07:57:58 <nssy> lol
08:54:39 <sech1> Inge- "wrmsr": ["0xc0011020:0x0", "0xc0011021:0x60", "0xc0011022:0x510000", "0xc001102b:0x1808cc16"],
08:54:54 <sech1> For buggy 1st-gen Ryzens, in config.json for XMRig 5.4.0
09:26:45 <sech1> "Exception thrown at 0x000001828D7613D2 in xmrig.exe: 0xC0000005: Access violation reading location 0xFFFFFFFFFFFFFFFF" and it points to "subpd xmm1,xmm9"
09:26:56 <sech1> These buggy Ryzens are fun
09:27:50 <sech1> the nearest instruction reading from memory is like 30 instructions before it
09:48:27 <tevador> is this the opcache bug?
09:48:38 <tevador> it's probably executing some nonsense
13:05:32 <sech1> tevador yes, opcache bug
13:19:30 <sech1> Interesting. It doesn't crash if I remove CBRANCH instructions from generated code.
13:20:43 <sech1> and crashes within a few seconds if I have 25 CBRANCH per program.
13:40:32 <sech1> So it's not about self-modifying code, but about switching between L1 code cache and Opcache after the jump.
13:41:05 <sech1> just commenting out "jz" instructions in "test/jz" sequence fixes crashes
21:50:06 <tevador> https://www.reddit.com/r/Monero/comments/eekuue/monero_mykings_botnet_has_minted_3m_worth_of_xmr/
21:50:11 <tevador> another completely trash article