Inferno Distributed Operating System

Geoff Collyer has provided news on the arm ports From: geoff@pla... Subject: changes to the ARM SoC ports Date: Tue, 27 Apr 2010 15:05:02 -0400 booting(8) has been updated; take a look if you're using an ARM port. On the Kirkwood SoCs (Sheevaplug ... [More] and Openrd-client), USB now works. Nemo was helpful fixing this, as usual. One of the people here is working on using the Kirkwood crypto acceleration hardware. The OMAP3530 port is now available in /sys/src/9/omap. It currently runs on the IGEPv2 board. The hardware can execute VFPv3 floating-point instructions, but 5[cal] don't generate them yet, so floating-point is currently emulated. USB isn't quite working yet. Once it is, we should be able to use USB Ethernet and thus run on Beagleboards. The ohci and ehci controllers are seen, but no devices yet. There are several USB errata that need to be looked into. From the latest omap3530 errata: - 3.1.1.130 only one usb dma channel (rx or tx) can be active at one time: use interrupt mode instead - 3.1.1.144 otg soft reset doesn't work right - 3.1.1.183 ohci and ehci controllers cannot work concurrently - §3.1.3 usb limitations: all ports must be configured to identical speeds (high vs full/low) This port is being made available now primarily as a basis for GSoC students; we expect it to improve. Rae McLellan of Bell Labs deserves thanks for helping to decrypt what passes for hardware documentation these days. Mail thread here. Information on the Plan 9 GSoC project is available here. From: geoff@pla... Subject: arm ports update Date: Wed, 12 May 2010 23:42:50 -0400 The kw port now supports the Guruplug Server Plus, including both Ethernet interfaces, and probably the other Guruplugs. booting(8) now has the necessary instructions to get started. They are more diverse than one might like because every version of u-boot we get for a new board seems to have had the dhcp, bootp and tftp commands tinkered with to behave slightly differently. We have two Guruplugs and one has been stable but the other is prone to random resets (and runs much warmer than the Sheevaplugs). I'd be interested in hearing from anyone else who sees random resets. I've imported the flash memory support from native Inferno, other than the flash translation layer, which was developed for nor flash and is suspect with nand flash. flash(3) describes the interface. It seems to work on the Kirkwood boards, but I haven't exercised it extensively. It does implement software ECC. /dev/flash looks like it always returns zero bytes on the igepv2 board, but lack of documentation makes it a little hard to tell what to expect. Mail thread here. Information on the kirkwood is available in this pdf. (5495468kb) [Less]

The interesting part of the Google Summer of Code has finally started, and we have six interesting and prospective projects this year. Venkatesh Srinivas is working on an improved virtual memory manager for Plan 9. I am proposing to improve the ... [More] virtual memory system of Plan 9 and to use the improved VM to implement a better libc memory allocator. Currently, Plan 9 allows programs to create anonymous memory segments; it has a limit of less than five segments per-process, however. I propose replacing the limited per-process segment array with a structure that allows for a much larger number of segments and implementing a memory allocator that takes advantage of the liberal number of anonymous segments. Mentor: Erick Quanstrom Project page. Jesus Galan is going to be intergrating some improvements to 9vx. When 9vx was released some years ago it was presented as an experiment, a demostration of what was possible combining vx32 and Plan9. Time passed by and, today, the 9vx experiment can be considered a sucess: it is replacing drawterm in many Unix systems, is a good test bed for experimentation and trying Plan 9 is easier than ever. The objective of this project is to integrate into the 9vx source some improvements done by 9fans and explore how to make it better. Mentor: Devon O'Dell Project page. John David is working a Plan 9 approach to Heirarchical Patch Dynamics (9hpd) This project builds simulation and modeling tools used for spatially-explicit multiscale integrative modeling using a Plan 9 reimplementation of the (HPD-MP). Successful completion will include a series of multiscale modeling examples running on 9vx and/or Inferno in a distribution that focuses on ease of use and accessibility for non-computer scientists. Hierarchical Patch Dynamics Modeling Platform Mentor: Ron Minnich Project page. Michael Block is doing work on input methods in Plan 9. I propose implementing a system that enables Plan 9 to accept non-English input. Mentor: Noah Evans Project page. Andre Guenther is planning on porting drawterm to the iPhone. Goal of this project is to port drawterm to the iPhone. Drawterm is a mean of making a cpu connection to a Plan 9 machine without running a Plan 9 terminal. With the possibility to mount client devices into the host system enables interesting interaction with the iPhone platform. Thus, this port should also encourage future experimentation with the mobile usage of Plan9. Another interesting aspect is, how the mouse and keyboard driven interface of rio translates into a touchscreen based device. Mentor: Jeff Sickel Project page. Per Odlund is working on another arm port, this time to the IGEPv2. All though how great software might be, if we can't use it, it's value deteriorate. One of the most common questions of people that want to try or expand their usage of plan9 is: does it work on x or y? As the mobile devices get more and powerful it would be intressting to see the usage of plan9 in this area. Adding support for IGEPv2 , Cortex A8 and armv7 is a first step in this direction. Mentor: Charles Forsyth Project page. [Less]