[PATCH] D63936: [clang][Driver][ARM] Favor -mfpu over default CPU features

Alexandros Lamprineas via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Wed Jul 3 05:58:03 PDT 2019


labrinea added inline comments.


================
Comment at: llvm/lib/Support/ARMTargetParser.cpp:412
 
-  if (Extensions & AEK_CRC)
-    Features.push_back("+crc");
-  else
-    Features.push_back("-crc");
-
-  if (Extensions & AEK_DSP)
-    Features.push_back("+dsp");
-  else
-    Features.push_back("-dsp");
-
-  if (Extensions & AEK_FP16FML)
-    Features.push_back("+fp16fml");
-  else
-    Features.push_back("-fp16fml");
-
-  if (Extensions & AEK_RAS)
-    Features.push_back("+ras");
-  else
-    Features.push_back("-ras");
-
-  if (Extensions & AEK_DOTPROD)
-    Features.push_back("+dotprod");
-  else
-    Features.push_back("-dotprod");
+  for (const auto AE : ARCHExtNames) {
+    if ((Extensions & AE.ID) == AE.ID && AE.Feature)
----------------
ostannard wrote:
> labrinea wrote:
> > SjoerdMeijer wrote:
> > > This could be a little local helper function, share the code, as exactly the same is done in `ARM::appendArchExtFeatures`
> > We are not doing exactly the same thing in these functions. Here we extract features out of a bitmap, which is a map containing a bitwise OR of separate feature bitmasks. If a bitmask that corresponds to a known feature is present - and here I mean all the bits of that mask are present - then we push the feature, otherwise not. 
> > 
> > In `ARM::appendArchExtFeatures` we compare a given bitmask, which corresponds to a specific feature, against all the known bitmasks (individual features) one by one. In contrast to `ARM::getExtensionFeatures` here we also handle negative features, and that means the prohibition of a feature can disable other features that depend on it as well.
> Does this correctly handle the combination of integer-only MVE with a scalar FPU? I think -march=...+mve+fp should enable AEK_FP and AEK_SIMD (+mve), but shouldn't enable +mve.fp.
The target features explicitly specified on the command line are handled by `ARM::appendArchExtFeatures` (see [[ https://reviews.llvm.org/D64048 | D64048 ]]). That said, yes, -march=...+mve+fp does enable scalar floating point and integer-only mve, but doesn't enable mve with floating point. I could possibly add such a test on that revision.

On the other hand, `ARM::getExtensionFeatures` cannot tell the difference between the two configurations you describe, and it's not possible to do so, because they are represented on a bitmap returned from `ARM::getDefaultExtensions`, which reads the table. That said, if there was an entry in the table containing `AEK_FP | AEK_SIMD` that would enable mve.fp. However, I don't think this is a problem in practice. My understanding is that the table indicates FPU support with `FK_*`, and Extension support with `AEK_*`.  That said, I believe AEK_FP is only used for command line option handling.

Maybe a fix for this problem would be to replace `AEK_FP | AEK_SIMD` with something like `AEK_MVE_FP` but then we wouldn't be able to do what is proposed in [[ https://reviews.llvm.org/D64048 | D64048 ]].


================
Comment at: llvm/unittests/Support/TargetParserTest.cpp:574
 
-  Extensions[ARM::AEK_CRC]        = { "+crc",       "-crc" };
-  Extensions[ARM::AEK_DSP]        = { "+dsp",       "-dsp" };
+  for (auto &Ext : ARM::ARCHExtNames) {
+    if (Ext.Feature && Ext.NegFeature)
----------------
ostannard wrote:
> SjoerdMeijer wrote:
> > I like this  approach.
> I'm not sure this is a good idea - we are now testing the implementation by checking that it matches the table used by the implementation, so if there's a bug in the table this will still pass.
Surely, but is the purpose of this test to check that the table is correct, or that `ARM::getExtensionFeatures` reads the table correctly? I'd say the latter. Also, with this change we won't need to update the test every time there's a new entry in the table.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D63936/new/

https://reviews.llvm.org/D63936





More information about the llvm-commits mailing list