[ from the oh-no-not-again dept. ]
We are watching the Discovery Channel. Captain Cyborg is on spouting about how his pet chip impant did blah blah blah blah. Someone get this guy off the air talking about this yet again! If we have to see that video of him controlling a robotic arm over the internet one more time…
Jon.
[ from the arrgh-they-will-get-you-arrrrrrggghhhh! dept. ]
Hmmmm…copy configuration in to Upper and Lower BAT registers, play with SDR1, page tables, blah blah blah blah. What do you get? Well about 3K lines of code which doesn’t do what I want. I really really need a hardware debugging kit of some kind or a lot more time – that’s not because I started writing code later but because the task is complex. I will get enough resolved for the deadline though there is some serious summer work ahead.
The report should have 5-7K words sorted by the end of the weekend and enough for a draft to be ready. Then I have a week to get it wrapped up.
There so now you all know you can stop asking me how it is going. 
Jon.
[ from the what-are-they-taking? dept. ]
So this Unisys guy gives yet another boring seminar on “Web Security”. Covers the basics with little extra adventure and uses the wrong terms, as usual, for the benefit of the Computer Scientists listening. All based upon Microsoft, IIS, and other horrible shit. Quite sickening.
Now this was not as bothering as the session afterwards where there was some live demonstration. He proceeds to login as “Administrator” on a “hardened laptop” (it had Zone Alarm software installed) and spout fantastical shite about Microsoft security. I offered to secure an Apache system in 5 minutes (and I meant it) which would compete with his IIS setup…apparently it took six months to define templates to prevent access to the filesystem. In any case when I asked why he could not simply use a chroot jail my question was brushed over. Microsoft Windows does not do that kind of thing properly, you see…
Either this guy had never heard of capability sets on Windows and sudo on UNIX or he did a great job of ignoring it. Extolling the virtues of ACLs in response to my questions was interesting – given that most modern UNIX systems also support ACLs and Extended ACLs, and even permissions affecting other system objects – Linux has Trusted Path Execution, NSA Linux does funky shit and then there’s always Trusted Solaris.
I did point out that his repeated mentioning of buffer overflows was a moot point on systems where adequate protections are in place – through stack guarding software combined with non-Intel stack frame layouts – and that he was ignoring the potential for vulnerabilities in third party software and libraries which he had no access to.
Apparently “nobody buys Linux stuff” or Solaris, or infact any non-Microsoft products – at least that was the view of one Unisys drone.
Vomit inducing.
Jon.
[ from the this-will-be-fun dept. ]
This week we are graced with another guest lecture from Unisys, a fantastic opportunity to once again ask about corporate sponsorship of Software Patents such as that used by the LZW compression algorithm. I enjoy these talks…
Jon.
[ from the RMS-really-does-rule dept. ]
Hi,
Revolution OS has arrived. We watched it over the weekend and there are plans to arrange a screening.
The Memory Managemend issues are still ongoing however I have now written the equilavent of an LED attached to the serial port and have some position independent hard coded assembly to flash the framebuffer so that I know where things are going stray.
Jon.
[ from the giant-pandas-are-everywhere dept. ]
Well the project code is a good few thousand lines now and can talk to the prom, framebuffer, etc. though the memory management code is starting to get really bothering – at the moment everything I try to do seems to be failing, and it is very difficult to determine where when you have no hardware debugger. I am writing a minimal pixel painting assembly routing for emergency display blanking to determine whether the code is still running at certain points or if it has gone for a nice little trip through the memory to execute whatever it finds there again…
The memory concept at the moment is to have a single global free list which is initialised to contain a double(y) linked list of structs describing pages of memory – effectively all memory is one giant array of page frames with mimimal cleaverness implemented at this stage.
Jon.
[ from the got2-find-the-answer dept. ]
The problem that was happening with figuring out how to generate a got2 section for relocating data between kernel link address relocation has been resolved by using the gcc CFLAGS entry“-mrelocatable-lib”. Perhaps there is some documentation somewhere though it remains quite elusive.
Jon.