > /vol/dev/rdsk/cXtYdZ/unknown_format instead. This in turn means that > everything cdrecord throws at it passes Volume Management kernel driver > (hereafter "the V-driver") which is free to do certain magic. Now > 'cdrecord -eject/load' uses USCSI ioctls to open/clode tray which are > passed through verbatim. 'eject cdrom' in turn uses DKIOCEJECT ioctl > which also makes the V-driver throw an exception at vold which triggers > it to do certain magic in the user-land. You may ask why not make > cdrecord issue DKIOCEJECT to eject and then USCSI to load? Well, the > problem is that once DKIOCEJECT passes the V-driver it (so to say) > interrupts all communications and awaits for further instructions from > vold, meaning that the "load" USCSI never gets through and you find > yourself in "catch-22" situation...
Sun should provide a "close tray" command for vold. But yeah I see your point. vold does work rather nicely on solaris in general and disabling it would be unfortunate. Better to get it to work nicely with cdrecord. Len Sorensen -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
