.
Annunci online

All that comes in the mind of an italian guy moved to california
2 novembre 2010
Technologies: AROS: Closing the loop

What is this? Read and see....

After at least three failed attempts (Evilrich and then Tigger, disappeared almost immediately,Greg 'bheron' Casamento - that had a good amount of bad luck and, very shortly, Gary 'Gaz' Pearman), finally the Kickstart integration bounty started to deliver results and code.

The latest person to get the task is called Jason McMullan; this time he started to submit patches for 68k ROM in the developers mailing list around the begin of october. In october 6, deadwood asked him what he was working on, and he answered:


Frankly, I'm attempting KickStart Replacement Bounty I and II.

I'm working on porting AROS to the Commodore Amiga 1200, specifically
the UAE emulation. I also have a real A1200 I would like to get
this running on, but I'll need a flash burner for that.

I have a ROM image building with stock ELF binutils 2.20 and gcc 4.5.1,
with (I think) the correct register based ABI for libcalls. Total
ROM size is currently ~430K, so I should have plenty of space for the
Amiga AGA/OCS/ECS HIDDs and whatever other device drivers I need to write.

My short term goal is to get all the 'support' patches into the mainline
first, then bring in the m68k-amiga specific code, so that I have a
working build when I check in the m68k-amiga support.

I plan to complete this goal withing the next few weeks, depending on
how badly my changes clash with the existing architectures. I have
tried to be clean with this, but accidents happen.

My medium term goal is to boot an AROS Workbench disk, and have all
the AROS/C applications work on A1200 on UAE.

I hope to complete *that* goal by December 31, 2010.

( No HUNK support, it's all ELF right now, but HUNK backwards compatibility
  should be doable in the future.)

My long term goal is to be able to boot a Amiga WorkBench 3.1 disk, and
have all the Amiga HUNK apps work. This would allow AROS-m68k-amiga to
fully operate as a KickStart replacement ROM.



Jason did not simply applied for the bounty: he worked on a proof of concept  achieving, thanks to the latest UAE developments that finally allowed amiga UNIX buiilds to run, the build of a debian image to be used for compiling AROS on 68k. From this point he begun to submit a big amount of patches for getting the AROS 68k version back on track using GCC 4.5.0.

Its plan is to first be able to compile AROS then, slowly, replace in ROM images original amiga libraries with AROS ones in order to obtain a bootable AROS rom; furthermore Jason thinks to include, together with AROS ELF file format, also Amiga Hunk format in order to obtain the binary compatibility with Amiga software.

Lately Jason is no more alone in working in the bounty, even Toni Wilen, one of the WinUAE team coders, started to work on the bounty helping Jason in setting up the rom, especially taking care of the amiga-only hardware device drivers; he announced its intention to help Jason in this letter in the developer mailing list on October 21st:


Hello, it is WinUAE guy here. I think it is time to finally post something
here now that m68k build is really happening (great job!)

Sooner or later I am going to start working with 68k rom, mainly for (game)
emulation use, including add/fix missing Amiga specific drivers (I know
Amiga custom hardware inside out and also how drivers work) Unless someone
else have already started writing boring driver code..

I can help with low level drivers like Amiga floppy hardware driver, display
(+copper/blitter), CIA+keyboard, mouse etc.. anything that is close to
hardware and needed to boot not-too-badly programmed disk based games
without need for original rom. Of course many really old games will never
work without some kind of KS1.3 "compatibility" mode but that isn't
important now, perhaps someday..

Any list(s) of missing/not yet working 68k specific device drivers?

I think there are 2 phases:

Phase 1: support programs that have custom bootblock, possibly loads few
more tracks using trackdisk.device and then the rest uses hardware banging
loader.
Phase 2: support programs that boot to CLI, uses startup-sequence.

btw, I don't care about the bounty (or about "new" Amigas), I "only" want to
finally have (win)uae with built-in legal more or less "retro-compatible"
rom replacement.


Before to go on, is better to remind: Why the Kickstart integration phase 1 and 2 are so important?

