Re: ssh working as designed?
From: Scott Schwartz (schwartz+@usenet)Date: 04/26/02
- Next message: Richard E. Silverman: "Re: Disallow logins, but allow ftp?"
- Previous message: Kenneth Simpson: "Re: deny non-tunnel'd services?"
- In reply to: Neil W Rickert: "Re: ssh working as designed?"
- Next in thread: Markus Friedl: "Re: ssh working as designed?"
- Reply: Markus Friedl: "Re: ssh working as designed?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
From: Scott Schwartz <"schwartz+@usenet "@bio.cse.psu.edu> Date: 26 Apr 2002 01:50:19 -0400
Neil W Rickert <rickert+nn@cs.niu.edu> writes:
> >what causes this behavior?
>
> A flaky network.
On a slightly different topic, openssh 3.1.1 has lots of code like
this:
res = (f) (fd, s + pos, n - pos);
switch (res) {
case -1:
if (errno == EINTR || errno == EAGAIN || errno == EWOULDBLOCK)
continue;
but isn't it true that there are other errno values, like
EHOSTUNREACH, that also mean that the operation should be retried?
Ssh certainly shouldn't drop a connection just because a router flaps.
For that matter, why are all these "if (....)" expressions duplicated
throughout the program? Maybe there should be one function,
try_again(errno), that puts the logic in one place.
- Next message: Richard E. Silverman: "Re: Disallow logins, but allow ftp?"
- Previous message: Kenneth Simpson: "Re: deny non-tunnel'd services?"
- In reply to: Neil W Rickert: "Re: ssh working as designed?"
- Next in thread: Markus Friedl: "Re: ssh working as designed?"
- Reply: Markus Friedl: "Re: ssh working as designed?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]