[PATCH] D57763: [ARM] Add Cortex-M35P

[Oliver Stannard via Phabricator via llvm-commits]

olista01 requested changes to this revision.
olista01 added inline comments.
This revision now requires changes to proceed.


================
Comment at: lib/Target/ARM/ARM.td:756
 
-
 //===----------------------------------------------------------------------===//
----------------
Whitespace change.


================
Comment at: lib/Target/ARM/ARM.td:1018
+                                                         FeatureVFPOnlySP,
+                                                         FeatureUseMISched,
+                                                         FeatureHasNoBranchPredictor]>;
----------------
Why are you enabling FeatureUseMISched? This is using the Cortex-M3 sched model, and Cortex-M3 doesn't have it.


================
Comment at: test/CodeGen/ARM/build-attributes.ll:111
+; RUN: llc < %s -mtriple=thumbv8-linux-gnueabi -mcpu=cortex-m35p | FileCheck %s --check-prefix=CORTEX-M35P
+; RUN: llc < %s -mtriple=thumbv8-linux-gnueabi -mcpu=cortex-m35p -enable-unsafe-fp-math -frame-pointer=none -enable-no-infs-fp-math -enable-no-nans-fp-math -fp-contract=fast | FileCheck %s --check-prefix=CORTEX-M33-FAST
+; RUN: llc < %s -mtriple=thumbv8-linux-gnueabi -mcpu=cortex-m35p -enable-sign-dependent-rounding-fp-math | FileCheck %s --check-prefix=DYN-ROUNDING
----------------
It's confusing to have this share the CORTEX-M33-FAST check prefix, could you rename it to something more generic? (or remove it, I don't think these build attributes are related to the core).


================
Comment at: test/CodeGen/ARM/build-attributes.ll:1413
+; CORTEX-M35P:  .eabi_attribute 17, 1
+;;SDCOMP-45836: We default to PreserveFPSign
+; CORTEX-M35P:  .eabi_attribute 20, 1
----------------
This is an Arm-internal ticket number, this should be fixed to check standard LLVM behaviour.


Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D57763/new/

https://reviews.llvm.org/D57763