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
When lsquic serves as the server and handles chromium's(client) ietf quic v1 requests, we encountered a problem.
The error log from client-side is as follows:
I chromium: [918:1415:1119/151718.781863:INFO:quic_connection.cc(2153)] Client: Received Transport ConnectionClose for connection: 7330ae0c3284cfd2, with error: QUIC_IETF_GQUIC_ERROR_MISSING (received MAX_STREAM_DATA frame on never-opened stream 0), transport error code: STREAM_STATE_ERROR, error frame type: 0
It seems lsquic close the connection, as stream 0 was not opened, but the MAX_STREAM_DATA frame arrived first.
Following is the code:
When we captured the packet, we saw that the MAX_STREAM_DATA frame encoded before the request for stream 0
It seems normal.
My question:
But, when I read the RFC 9000, it writes:
So, I think when checking whether the stream is already opened at the point receiving the MAX_STREAM_DATA frame, the lsquic need to check if the stream belong to its side first?
We can discuss more about this.
Wait for your reply.
Yours sincerely
Michel Kong
The text was updated successfully, but these errors were encountered:
Lingtaonju
changed the title
It seems lsquic does not comply with the requirements of RFC 9000 when processing the "MAX_STREAM_DATA frame"
[Suspected Bug]It seems lsquic does not comply with the requirements of RFC 9000 when processing the "MAX_STREAM_DATA frame"
Nov 27, 2024
Dear all:
The story begins here:
When lsquic serves as the server and handles chromium's(client) ietf quic v1 requests, we encountered a problem.
The error log from client-side is as follows:
It seems lsquic close the connection, as stream 0 was not opened, but the MAX_STREAM_DATA frame arrived first.
Following is the code:
When we captured the packet, we saw that the MAX_STREAM_DATA frame encoded before the request for stream 0
It seems normal.
My question:
But, when I read the RFC 9000, it writes:
So, I think when checking whether the stream is already opened at the point receiving the MAX_STREAM_DATA frame, the lsquic need to check if the stream belong to its side first?
We can discuss more about this.
Wait for your reply.
Yours sincerely
Michel Kong
The text was updated successfully, but these errors were encountered: