Thank you for verifying that hald still is a major problem. You could reduce the problems that appearh while writing by opening the devices with O_EXCL but this does not fix all problems with hald and please note that cdrecord calls libscg for the SCSI command transport. The same library is used by many other programs with different constraints. Cdda2wav will e.g. be unable to do audio extraction with many CDs in case libscg did open the devices with O_EXCL. Also note that there is a general conceptional problem in Linux that is a result from the fact that you may have more than one device driver for a given device. If you open one device with O_EXCL, you may still open the other device withour any problems. Let me add some notes: As the way hal has been introduced into Linux did cause a lot of trouble: - Hald was introduced on Linux without contacting the authors of the software it influences. This is one reason why hald interrupts the CD/DVD/BluRay burning process. - Hald (on Linux) uses an algorithm for detecting media change conditions that is wrong (it hits even when there never was a media change). As a result, hald triggers read operations on partially written media that is currently written to in "uninteruptable streaming mode". This creates coasters. As it caused the Optical Media software to see strange SCSI error codes, the users are confused and do not easily understand that a related write error was caused by hald. - It seems that there is no concept for an interface that would allow a safe way to write CD/DVD/BluRay media in multi-session or multi-border mode. - It seems that hald is based on OS specific implementation details and for this reason would have portability problems in case that there would be an interface for optical media software. What I need is: - A media change detection system that stays quiet (after some time - e.g. after 3 seconds) in case that it detects the insertion of blank media. This is implemented on Solaris since 1992. - A system that can be told that there is new content without a media change (write process finished) and mounting or multi-media playing is possible. - A working way to claim exclusive access to a drive that still allows to send e.g. TEST UNIT READY and INQUIRY SCSI commands. - A working way to tell the system to forcibly unmount the media at some time in order to allow reliable multi-session operation. Active multi media playing software should be shut down from this interface. - The way the related drive "name" is specified must not be OS platform specific. Note that most platforms (including several "UNIX" platforms) do not not have a UNIX like /dev/* entry for the device and many (including Mac OS X) do not even associate a file handle to the interface that allows to send SCSI commands. For this reason, /dev/* or file handle based interfaces are highly non-portable.