Re: Getting paranoia to quit on CD eject

Peter Jones wrote:
> On Thu, 2004-10-14 at 14:31 +1000, Douglas Gilbert wrote:
> 
>>Monty,
>>BTW There is still a bug in the SG_IO ioctl block layer implementation
>>in the linux kernel 2.6 series (e.g. when /dev/scd0 is used as the
>>device name). The SCSI status value is returned but not the host or
>>driver values (i.e. errors detected at the near end or along the wire
>>are not reported). Hence if one had an external cd reader and the
>>cable was pulled during the transfer then the transport (e.g. USB,
>>iee1394, SPI) knows but cdparanoia won't get a sensible error from
>>the SG_IO ioctl.
>>
>>Hopefully this problem will soon be addressed.
> 
> 
> So what's the plan for how this sort of error will look in the
> sg_io_hdr?

As per the scsi generic (sg) driver in linux where the SG_IO
ioctl works as advertised.
See http://www.tldp.org/HOWTO/SCSI-Generic-HOWTO

Doug Gilbert