Preserve EDNS OPT record in a response if truncation occurs #1364
Conversation
| @@ -780,10 +797,10 @@ where | |||
| let mut additional_count = count_was_truncated(additionals.emit(encoder))?; | |||
|
|
|||
| if let Some(edns) = edns { | |||
| encoder.set_max_size(real_encoder_max_size); | |||
| // need to commit the error code | |||
There was a problem hiding this comment.
What does this comment mean btw?
There was a problem hiding this comment.
I think that is referring to the fact that the response error code may need to be committed to the EDNS record... i.e. in EDNS the response (error code) is has high-order bits stored in the EDNS record that augment those in the original DNS Header. So I think that's what this may be talking about...
Codecov Report
@@ Coverage Diff @@
## main #1364 +/- ##
=======================================
Coverage 86.42% 86.42%
=======================================
Files 131 131
Lines 13766 13826 +60
=======================================
+ Hits 11897 11949 +52
- Misses 1869 1877 +8 |
vlad20012
force-pushed
the
edns-truncation
branch
2 times, most recently
from
97ad6da
to
15ecb3f
Compare
| @@ -113,6 +113,18 @@ impl Edns { | |||
| pub fn set_option(&mut self, option: EdnsOption) { | |||
| self.options.insert(option); | |||
| } | |||
|
|
|||
| pub(crate) fn len(&self) -> u16 { | |||
| 1 // Name::root | |||
There was a problem hiding this comment.
I wonder if we should consider adding some consts for these things, or perhaps instead use std::mem::size_of It would make the code a bit more self-documenting. Thoughts?
I know we have a similar practice elsewhere in the code, maybe I'll file a cleanup issue for this.
There was a problem hiding this comment.
Hmm, could we use EncodedSize for this?
Let's do this in another PR anyway
|
@vlad20012 please confirm that this is ready to land. If it is, either update the branch to current main, or confirm and I can update this branch. Thanks for this PR! |
vlad20012
force-pushed
the
edns-truncation
branch
from
15ecb3f
to
d65008d
Compare
vlad20012
force-pushed
the
edns-truncation
branch
from
d65008d
to
0e6682a
Compare
|
hm, it looks like this got approved and reviewed, and then we missed the fact that it was ready to go back then. Is this still worth doing? |
|
@vlad20012, do you think this is something worth updating and getting in? I think I missed when you updated this before... |
According to RFC 6891
Hence, we must preserve EDNS OPT record in an answer when truncation occurs