Re: [Libcdio-devel] problems with libcdio on openbsd (and probably netbsd too)

[Robert Kausch]

Am 27.05.2018 um 14:52 schrieb Thomas Schmitt:

> Is there documentation about what a raw partition can do which the others
> cannot ?

I don't know, sorry. I'm not a NetBSD developer. Just ported fre:ac to 
the BSDs back then and most of my knowledge is from trying it out and/or 
reading the source code.

> libburn tries with 'c' and 'd' :
>
>    fd = open(path, O_RDWR | O_NDELAY);
>    ...
>    ret = ioctl(fd, SCIOCIDENTIFY, &addr);
>
> and wants to see after successful ioctl():
>
>    addr.type == TYPE_SCSI
>   
>  From then on, the drive is regarded as usable for SCSI transactions
> via ioctl(SCIOCCOMMAND).
>
> Is this too daring ?

I *think* this should work, but it would probably be better to use 'a' + 
RAW_PART. Two reasons:

1. Future architectures might define RAW_PART to 4 or something else, so 
you'd suddenly get devices like /dev/rcd0e on such systems.
2. Not sure if it's actually possible, but on architectures with 
RAW_PART == 3 there might still be a /dev/rcd0c device in some cases, 
but with a different meaning. Treating it like a raw disc device could 
possibly cause bad things to happen.

----
Robert Kausch
address@hidden