Hi, Max Nikulin wrote: > I assume that nowadays it is more reasonable to expect kernel >= 2.6.
Historically /dev/sg* is the safest blind bet. It is usable for ioctl(SG_IO) since at least kernel 2.4 and not foreseeable to vanish. The only disadvantage is with coordination of drive access, because all reader use cases use the /dev/sr* device files. Before 2.4 i used cdrecord by its peculiar numerical Bus,Target,Lun addresses (e.g. 2,0,0) without knowing what it does internally. I'm not sure what kernel i had when i began to use growisofs which always operated on /dev/sr* or their links /dev/cdrom, /dev/dvd, ... But i remember that growisofs switched from using ioctl(CDROM_SEND_PACKET) to ioctl(SG_IO) when i already was working on libburn at kernel >= 2.4 time. > <https://docs.kernel.org/admin-guide/devices.html> > The prefix /dev/sr (instead of /dev/scd) has been deprecated. Yes. That was the plan 20 years ago. For a while i had both names on SuSE. I don't know why /dev/scd* failed to persist. (The kernel code part was alreay named "sr" when the attempt was made.) > Should it be considered as a documentation bug? In comparison to contemporary distro reality it is obviously bit rot. > I am lost which way /dev/scd<n> were created in the past. Digging in my dim memories ... It happened after Linux gave up the large unconditional collection of device files in favor of files which show up only if they have a drive attached. I first experienced that novel behavior with a USB attached CD drive. It may be that HAL, kindof a predecessor of udev, was already at work. (It was an infamous groper of CD drives in the middle of burn runs which resulted in wasted CD-R media.) Have a nice day :) Thomas
