[SOLVED] Why does badblocks check more blocks than the disk has?

RandomTroll · 04-04-2020, 05:50 PM

badblocks finds 80,475,544 bad blocks.

Code:

dumpe2fs -b /dev/sdb1

returns no blocks.

Code:

 e2fsck -L BadBlockList /dev/sdb1

returns

Quote:

Bad block xxxx out of range; ignored

for every entry in BadBlockList.

Code:

 dumpe2fs -x /dev/sdb1

returns

Quote:

Block count 29,304,950

.

The lowest block in BadBlockList is 36,744,256.

I see a problem:

Code:

 badblocks /dev/sdb1

reports

Quote:

Checking blocks 0 to 117219799

- way too many. Where does it get this number?

scasey · 04-04-2020, 06:00 PM

Please show us the exact commands you're running. On my 'puter, all of those commands require the specification of a device name, and simply return syntax instructions without one.
I also note this in the dumpe2fs man page

Code:

Note: When used with a mounted filesystem, the printed information may be old or inconsistent.

and

Code:

-x     print the detailed group information block numbers in hexadecimal format

RandomTroll · 04-04-2020, 10:52 PM

Quote:

Originally Posted by scasey

Please show us the exact commands you're running. On my 'puter, all of those commands require the specification of a device name, and simply return syntax instructions without one.

I thought anyone could figure that out, but I put them in for you.

Quote:

Originally Posted by scasey

I also note this in the dumpe2fs man page

Code:

Note: When used with a mounted filesystem, the printed information may be old or inconsistent.

The filesystem was not mounted.

Quote:

Originally Posted by scasey

and

Code:

-x     print the detailed group information block numbers in hexadecimal format

[/quote]
group information block numbers, not the summary information. -x was irrelevant to the information I reported.

scasey · 04-05-2020, 01:12 AM

OK. Not at my ‘puter just now, but please review the man pages about the differences in default block sizes used in the various tools.

RandomTroll · 04-05-2020, 09:05 AM

Quote:

Originally Posted by scasey

please review the man pages about the differences in default block sizes used in the various tools.

I had to do that in the first place, as I rarely use these tools. The information I reported is correct and in the same base. The question is 'whence gets badblocks the number of blocks to check?' It uses too-high a number for this disk, more than 3 times the number dumpe2fs reports, which corresponds to the size of the disk as reported by the documentation, thus reports 87 million bad blocks, because they aren't there.

rknichols · 04-05-2020, 09:35 AM

Read them again, especially the part of the e2fsck manpage that reads:

"Note that the block numbers are based on the blocksize of the filesystem. Hence, badblocks(8) must be given the blocksize of the filesystem in order to obtain correct results."

and the part of the badblocks manpage that reads:

"-b block-size
Specify the size of blocks in bytes. The default is 1024."

and note that "0 to 117219799" is 117,219,800 blocks, which is exactly 4 times the filesystem block count of 29,304,950.

RandomTroll · 04-06-2020, 01:05 AM

Quote:

Originally Posted by rknichols

Read them again, especially the part of the e2fsck manpage that reads:

"Note that the block numbers are based on the blocksize of the filesystem. Hence, badblocks(8) must be given the blocksize of the filesystem in order to obtain correct results."

and the part of the badblocks manpage that reads:

"-b block-size
Specify the size of blocks in bytes. The default is 1024."

and note that "0 to 117219799" is 117,219,800 blocks, which is exactly 4 times the filesystem block count of 29,304,950.

That portion of the man page of e2fsck goes on to say

Quote:

it is much simpler and safer to use the -c option to e2fsck.

I ran badblocks only because 'e2fsck -c' failed in the same way, tried to correct all 87 million errors. And I had an off-by-one error dividing 117219799 by 29304950. So I really need to use the -B on e2fsck. Thanks.

RandomTroll · 04-06-2020, 01:31 PM

dumpe2fs reports the correct block size - how can e2fsck & badblocks get it wrong?