You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The SYN+ACK produced by libslirp lacks window scaling. This means we have max ~1 in flight packet (over links with MTU=64K), which is not happy.
I'm not entirely sure if having window scaling (ie: sliding window > 64K) would benefit performance, since in most cases the link between slirp and application has small RTT, but I guess it could reduce at least the amount of ACK's flying back.
The text was updated successfully, but these errors were encountered:
@AkihiroSuda feel free to open / move the bug to libslirp.
fwiw, I am not exactly impressed by libslirp. The code is showing age, and has seen a number of vunlerabilities. Yet, we don't have anything better for now. I have done some PoC based on rust smoltcp, which is imho a far more "impressive" work ;)
I'm looking at slirp4netns, and so far I'm very impressed with it (mostly libslirp).
Having said that, it would be nice if libslirp supported window scaling. The TCP handshake:
The SYN+ACK produced by libslirp lacks window scaling. This means we have max ~1 in flight packet (over links with MTU=64K), which is not happy.
I'm not entirely sure if having window scaling (ie: sliding window > 64K) would benefit performance, since in most cases the link between slirp and application has small RTT, but I guess it could reduce at least the amount of ACK's flying back.
The text was updated successfully, but these errors were encountered: