Hi, i wrote: > > Back in 2020 i would quite surely have noticed > > if that behavior had been shown.
Richmond wrote: > lsblk -b -o VENDOR,MODEL,SIZE,PHY-SEC,LOG-SEC /dev/sr* > VENDOR MODEL SIZE PHY-SEC LOG-SEC > TSSTcorp TSSTcorp_DVD+_-RW_TS-L632H 1073741312 512 512 > > 4.19.0-23-amd64 #1 SMP Debian 4.19.269-1 (2022-12-20) x86_64 GNU/Linux So this behavior is not new. > This is a laptop where I rarely use the CD/DVD. Note it is not the same > computer as was getting errors before, that was debian 11. I assume that you will see the same result there. It would explain the block address of the first read attempt and the log messages about unaligned access. kernel: [ 9.507304] sr 3:0:0:0: [sr0] tag#12 CDB: Read(10) 28 00 00 07 ff fc 00 00 02 00 kernel: [ 9.507731] sr 3:0:0:0: [sr0] tag#31 unaligned transfer If you have some blank optical medium, then try whether the emitter of the read attempt can be discouraged if the drive is perceived as offering just one block of 2048 bytes. ---------------------------------------------------------------------- The unresolved riddle is which software began to try reading from the optical drive after your recent package installations or upgrades. It might be a different one than the software which does similar read attempts on piorunz' machine since years. In any case these softwares inquire the device capacity from systemd or directly from the kernel. Then they try to read end and start of the obtained capacity. There are many motivations to read the start of the device and fewer to read its end. One reason to read the end is the GPT backup header, which would sit 512 bytes before the end. The main GPT header block is at byte 512 of the storage device. This would explain the next failed read attempt: kernel: [ 9.620170] sr 3:0:0:0: [sr0] tag#3 CDB: Read(10) 28 00 00 00 00 00 00 00 02 00 I wonder why the caller did not want only 512 or 1024 bytes. The number of 2 requested blocks cannot come from the read-ahead mechanism of the Linux block layer. I would still place my bet on something like blkid, which wants to know whether there is a partition table. Maybe there was a change about a customer of libblkid. Have a nice day :) Thomas
