On Thu, 2004-11-18 at 11:09 -0500, Alan Cox wrote: > On Thu, Nov 18, 2004 at 10:54:49AM -0500, David Zeuthen wrote: > > Really? Because using the BLKGETSIZE64 ioctl() on the device file we get > > Yes - see the specification. > I'm curious why BLKGETSIZE64 ioctl() then succeeds and give me the size if you say it's bogus? Care to explain the 4096 byte difference? Hmm, it looks like I should put code in hal to set size as zero for CD-R and CD-RW media [1] and then make the fs detection code in hal not look at the end of the disk if size is zero. [1] : Exactly what media does this apply to? All non DVD-ROM and CD-ROM? Btw, wouldn't it be a lot nicer if the kernel started to give out e.g. /sys/block/hdc/hdc%d for the various partitions and sessions (assuming hdc is an optical IDE drive)? I guess the kernel could safely give the right size of /sys/block/hdc/hdc1. With that it would be a lot easier to support e.g. PhotoCD discs (which are multisession). FWIW, that's what Darwin does. Cheers, David