The integration bounty are important for a good serie of reasons: first of all, once the AROS 68k port will be completed, it also will include binary compatibility with the HUNK file format, and most of the existing amiga applications written according to the Amiga OS guidelines will work directly on AROS: that means is possible to use AROS in place of the original Amiga OS; unlike the original Amiga OS 3.x, AROS is actively developed and therefore there is the advantage of having new features and patches;

second, AROS is open source and free of charge, and a 68k version b inary compatible with Amiga programs will come out handy for all those alternate and homebrew projects that require Amiga OS compatibility without the need of an official Amiga ROM, names as Natami, UAE/WinUAE.J-UAE, Minimig and the FPGA Toaster-on-a-chip comes to mind;

third but not less important, the existence of an Amiga version of AROS has an emotional factor, binding to the original Amiga Replacement OS project of AAron Digulla - sort of the return home (Amiga Hardware) of the Prodigal Son (AROS) and will add even more legitimacy to the project form those people that so far were against AROS for the fact that does not run on Amiga Hardware; potentially can also add for some more developers that might decide in contribute for improve the compatibility and the performance, just as Tony Wilen did.

Today November second, Jason communicated another significant progress:

I have successfully loaded my first disk block using Frankenrom with KS 3.0!
With Toni's help, I realized that I needed to enable DMA (DUH!) early on after IRQs have been enabled, and KS 3.0 + AROS exec will now get to the point where it *sucessfully* loads the first disk block.
However, the screen is still black. Is any of this debugging useful? (I disabled alert.hook and exec/SetFunction() Just In Case, but no change in behaviour)


And today even better:

KS 3.0: Getting there! exec will now get to the boot screen! http://www.evillabs.net/wiki/index.php/AROS_m68k-amiga
I need to make a disk with *just* the Install boot block now.

and the result is the boot image seen above; it is just AROS exec and the rest is Amiga routines but is a significant step forward!

Jason recommmended Tony Wilen for the assignment of the Phase 2 of the bounty since he has more experience in the Amiga OS hardware internals and is able to code effectively drivers. The outcome of this bounty are keeping mu attention level high nd am really excited to see the progresses!

TECNOLOGIE
17 luglio 2009
Diary:Technologies:Amiga / AROS: Hacko ergo sum

This time will not talk about lua programming, also because since many of the project that i tried to work in either got covered by somebody else, (like the network control panel or the amistart calendar). Am also still thinking on how to build some tutorial to explain the base concepts on how to generate a zulu interface.

Andrea "Afxgroup" Palmete' accepted the OS4 Depot Gnash os 4 port bounty: considered that  Gnash is a project under GPL license, it might be possible, once the bounty has been satisfied, to port it in other amiga-like systems with adequate modifications (being a os4 project it uses reaction for its - thank God quite basic - GUI); so far results are pretty good: youtube movies are working properly and flash applications up to version 7 are quite supported too; a summer of code task should bring to the mainstream project the actionscript 3 classes support. Personally i still hope that somebody might build one day an open source builder for flash contents: in the past there were attemts such as f4l but now the projects looks abandoned and the project that should have replaced it, called UIRA, was actually suspended for political/legal issues; the original developers made a call for new people that would have liked to bring it on....

In the meanwhile Stanislaw Sszymczyk continued to update OWB that, so far, reached version 0.9.7 [edit: 0.9.8 is already out]: it fixes some errors that brought 0.9.6 version (like inability to submit forms in some pages such as the aros-exec forum one) and features also a basic bookmark manager. Beside that, the cookie handling has been updated in the mainstream version and implemented (fixed a bit more too) in the AROS version.

I am also glad to announce that now the AJAX editor of my blog platform is working too unde OWB, beside the minor annoyance to need to refresh the page once i saved the content.



OWB 0.9.7 on my blog edit platform - the bookmark window is in the bottom right.


Paolo Besser is working too in try to improve its Icaros Desktop distro: recently, beside the new Icaros Desktop 1.1.2 release, he published a new Automatic Upgrade tool that will detects upgrades either from an .iso file, a CD-ROM or from internet, downloading a nightly.iso and using it to upgrade the system I tried it yesterday but it makes my QEMU environment close, guess because the disk ran out of space (my system partition is small) while downloading the nightly .iso from the net.

