[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <45C5D7AD.9080704@joow.be>
Date: Sun, 04 Feb 2007 13:55:09 +0100
From: Pieter Palmers <pieterp@...w.be>
To: Stefan Richter <stefanr@...6.in-berlin.de>
CC: Dan Dennedy <dan@...nedy.org>,
linux1394-devel@...ts.sourceforge.net, linux-kernel@...r.kernel.org
Subject: Re: [PATCH update] ieee1394: cycle timer read extension for raw1394/libraw1394
Stefan Richter wrote:
> Pieter Palmers wrote:
>> Stefan Richter wrote:
> ...
>>> - Fix integer overflow.
>> I had to use 1000000ULL instead of USEC_PER_SEC to avoid weird behavior.
>
> OK, I'll change that and will wait for...
>
>> I can't test it right now, but I'll report later.
>
> ...your and Dan's ACK before I commit the patch.
Stefan,
I tested the patches as posted on bugzilla, and it looks like it is
working fine.
I attached a test program I used while writing/debugging the patches.
Might be useful for other people to test if this works correctly.
Compile:
$ gcc -g -o ctr_test -lraw1394 ctr_test.c
Run:
$ ./ctr_test
libraw1394 Cycle Timer API test application
using port 0
init rate=24.5837
Local time: 1170593509294313us, 1170593509294.313ms (approx 38year,
20day since epoch)
CycleTimer: 67s, 2323cy, 1894ticks
rate: 24.583702ticks/usec
Local time: 1170593510294462us, 1170593510294.462ms (approx 38year,
20day since epoch)
CycleTimer: 68s, 2328cy, 2044ticks
rate: 24.587107ticks/usec
The rate should be something around 24.576.
Since epoch is somewhere around 1970, the approx 38 years looks rather
correct.
Greets,
Pieter
View attachment "ctr_test.c" of type "text/x-csrc" (7531 bytes)
Powered by blists - more mailing lists