With the SG_IO patch applied, cdparanoia does the SG_EMULATED_HOST ioctl on /dev/srX instead of /dev/sgX. That ioctl is executed in linux >=2.6.15 in block/scsi_ioctl.c's sg_emulated_host() (instead of sg.c's sg_ioctl()). The comment above that function tells us that the author knew no reason to differentiate between SCSI and emulated-SCSI and thus always returns emulated-SCSI. This makes cdparanoia choose a wrong command set for my SCSI Plextor PX-40TS. So either sg_emulated_host() needs to be fixed or cdparanoia must work around it.
Created attachment 92850 [details] cdparanoia_vs_Plextor_PX-40TS.txt Hi, I second this. I have a PLEXTOR PX-40TS and a PLEXTOR PX-W124S and cdparanoia-3.9.8-r4 (which introduced the SG_IO patches) doesn't work with both drives. I attached the output of cdparanoia so you can see the errormessages I can see when I try to rip an audio-CD. cdparanoia-3.9.8-r3 works flawlessly. Cheers Poly-C
Created attachment 92851 [details] cdparanoia_vs_Plextor_PX-W124S.txt
http://forums.gentoo.org/viewtopic.php?p=3494216#3494216 Another case in the above thread, this time an old TEAC CD-R55S 1.0R. Maybe for the time being put the patches under a use flag?
Hi, could you please test if cdparanoia works when you use the generic scsi-device /dev/sg? instead of /dev/sr? (or any /dev/cdr{om,w} symlink pointing to /dev/sr?)? It seems as I have to use /dev/sg3 for my PLEXTOR PX-40TS to get cdparanoia ripping audio-CDs Cheers Poly-C
I've removed the SG_IO patches from 3.9.8-r5, they caused more pain than good. I'll try to bump to the latest version as soon as I can, and see to merge the current patches. New rule about cdparanoia: don't apply redhat patches.