Hacker News new | past | comments | ask | show | jobs | submit login
Clever Hack Finds Mystery CPU Instructions (eejournal.com)
144 points by weare138 on April 24, 2021 | hide | past | favorite | 15 comments



On x86.

The linked article loaded very, very slow. Original (?) bloppost: https://blog.can.ac/2021/03/22/speculating-x86-64-isa-with-o...


Previously for that link (208 points / 65 comments):

https://news.ycombinator.com/item?id=26584118


Thank you. Said blog post loads very, very fast!


This reminds me of sandsifter, the x86 processor fuzzer:

https://github.com/xoreaxeaxeax/sandsifter


Archive.today cached version: https://archive.ph/wi3Hp


thanks


While I certainly have no need to find undocumented x86 instructions, I do find this style of explanation great learning a bit about how processors work (something I would like to learn more about).


I know very little about this field, but I guess I’m surprised that the table of instructions can’t be found by looking at a microcode update file.


Aren't those encrypted?


Yes, but also: the update file doesn’t contain everything, but just what needs changing. So if these undocumented ones aren’t in the file, that doesn’t mean they don’t exist.


Isn't having undocumented opcodes a huge red flag for security?


HN'dotted?


I like the phrase "hugged to death" as a more generic version of "slashdotted" personally. It's a shame but /. is no longer the web powerhouse that it used to be, though I still read it regularly just because it will sometimes have stories and things I don't otherwise find in places like reddit or hn.


agreed!


Contrary to popular belief, HN traffic is not as huge as people think for the hug of death. Its the reddit traffic, when people either repost or post at both places. HN can barely send 5 users per second, which can surely be handled by a potato.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: