On 04/27/2015 06:04 PM, Brian Gerst wrote: > On Mon, Apr 27, 2015 at 11:56 AM, Andy Lutomirski <l...@amacapital.net> wrote: >> On Mon, Apr 27, 2015 at 8:46 AM, Borislav Petkov <b...@alien8.de> wrote: >>> On Mon, Apr 27, 2015 at 07:57:36AM -0700, Linus Torvalds wrote: >>>> On Mon, Apr 27, 2015 at 4:35 AM, Borislav Petkov <b...@alien8.de> wrote: >>>>> >>>>> /* >>>>> * Change top 16 bits to be the sign-extension of 47th bit, if >>>>> this >>>>> * changed %rcx, it was not canonical. >>>>> */ >>>>> ALTERNATIVE "", \ >>>>> "shl $(64 - (47+1)), %rcx; \ >>>>> sar $(64 - (47+1)), %rcx; \ >>>>> cmpq %rcx, %r11; \ >>>>> jne opportunistic_sysret_failed", >>>>> X86_BUG_SYSRET_CANON_RCX >>>> >>>> Guys, if we're looking at cycles for this, then don't do the "exact >>>> canonical test". and go back to just doing >>>> >>>> shr $__VIRTUAL_MASK_SHIFT, %rcx >>>> jnz opportunistic_sysret_failed >>>> >>>> which is much smaller. >>> >>> Right, what about the false positives: >>> >>> 17be0aec74fb ("x86/asm/entry/64: Implement better check for canonical >>> addresses") >>> >>> ? We don't care? >> >> The false positives only matter for very strange workloads, e.g. >> vsyscall=native with old libc. If it's a measurable regression, we >> could revert it. >> >> --Andy > > Another alternative is to do the canonical check in the paths that can > set user RIP with an untrusted value, ie, sigreturn and exec.
It is already done only on that path. Fast path doesn't check RCX for canonicalness. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/