Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Scientists find security risk in RISC-V open-source chip architecture (scmp.com)
10 points by msolujic on June 6, 2024 | hide | past | favorite | 9 comments


It's astonishing how few technical details are provided in this article (like the nature of the vulnerability, whether it affects RISC-V's design in general or only a specific implementation, and whether it is readily fixable).

I tried to find more details and found that slashdot user st0nerhat had found a more technical summary in Chinese, which you can see translated here:

https://mp-weixin-qq-com.translate.goog/s/ke8tBpJ7NpvUEAecov...


dupe: https://news.ycombinator.com/item?id=40591280

This is a bug in a student-built microarchitecture, not an issue with the ISA.

Not a great article, as it attacks RISC-V for no good reason.


They had a pun, they wanted to use it.


Yeah looks like bugs in riscv-boom rather than the RISC-V ISA itself


In particular this is a timing attack like Spectre/Meltdown - looks like there are timing issues with allocation of write port slots when writing to the register file - because division takes so long it's possible to use a late write from division to tell whether a particular speculative branch has been taken or not



The article has little useful data - this looks like it's the original paper:

https://mp.weixin.qq.com/s/ke8tBpJ7NpvUEAecov--UQ


There's a CVE table in that article and they seem to have CVEs from a microarchitecture implementing openRISC (entirely different ISA) mixed in there somehow.


This describes a timing attack like Spectre/Meltdown on a particular O/S architecture, not all RISC-V cores - looks like there are timing issues with allocation of write port slots when writing to the register file - because division takes so long it's possible to use a late write from division to tell whether a particular speculative branch has been taken or not




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

Search: