Sunday, June 29, 2014

Notes onto RTOS Linux

Personally, I'm not one to make any sort of lengthy association between culture and computing. It's my point of view that a computer is a practical thing -- nothing of "style" or of too much of conversation, to me. I also don't talk for long about my car, or my experiences in driving, or my experiences in operating heavy engineering equipment. Those are all practical things, to my point of view. If I might have any sort of a "style" about computing, regardless, I suppose "Tight lipped" would be the label that I would prefer, at that -- "Tight lipped" as contrasted to my sometimes bouts of rambling inconherently, and likewise of writing, though without a great sense of structure.

I understand that as I would continue to pursue a course of formal, academic study in computing and -- correspondingly -- a study in digital electronics, that I should wish to learn to develop a normative style of writing, then -- such that I would not be thought to be inept at communication, at least, though I am inept in much of "Style," I think. It's on the premise of computing being a practical thing, to me, that I endeavor to not be "Self conscious" about writing about computing -- even when I am writing without a great sense of structure or of focus.

If I would begin to write with more of a sense of structure and/or a sense of focus, then, I think it could go well if I was to begin to write about a single, linear process in development. At that, the first thing that comes to mind that I would like to write about: A manner of study, with regards to applications of the Linux kernel in realtime systems on thin client platforms. Of course, that in itself might not be so much of a simple topic to write about. If there is even a single vocabulary to the concept, I have not learned it yet, and if there is single, discrete set of technologies to the applications of such a concept, realtime systems on thin client platforms, I have not found that set of single, discrete technologies, as yet. Candidly, being a student, I have found at least a small assembly of hetereogenous components, some of which share some common qualities -- for instance, that a BeagleBone Black and an Arduino board can both conduct communications using the I2C protocol,  and that the Altera Cyclone III on eSOC FPGA board manfucatured for DeVry University students (a group which I am currently one of)  .... that it has IO headers at least, though he eSOC 3 is currently lacking any kind of a user manual, schematic, or other design documents, and therefore there's not much I can do with it.

In a practical regards, I think that a realtime Linux platform could be of some interest with regards to parallel computing models using single board computers, if not furthermore of Cubesat design -- neither of which would seem to be hot topics in "the consumer market," at this time, however. I'm sure that there could be an app made for news about it -- maybe it could be more popular, then. Certainly, it's a bit more complex than what can be developed on a mobile phone.

