@FundMECFSResearch@Tuuktuuk I'm reminded of this quote from Wasp, by Eric Frank Russell: "For months now, we have been making triumphant retreats before a demoralised enemy who is advancing in utter disorder."
@riplin This one was on ARM3, and the RiscOS debugger wrote a trap instruction to the breakpoint, so when the self-modifying code modified it, it mangled the breakpoint, and the debugger failed.
Fortunately, the self-modifying sequence was short enough to run by hand, and you could grab enough state out of memory to work out what the instruction should have been.
The cache thing is another level of awkward, though.
@harrysintonen@infosec.exchange The one thing I ever cracked (30+ years ago) involved self-modifying/decrypting code -- you'd set a breakpoint, and it would crash because it modified that instruction before it reached the breakpoint. Work it through by hand from there, and you'd get the right instruction, which got you to the next section of self-modifying code.
Six times.
I eventually got through it, and had a version that I could run from the hard disk. That was all I wanted.
@yuman @zdhzm2pgp I was a mobile phone refusenik for a long while, and I still barely use the thing, for anything. But apparently that was just me.
The scale of the rejection of AI right now is a whole different thing, though.
@gerikson @BlueMonday1984 The only thing that could have made that article better is if he'd literally ended it mid-sentence.
@FundMECFSResearch @Tuuktuuk I'm reminded of this quote from Wasp, by Eric Frank Russell: "For months now, we have been making triumphant retreats before a demoralised enemy who is advancing in utter disorder."
@riplin This one was on ARM3, and the RiscOS debugger wrote a trap instruction to the breakpoint, so when the self-modifying code modified it, it mangled the breakpoint, and the debugger failed.
Fortunately, the self-modifying sequence was short enough to run by hand, and you could grab enough state out of memory to work out what the instruction should have been.
The cache thing is another level of awkward, though.
@harrysintonen@infosec.exchange The one thing I ever cracked (30+ years ago) involved self-modifying/decrypting code -- you'd set a breakpoint, and it would crash because it modified that instruction before it reached the breakpoint. Work it through by hand from there, and you'd get the right instruction, which got you to the next section of self-modifying code.
Six times.
I eventually got through it, and had a version that I could run from the hard disk. That was all I wanted.
Fun times. :)
@Vivaldi@vivaldi.net Desktop... icons? I don't understand. :ablobcatwink: