nydusd: fixed getting the wrong chunk due to using the wrong blob index #573
Conversation
zyfjeff
changed the title
|
The integration test for this PR is covered by #572 |
|
@zyfjeff , a new test job has been submitted. Please wait in patience. |
|
@zyfjeff , The CI test is completed, please check result:
Congratulations, your test job passed! |
||||||||||||||||||||||||
|
could you leave some words on what it will impact? it impact chunk info? |
|
Why does this patch relate to nydus-image? It looks like a change of nydusd runtime |
You're right, I'll change it. |
zyfjeff
changed the title
zyfjeff
force-pushed
the
fix-wrong-blob-index
branch
from
29bd1be
to
1ea7e3c
Compare
|
@zyfjeff , your pull request has been updated. A new test job will be submitted. Please wait in patience. |
|
@zyfjeff , your test job has passed, and no need to test again. |
This will cause the contents of the file to be confused, or the chunk index to go out of bounds |
|
I don't quite understand why data is corrupted. I think it can only cause wrongly generating |
This will result in a BlobIoVec containing different Blobs, but where BlobIoVec is used, it is assumed that they all belong to the same Blob, but it is not, which will result in chunks being taken from the wrong blob |
…blob index The blob index on the disk where erofs exists starts at 1, but the blob index in BlobInfo starts at 0, and the blob index on the disk is mistakenly used here instead of blobInfo's index Signed-off-by: zyfjeff <zyfjeff@linux.alibaba.com>
zyfjeff
force-pushed
the
fix-wrong-blob-index
branch
from
1ea7e3c
to
118a287
Compare
The blob index on the disk where erofs exists starts at 1, but the blob index in BlobInfo starts at 0,
and the blob index on the disk is mistakenly used here instead of blobInfo's index
this will eventually result in the wrong file contents being obtained.
Signed-off-by: zyfjeff zyfjeff@linux.alibaba.com