In all that complexity, perhaps there are some discrete qualities, however -- such that one may endeavor to "Snapshot," in a manner of writing -- resources, such as:
  • Article: Jones, M. Tim. Anatomy of real-time Linux architectures. IBM Developerworks. 2008
    •  In a structural view: The article denotes three primary types of Realtime Linux kernel, namely: Thin-kernel; Nano-kernel; Resource kernel. The article also describes the baseline architectures of the Linux 2.4 and Linux 2.6 kernels.
  • Project: ADEOS Project
  • Realtime Linux distribution:  Timesys LinuxLink
    • This item, in particular, has really "Caught my attention," this week. I denote it, here, as a realtime linux distribution considering that it is an apparent descendant of a Linux distribution denoted in "that IBM article" (Jones op. cit.) namely, Linux/RK, denoted there as a realtime Linux distribution implementing a resource kernel model
    • The Linux/RK project page, itself, makes reference to a number of additional resources, describing concepts of the architecture of a resource kernel model as implemented in Linux/RK. (Note that the resources are dated around 2001, and are available in Postscript format, but may be converted to PDF format with some GPL licensed tools such as 'ps2pdf', cf. Cygwin)
    • In a practical regards, I've noticed that the Timesys web-based Build Factory provides support for the BeagleBone Black single-board computer and the TI Sitara microcontroller unit used on the BeagleBone platform. I'm well impressed, moreover, to see how well architected the Timesys Build Factory is. I've compiled a single build with it, and have yet to install that to my BeagleBone Black, but I look forward to that -- though catching up on some other  items, in the meanwhile.
    • This item, in particular, I think I may be able to incorporate into a senior project. Though the project would be focusing primarily on developing an I2C I/O model using the BeagleBone Black, but with there being a solid baseline OS avaialble -- namely, LinuxLink -- as already catered towards realtime and embedded systems, that substantial bit of existing work could simply help -- by a long shot -- in defining a modular design, in the project.
    • For a bit of historical context, there's an article about Timesys' original Linux/RT, at the Linux Journal
  • Standard: POSIX 1003.1
 From there, the snapshot begins to focus more on the development environment, less on the OS itself.
  •  Virtualization platform (x86 architectures): Virtualbox
    • My having recently begun to develop a Windows 8.1 desktop on an x86 (amd64) PC, such that uses a UEFI architecture in its boot process -- that, adding some distinct complexity for any sort of a "Dual boot" configuration -- and the same PC has an intel graphics card that doesn't work ideally well, on Mint LMDE, "Out of the box", it's therefore my impression that Virtualbox provides a "Nearest to ideal" methodology, then, for both developing software in the Linux environment and continuing with the typically Microsoft Windows-dependent lab materials in the courses at DeVry University Online (DVUO)
    • I've installed Linux Mint (LMDE edition 201403) into that Virtualbox image, and have begun to use that desktop as an overlay onto my Microsort Windows 8.1 desktop. 
    • Virtualbox provides a nice seamless view mode for graphical applications in compatible Virtualbox virtual guest instances. When Virtualbox is using the seamless mode -- as for displaying GUI applications running directly in the compatible virtual guest instance -- it makes for a relatively smooth integration with the virtual host PC desktop
    • There are some limitations -- such as namely:
      • That the desktop menu of the virtual guest instance is inaccessible in seamless mode. However, there are some quick hotkeys that Virtualbox provides for changing the view mode, such that allows for access to the desktop men, as in fullscreen view mode.
      • That it does somewhat "lag the PC", it being a virtual OS image for an OS that would typically be installed directly onto "Bare metal" hardware
      • Those are, effectively, the only limitations I can observe of that configuration
    •  Of course, it would also work "Swell" with the build factory image created by the Timesys web-based Build Factory
      • Virtualbox can be easily configured to provide "pass through" I/O for USB devices, such as would allow for the BeagleBone to be configured directly via USB from within the Virtualbox virtual guest instance
      • With or without an window server running in the Virtualbox virtual guest image, Virtualbox nonetheless provides a framework for a full Linux development environment, without the characteristically "hairy" qualities of the process of installing a second bootable OS diretly onto a PC that uses UEFI.
      •  The virtual media device as used by Virtualbox for its machine and block storage device emulation procedures, that media device is essentially contained within a single virtual media image, such that can easily be backed up and/or relocated to another PC.
    • Advice: To use Samba and CIFS instead of the Virtualbox "Shared folder", for sharing files between the Virtualbox virtual guest image and the virtual host OS.
      • "Some limitations would apply"
  • Programming Language: Common Lisp
    • This topic necessarily would entail a number of distinct sub-topics, such as at least one topic for each Common Lisp implementation being used in a single Common Lisp development environment. 
    • Of course, one cannot simultaneously use every Common Lisp implementation that exists. Personally, I wish to focus on:
      • Steel Bank Common Lisp  (SBCL) 
        • towards: Defining a "New" Lisp OS
        • Focusing on platforms:
          • amd64
          • armhf (where applicable, else armel)
            • See also: Connors, Jim. Is it armhf or armel?. Jim Connors' Web Log. 2013
            • Shell command: readelf -A /proc/self/exe | grep Tag_ABI_VFP_args
            • NOTE: The distro might be compiled as armel, though the MCU might support armhf features, regardless
      • Clozure Common Lisp (CCL)
        • towards: A Common Lisp implementation providing close integration of existing software components on the Linux platform -- typically via FFI -- including the Linux Kernel itself
  • Concept: MPI
    • Resource: ParGCL, another Common Lisp implementation
  • Concept: I2C
    • Short summary: Standard I/O protocol for digital electronics devices onto a common digital signal bus
    • Supported in the two PRU-ICSS modules in the TI Sitara MCU (cf. BeagleBone Black)
  • Concept: CORBA
    • Short summary: Platform-neutral architecture for remote data procedures and remote procedure calls within heterogenous networks
    • Concepts: CORBA GIOP, CORBA IIOP
    • Concept: CORBA IDL
 That, then, would effectively serve to complete the "Snapshot," this morning -- the "Snapshot" amended, then, with subsequent updates.