2002-10-11 00:41:37

by Andries E. Brouwer

[permalink] [raw]
Subject: tcp urgent data broken since 2.5.34?

Since 2.5.34 I see a lot of processes hanging for a while.
I just looked at what might cause this, and I get the strong
impression that something went wrong with the 2.5.34 patch
that introduced sk_send_sigurg() (or possibly the signal
handling).

Phenomenon:
B->A: urg 1
B->A: some data
A->B: ack for the urg
half a minute later:
B->A: resend of some data
A->B: ack

This is using rlogin/rlogind.
Will look further later, but in the meantime somebody might
see immediately what causes this.

Andries