And Chris "platon42" Hodges is working on fixing the port of Poseidon on AROS: one of the main problems that Chris is having now is that he is lacking a real x86 hardware where to test on: its actual method of testing is to develop on VmWare, upload its latest binaries on the nightly buillds and have users with real hardware report about any problems it might arise.  So far some peripherals are recognized and some can also be mounted, but in other cases we still have either system freezes and hangups or the new inserte dperipheral might not be recognized properly by the stack. Once that is done, seems there still are some problems with the fat.handler, where in example the OS command copy does not work properly from FFS/SFS to FAT yet...

Matthias "Mazze" Ruster is busy, together with the YAM developing team in porting finally the well-known Amiga mail client on AROS; the port is virtually finished, beside a necessary bug hunting for some specific system-related flaws.

Last June 6 I had the occasion and the privilege to participate at a monthly reunion of the Southern California Commodore/Amiga Network (SCCAN) in the town of Castaic, north of the Los Angeles County and San Fernardo Valley, almost at the limit of the huge LA urban area. In the garage of one of those almost-cloned suburb houses i was greeted from Joew May, one of the members and the house owner. On ce the garage door slide up i faced its "laboratory": on the various crowded shelf i seen a c64, a sx-64 (c64 executive), several amiga 500, 1200 and a 3000 with or without external hard disk, beside three or four a2000 with toaster piled in a corner. After a while other members of the group arrived, included Robert Bernardo, the founder and brought further hardware and software; i therefore seen UAE on a eee-pc 700, i was able to see a custom version of the c-one board which also supported Minimig in its FPGA programming, the sx-64 mentioned above and other relics from the 64 era, including a plastic musical keyboard overlay similar to the one distributed by SIEL in italy.


An overall view of the SCCAN : from the left, standing Jerold - a guest movie operator; then Robert Bernardo sitting with white shirt and glasses, then standing Tim, the c-one user and, sitting on the right Matt with its eee-pc running UAE. Just next to Robert look st the pile of a2000 equiped with Video Toasters:)

It was a pity that, since my wife and my mother in law were with me, i wasunabvle to stay more than a couple of hours but it was nice to see some other people with a common ground to share; beside those little reunions, the SCCAN and the Fresno Commodore Club also organize in Las Vegas the Commodore Vegas Expo (or CommVEx) on the last week-end of July; as usual for budget reasons i cannot partecipate: this was a good edition with excellent guest (among those,Dave Haynie, RJ Mical and miss Jeri "C-One" Ellisworth !!!) but at least it is my intention to go to Castaic again next end of August/begin of September.

In order to show AROS to the SCCAN members, I thought it was about time to introduce it in a much better way than it happened at SCALE, where I introduced VmWAROS on virtual machine, full screen but no sound and the network still to be fixed; that of course was unprofessional and, am afraid, even unconvincing.

The stimulus to do that also came from the IRC channel, where an user asked how Icarosperformed on Virtualbox;  nic answered that he has been unable so far to make Icaros run on virtualbox properly and remembered - and me too -  that last year Michal Schulz made some fixes for AROS to run properly on it; other reason has been due to my subscription in the FSUGItalia board, where I introduced AROS and one of the users complained about the fact that VmWare was not an "open" virtual machine and was looking for alternatives such as Virtualbox.

So, at the end I downloaded and installed Virtualbox on my own machine and proceedted to look for make Icaros work on it.
According to an old multiplatform tutorial I should have first joined together all .vmdk files with the
vmware vdiskmanager, usually buindled with VmWare server but not with the player (thanks again to Kalamatee that provided me the file via IRC saving me from an approximate hundred megabytes download) and then I should have used qemu and convert the virtual file in a middle format that, using a Virtualbox tool, should then become readable from this one; and therefore I proceeded in all the iter to the Virtualbox converter... that was not there! Hoiw come?

A further search in google clarified that latest versions of Virtualbox can read VmWare virtual drives (nice to know having spent half an hour converting the files...) and so, once deleted the converted files I do the attempt to configure Virtualbox in run Icaros. The first boot attempt ends with the red pointer on the black screen and nothing else; considered that, whan I made my tests, the ata.device still were having big problems with AMD boards (and my Dell laptop has an AMD Sempron), i decided to follow instructions on Icaros site for unsupported SATA drives and so I  changed in GRUB the ATA=32bit string in ATA=nopci (ATA=nodma works too btw) and doing so i finally boot on Wanderer.


Icaros on Virtualbox: it works, despite some problems...
 
