On Tue, 2018-01-09 at 16:05 +0000, David Woodhouse wrote:
>
>
> Razvan points out that the #ifdef there is redundant; in older kernels,
> bugs.c is only built on 32-bit anyway.
>
> We're working on backporting the other CPU_BUG_* and sysfs
> vulnerabilities bits to 4.9 (first), and will probably end up
> cherry-picking 62a67e123e ("x86/cpu: Merge bugs.c and bugs_64.c").
s/will probably end up/ended up/
:/sys/devices/system/cpu/vulnerabilities# cat spectre_v2
Mitigation: Full AMD retpoline
:/sys/devices/system/cpu/vulnerabilities# uname -a
Linux localhost 4.9.76-rc1+ #2 SMP Tue Jan 9 17:37:55 GMT 2018 x86_64 x86_64 x86_64 GNU/Linux
http://git.infradead.org/users/rga/retpoline-stable.git/shortlog/refs/heads/linux-4.9.y
passes basic smoke testing so far; will audit it and check for
additional indirect branches which didn't exist in 4.14, tomorrow. And
continue to 4.4 and 3.2 kernels too.
We will also backport the IBRS patches on top, once they're ready.
What we *haven't* included is PTI-specific changes, like (to pick a
random example) the tboot one.Attachment:
smime.p7s
Description: S/MIME cryptographic signature
