This is an archive of the discontinued LLVM Phabricator instance.

Handle ARMv6KZ naming
ClosedPublic

Authored by tyomitch on Nov 11 2015, 4:08 AM.

Diff Detail

Event Timeline

tyomitch updated this revision to Diff 39901.Nov 11 2015, 4:08 AM
tyomitch retitled this revision from to Handle ARMv6KZ naming.
tyomitch updated this object.
tyomitch added reviewers: rengolin, joerg, bogden.
tyomitch added a subscriber: cfe-commits.

Why isn't this just part of D14568?

It's in the other repository.

From: James Molloy [mailto:james@jamesmolloy.co.uk]
Sent: 11 November 2015 12:42
To: reviews+D14570+public+1de1f7f06e3d3585@reviews.llvm.org; Artyom Skrobov; renato.golin@linaro.org; joerg@netbsd.org; Bernard Ogden
Cc: kanheim@a-bix.com; cfe-commits@lists.llvm.org
Subject: Re: [PATCH] D14570: Handle ARMv6KZ naming

Hi Artyom,

Why isn't this just part of D14568?

Cheers,

James

joerg edited edge metadata.Nov 12 2015, 6:02 AM

The specific CPU was only chosen to match what the backend supported at the time. As long as the feature set is identical, it doesn't really matter.

The specific CPU was only chosen to match what the backend supported at the time. As long as the feature set is identical, it doesn't really matter.

The feature set isn't quite identical; default ARMv6 CPU for non-NetBSD targets is "arm1136jf-s", which differs from "arm1176jzf-s" by not having the HasV6KOps and FeatureTrustZone backend features.

I don't understand why NetBSD is special in this respect; but anyway, my patch preserves its special handling.

compnerd accepted this revision.Nov 12 2015, 8:20 AM
compnerd added a reviewer: compnerd.
compnerd added a subscriber: compnerd.

Wow, this is tricky: the code change is in LLVM, and test change in clang :(. However, this does seem to preserve the features.

This is a dependent change so be careful about commit ordering.

This revision is now accepted and ready to land.Nov 12 2015, 8:20 AM
rengolin edited edge metadata.Nov 16 2015, 4:19 AM

Wow, this is tricky: the code change is in LLVM, and test change in clang :(. However, this does seem to preserve the features.

The problem is that Clang is the most important user of this library, which also has other users in LLVM. We just need to make sure that Clang is doing the right thing, and that the intentions are recorded on the right places. In this case, it's Clang. :)

Closed by commit rL253207: Handle ARMv6KZ naming (authored by askrobov). · Explain WhyNov 16 2015, 6:08 AM
This revision was automatically updated to reflect the committed changes.