First of all I need to make clear that, unlike VmWare and qemu that adapt a screen refresh rate to the host machine speed and virtualisation capabilities, Virtualbox will run at full video speed (else known as 1-1 screen refresh) and will run excellent if your hardware support virtualisation else,like in my single core laptop, expect to wait up to three seconds for opening a wanderer window and show its icons.

Second, might be a mistake in my configuration but the Virtualbox network bridge tends to make exclusive use of the network, therefore forget to browse from AROS and windows at the same time; ti make sure the bridge does not give me any problem, i disabled it when i dont use VB, and will keep it like that until i dont understand what to do about it, or somebody will suggest me a workaround.

Third and last, the sound "kinda" work if i set a compatible sound card (AC-97 with a supported chipser) in the virtual machine, but the sound test make a kind of attempt of a sound (sound like BEEreberebereberebeepp-p-p, instead of a two seconds Beep) and then it freezes badly the virtual machine.

So for now Virtualbox is still there, sitting in my disc; probably i will do some more experiments in the future; about me i decided to switch to plan B and try the QEMU option.

Hopeful to have a better result (also because to what an IRC member told me) i decided to install the VE version of Icaros, that also includes KQemu, and follow the instructions given by Paolone in its web site for installing a QEMU environment on a netbook here, with some interpretation, of course. And the final result was really good: double clicking on the executing .bat file shortcut, Icaros start at full screen (i have another.bat file for work in a window), the sound seems to work properly, tried with milkytracker  (this .mpeg file make freeze and quit qemu environment btw) and the network works fine thianks to DHCP.


Icaros on qemu: finally something to show for evangelisation :)


Despite that i still gave problems in transferring files: considered the (damn!) net sandboxing system of qemu, making a subnet with IP addresses like 10.2.xx that is not visible from outside, this mieans that in order to transfer files i have to FTP from inside qemu towards my machine; and that is complicated from the fact that my machine does not have a static address, being used mainly through wireless. I tried a simple FTP address on it but so far it did not work, art least i was unable to see it from inside qemu. When i have some time will install the XAMPP package (doing web sites i need it even for work) and will set filezilla server.

But those were not all the experimentations that I made so far: when the Network control panel came out had the pretty insane idea to make a personal "emergency live-CD" with all I needed to go on line, like OWB, a IRC client, YAFS and the FTP using as base a nightly; this because my old laptop has a malfunctioning DVD drive that can read only CD-ROMs. Therefore, I found a copy of Magic ISO maker and started to experiment burning images. For my perosnal disappointment i have been unable to make any progress: either the CD booted showing a mangled scren or they did not boot at all (but they were showing if i boot from another original nightly). So, seems that so far the only way to make a bootable CD is to build AROS using the toolkit from source;  knowing that the GRUB boot file for AROS is called eltorito (yuo, same as the mexican food restaurant chain) i tried to make point the bootable option to it, still without success so far. Any advice for me?

Going on with my activities, lately had occasion to work again on the translation of the Amiga Programming Guide by Gianfranco "ShInKurO" Gignina; he prepared a SVN repository on Google Code for allow other people to partecipate. Since am already susing SVN at basic level from my office in linux shell, i thought that was a piece of cake to handle. But, as usual, murphy decided to pay me a visit. To contribute I decided to download TortoiseSVN, a SVN client for windows that provide a graphical user interface integration. I already prepared the translation of the new parts of chapter 5 but, despite I did inserted the google code repository and the generated password in the settings (as https too) i have been unable to commit in the repository.My caontribution has been uploaded by ShInkUrO itself, which I sent it via email; i really hope to fix this soon.

And, to finish for this time, let me provide an interesting link that I found on the net: the "how to survive poisonous people" talk made by the SVN developer's team and hosted on Google video: it talks about problems that might arise when some community members are, to say it soft, "less cooperative"; thank God this happens less in AROS environment, but is still a good source of advice.



sfoglia
  

Rubriche
Link
Cerca

Feed

Feed RSS di questo 

blog Reader
Feed ATOM di questo 

blog Atom
Resta aggiornato con i feed.

Curiosità
blog letto 1 volte

Older Posts

Who links to me?
Get in touch with Simone Bernacchia

 




IL CANNOCCHIALE