Arrange buffer-sv2 fuzz and bench correctly #2035
Conversation
Shourya742
force-pushed
the
2025-12-19-arrange-fuzz-and-bench
branch
from
4cc64bc to
86e56ad
Compare
| name = "slower" | ||
| path = "fuzz_targets/slower.rs" | ||
| test = false | ||
| doc = false | ||
|
|
||
| [[bin]] | ||
| name = "faster" | ||
| path = "fuzz_targets/faster.rs" | ||
| test = false |
There was a problem hiding this comment.
I would recommend to change the name of this targets to a more descriptive one.
There was a problem hiding this comment.
Since you have more context of the buffer-sv2 crate itself, do you think these targets make sense?
There was a problem hiding this comment.
running the faster one I get a local execution of ~= exec/s: 29, which is extremely slow and not recommended.
In less then 30 seconds of execution we get an OOM:
SUMMARY: libFuzzer: out-of-memory
| framing_sv2 = { path = "../sv2/framing-sv2" } | ||
| codec_sv2 = { path = "../sv2/codec-sv2", features = ["noise_sv2"]} | ||
| common_messages_sv2 = { path = "../sv2/subprotocols/common-messages" } | ||
| affinity = "0.1.1" |
There was a problem hiding this comment.
Just a note: this crate blocks execution of all the targets on macos. This is used by the slower.rs
There was a problem hiding this comment.
If we are going to keep this fuzz targets, maybe we could put this behind features?
2 participants
part of #2042
This PR moves the buffer_sv2 fuzz targets to the top-level fuzz module. It also makes criterion a dev-dependency; it was previously optional, which caused cargo bench to break. In addition, the benchmarks were not compiling correctly this PR applies minimal fixes to get them building and running. Overall, this brings the benchmarking suite back to a working state.