In D136425#3874313, @SjoerdMeijer wrote:

In D136425#3874298, @dmgreen wrote:

I am not against this, but as @ktkachov said it sets a new precedence. We have not in the past added CPU names which are aliases.

If we are doing this, can we build a better way of adding aliases? Maybe a list in TargetParser.def that gets queried at the relevant places?

There are a few unknowns here I think (at least for me):

• Are we going to have (a lot) more aliases? I don't know, but I would guess not.
• Do we later want to distinguish V2 from Grace (for one reason or another)? I don't know yet, but maybe.

I am happy to investigate how we could introduce aliases, but then again, not sure how useful that would be given what I mentioned above; that's the reason I went for this approach, as a one-off, for now.

But let me know what you think, and am happy to look.

Doesn't have to be part of this patch, but can this be extended to FADDP as well?

I think the most important use of the __ARM_FEATURE_MOPS is to check whether it's safe to use the new intrinsics, so it should only be defined by the compiler once it supports the intrinsics proposed in https://github.com/ARM-software/acle/pull/38/files

In D75181#2193447, @danielkiss wrote:

Would it be better to add a new value to "sign-return-address" as "none"? I don't see any other alternative option, I'm open to any other idea.

In D69247#1719597, @peter.smith wrote:

I'm on vacation at the moment, will take a look next week as this will need some checking of the architecture manual. This LKML message https://lkml.org/lkml/2018/10/9/698 implies that in some cases it is insufficient to just read CTR_EL0.{IDC,DIC}, as a 0 also requires a check with CTR_EL1 which can't be done in userspace. Would you be able to take a look at that thread? For compiler-rt we need something that works on all CPUs at EL0.

In D64495#1587266, @sammccall wrote:

FYI, this change broke git-llvm for everyone with a different username :-)
Fixed in r366198

Fix comments plus an offline comment I had (copy-pasto in an assert message)

Add more CHECK-LABEL tests, test v8.4a and v8.5a features. Fix formatting in pattern.