[Yaffs] Recent yaffs changes and testing

Top Page
Attachments:
Message as email
+ (text/plain)
Delete this message
Reply to this message
Author: Charles Manning
Date:  
To: yaffs
Subject: [Yaffs] Recent yaffs changes and testing
Hi all

There have been some recent yaffs changes that fix various things and add some
improvements. Most of these are in yaffs_guts.c and can be seen here:

http://www.aleph1.co.uk/cgi-bin/viewcvs.cgi/yaffs2/yaffs_guts.c?view=log

While I've done reasonably extensive testing, I appreciate it when others find
other interesting tests and any feedback is more than welcome.

At this stage there is one issue that has been lurking for ages that I am
still investigating: yaffs fails fsx. With some tweaks I have been able to
get fsx to run for over an hour on simulated nand which is approximately
equivalent to approx 5 - 10 hours on real flash. It is interesting though
that many tests such as tests in the Android test suite, which use similar
calls, runs for over a week without crashing.

In the past I have had some stress tests only fail after over 100,000 test
cycles and my new benchmark is to run the main stress tests for 1 million
cycles every week end.

Testing is getting more and more focus with at least one day a week being
devoted to improving tests. At the moment I have a dual core laptop that
spends over 50% of the time running two parallel tests to keep both cores
maxed out. Over the next few weeks I intend to set up a quad core box that
will just run these tests on a continuous basis.

-- CHarles