News for hackers. Stuff that matters.
Re:Teardown of a Failed Linux LTS Spectre Fix by atheowaway4z on Wednesday September 04, 2019 @07:59AM (#20876252)
Does anybody know what triggered the original fix in the first place? Just grep-ing?

If the kernel folks did what this company did . I.e have the compiler detect vulnerable code . The issue would have been avoided as well right ?

Re:Teardown of a Failed Linux LTS Spectre Fix by temac on Wednesday September 04, 2019 @08:09AM (#20876323)
I think there are some static analysis tool. At first it was some proprietary tools ran by who knows who, but this was obviously an enormous problem in the context of Linux dev, so IIRC somebody wrote something (that might be quite primitive but that is better than nothing) to detect potential cases. Anyway, a compiler plugin is a neat idea. It won't detect everything but at least it can detect (and fix!) local cases. Not sure if the open source analyzers (if any) can detect less local cases, though -- but then with this pattern most cases should be quite local, so I might actually be the best to have something simple.
Re:Teardown of a Failed Linux LTS Spectre Fix by wyldfire on Wednesday September 04, 2019 @09:58AM (#20877239)
The article describes grsecurity's 'respectre' compiler plugin [1].

[1] https://grsecurity.net/respectre_announce.php

Re:Teardown of a Failed Linux LTS Spectre Fix by mannykannot on Wednesday September 04, 2019 @11:20AM (#20877991)
While they say that Respectre would have caught it, the actual discovery was apparently made this way:

"We independently backported the fix on July 9th 2019 and on noticing the warning, fixed it correctly. When the upstream kernels later backported their bad fix, it created a conflict in our git repo that led to us immediately spotting their flaw (and keeping our existing fixes)."