Haiku Archive

This week I continued work on moving Beta1 forward, fixing some important and less important bugs. To make things clear about what to expect in the upcoming weeks, I will spend more time on Beta1 tasks, but I'll also continue working on WebKit. However, my work there will focus on fixing bugs, rather than adding new features.

As I talked to many attendees about various things like our package management, scheduler update, WebPositive progress, Wi-Fi, ASLR/DEP, and anything else I could think of, and there was an overwhelming positive energy about Haiku by those who saw it in action. By far the most common question I got was "When will the next release be out?". In the past, I would say the most common question is "Why would I ever choose Haiku over any existing Linux distribution?", so it is nice to see that there was a lot more positive energy about Haiku, as well as excitement about the next release.

While there are some minor downsides to having the kits on top of Linux (or one of the BSDs), the upsides include all the drivers in the world (well, the gpu driver situation could be a tad better), a rock solid kernel that works on all kinds of devices (who says BeOS can't run on a phone, mine can), and a working BeOS clone with comparatively little effort (as a musical engineer, my biggest worry was sound system latencies, but it turns out many Linux schedulers can easily be tuned to handle the loads I expect in a BeOS system.)

I think the Haiku project made a monumental mistake in not using an existing kernel - it's simply no longer practical for a small project to keep up with the hardware evolution, handling security requirements and so on. Sad but true, and it was sad but true back in 2001.

My area of interest is broadly defined kernel development. In the spring of 2013 I implemented ASLR and DEP which caused minor confusion due to "activation" of bugs that have been hidden but I think that overall it worked out well for Haiku. Later I tinkered a bit with RTM (Restricted Transactional Memory), new extension introduced in Haswells but the code will need a lot of work before it will become usable. From October to mid-January, I was employed by Haiku, Inc. to work on the scheduler and adaptating Haiku for work on systems with more than one logical processor. Among other things, I got rid of the 8 processors limit, which was quite firmly rooted in the ABI inherited from BeOS.

The main scheduler logic has been completed and now I am concentrating mainly on bug fixes, adjusting tunables and some minor improvements. I also removed gSchedulerLock, a spinlock I mentioned in my last post, and replaced it with more fine grained locking. An new interfaces for cpufreq and cpuidle modules has been created together with a cpufreq module for Intel Sandy Bridge or newer cores and cpuidle module for all processors that support C-states and invariant TSC. Furthermore, IRQs (including MSI) can be now directed to an arbitrary logical processor. Implementation of inter-processor interrupts has been improved so that it avoids acquiring any lock if it is not necessary and supports multicast interrupts. And, last but not least, 8 processor limit has been removed.

I'm proud to announce that, at last, the package management branch has been merged into the main development line, aka master branch. The builds and nightly images from hrev46113 on will include the new feature.

I only notice now that I should probably have given Matt a heads-up upfront, since due to the somewhat changed build process the build bot will need an update. So there might be a bit of a delay until the first Haiku PM nightly images will hit the website. Sorry for that.