[edk2-devel] [PATCH v8 17/46] OvmfPkg/VmgExitLib: Add support for NPF NAE events (MMIO)

[Laszlo Ersek]

On 05/22/20 16:14, Laszlo Ersek wrote:

>   INT64 Displacement;

>   Displacement *= (1 << Ext->Sib.Scale);

> (10c) Multiplying a negative INT64 by 1, 2, 4, or 8 is well-defined
> (assuming again that the initial Displacement value is small enough,
> which depends on the original instruction).
> 
> If we wanted to be super-safe, we could replace this open-coded
> INT64 multiplication with a call to SafeInt64Mult(), from
> <Library/SafeIntLib.h>, and hang here, if the call fails.
> 
> Up to you.

Side comment: normally, even if we were 100% sure the result would be in
range, we shouldn't use the "*" operator for INT64 multiplication --
because of intrinsics.

BaseLib offers MultS64x64() for that problem.

But, this is X64-only code. So using "*" for INT64 multiplication should
be fine on all toolchains, regarded purely from an intrinsics perspective.

So it's a choice between "*" (if we trust the instruction being decoded
to be sane) or SafeInt64Mult() (if we don't).

MultS64x64() sits in the middle, and doesn't buy us anything here.

Thanks
Laszlo


-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.

View/Reply Online (#60146): https://edk2.groups.io/g/devel/message/60146
Mute This Topic: https://groups.io/mt/74336571/1813853
Group Owner: devel+owner at edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub  [edk2-devel-archive at redhat.com]
-=-=-=-=-=-=-=-=-=-=-=-