First of all let me add another "statistic" about the issue:
[tom@localhost ~]$ sudo shred -n 1 /dev/sda3
[tom@localhost ~]$ sudo blkdiscard /dev/sda3
[tom@localhost ~]$ sudo hexdump /dev/sda3 | wc -l
310635
[tom@localhost ~]$ sudo hexdump /dev/sda3 | pcregrep -M '0000 0000
0000 0000 0000 0000 0000 0000\n\*' | wc -l
2410
total untrimmed sectors: (310635 - 1 - 2410) / 32 = 9632
total ranges: (43 GiB * (1024 ^ 3) / 512) / 65535 ~= 1376
average untrimmed sectors per range: (9632 / 1376) = 7 = (65535 % 8)
Also, FWIW:
[tom@localhost ~]$ sudo shred -n 1 /dev/sda3
[tom@localhost ~]$ let step=(65535-65535%8)*512
[tom@localhost ~]$ echo $step
33550336
[tom@localhost ~]$ sudo blkdiscard -p "$step" /dev/sda3
[tom@localhost ~]$ sudo hexdump /dev/sda3
0000000 0000 0000 0000 0000 0000 0000 0000 0000
*
ac0000000
On 24 June 2015 at 02:26, Martin K. Petersen <martin.petersen@xxxxxxxxxx> wrote:
>
> What happens if you discard sectors 0-6 and then sector 7?
>
[tom@localhost ~]$ sudo shred -n 1 /dev/sda3
[tom@localhost ~]$ sudo blkdiscard -l 3584 /dev/sda3
[tom@localhost ~]$ sudo blkdiscard -o 3584 -l 512 /dev/sda3
[tom@localhost ~]$ sudo hexdump -n 4096 /dev/sda3
0000000 f06d 8365 5e1b 616c 7362 4d61 2182 02fb
...
0000ff0 54ef 9579 51bc 9042 115a 375e c28f 4dcc
0001000
>
> This is on the Intel 530? What does the drive report in
> /sys/block/sdN/queue/discard_zeroes_data?
>
Yes. It reports 0. In `hdparm -I`:
* Data Set Management TRIM supported (limit 1 block)
* Deterministic read data after TRIM
You may also want to check out and compare the two attached test case files.
Attachment:
sandisk_test
Description: Binary data
Attachment:
intel_test
Description: Binary data
