gpsd-dev
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: b32ff1a86 Breaks OSX and FreeBSD


From: Greg Troxel
Subject: Re: b32ff1a86 Breaks OSX and FreeBSD
Date: Sat, 21 Dec 2019 18:58:34 -0500
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/26.1 (berkeley-unix)

Bernd Zeimetz <address@hidden> writes:

> On 12/22/19 12:48 AM, Greg Troxel wrote:
>> I just did a run with WRITE_PAD=0 on NetBSD/8 amd64, running bare metal
>> on a fairly modern machine (5 years old, 4 cores, 24G of ram) that seems
>> pretty fast.
>> 
>> So I concur that there must be a bug lurking someplace.  This smells
>> like the writes to the pty master side not waiting or not returning
>> partial writes correctly (or having those partial writes accounted for).
>
> If you are using scons check or similar things - the default seems to be
> tcp, not pty. My first idea was that the pty stuff is broken, but after

I am using scons check.

> realizing we are using tcp, I doubt its an issue with the transport. I
> assume tcp is well enough tested on all BSD variants.

Yes.  The code dates from the 80s and it is pretty well shaken out :-)
And, more or less if the TCP code were buggy (other than not quite
retransmitting on the right schedule to exactly meet the congestion
control specs) sort of thing, it would show up and would have been
fixed.  This could be the one thing that trips a bug, but that seems
unlikely.




reply via email to

[Prev in Thread] Current Thread [Next in Thread]