[llvm-commits] [llvm] r110382 - in /llvm/trunk: lib/Analysis/BasicAliasAnalysis.cpp lib/Analysis/MemoryDependenceAnalysis.cpp test/Analysis/BasicAA/getmodrefinfo-cs-cs.ll

[Dan Gohman]

On Aug 30, 2010, at 2:32 PM, Chris Lattner wrote:

> 
> On Aug 5, 2010, at 3:09 PM, Dan Gohman wrote:
> 
>> Author: djg
>> Date: Thu Aug  5 17:09:15 2010
>> New Revision: 110382
>> 
>> URL: http://llvm.org/viewvc/llvm-project?rev=110382&view=rev
>> Log:
>> Fix memdep's code for reasoning about dependences between two calls. A Ref
>> response from getModRefInfo is not useful here. Instead, check for identical
>> calls only in the NoModRef case.
>> 
>> Reapply r110270, and strengthen it to compensate for the memdep changes.
>> When both calls are readonly, there is no dependence between them.
> 
> Hi Dan, if you're changing the semantics of the mod/ref queries here, please update this comment in AliasAnalysis.h:
> 
>  /// getModRefInfo - Return information about whether two call sites may refer
>  /// to the same set of memory locations.  This function returns NoModRef if
>  /// the two calls refer to disjoint memory locations, Ref if CS1 reads memory
>  /// written by CS2, Mod if CS1 writes to memory read or written by CS2, or
>  /// ModRef if CS1 might read or write memory accessed by CS2.
>  ///
>  virtual ModRefResult getModRefInfo(ImmutableCallSite CS1,
>                                     ImmutableCallSite CS2);
> 
> also here:
> http://llvm.org/docs/AliasAnalysis.html#ModRefInfo

My intent was to make the code follow what appeared to be the intent
of the documentation, not to change the intent. I've committed r112557
to clarify the documentation.

> Note that the call/call modref query should probably return its own (richer) enum than ModRefResult if you're interested in this going forward.

Yep. It's not an immediate priority for me though.

Dan