[PATCH] D13762: [Polly] Ensure unique implicit reads/writes at beginning/end of ScopStmts

Michael Kruse via llvm-commits llvm-commits at lists.llvm.org
Thu Nov 26 15:51:28 PST 2015


Meinersbur added inline comments.

================
Comment at: test/Isl/CodeGen/uninitialized_scalar_memory.ll:8
@@ -7,3 +7,2 @@
 ; CHECK-NEXT:   store float %ebig.0, float* %ebig.0.s2a
-; CHECK-NEXT:   store i32 %iebig.0, i32* %iebig.0.s2a
 ; CHECK-NEXT:   br label %polly.stmt.if.end.entry
----------------
Reminder to myself: Have to check why this is gone

================
Comment at: test/ScopInfo/read-only-scalar-used-in-phi.ll:22
@@ -21,1 +21,3 @@
+; CHECK:       MustWriteAccess :=  [Reduction Type: NONE] [Scalar: 1]
+; CHECK:           { Stmt_next[] -> MemRef_phisummerge[] };
 ; CHECK: Stmt_bb1
----------------
This is a EXIT_PHI access

================
Comment at: test/ScopInfo/same-base-address-scalar-and-array.ll:7
@@ -6,3 +6,2 @@
 ; CHECK:         Arrays {
-; CHECK-NEXT:        float* MemRef_out; // Element size 8
 ; CHECK-NEXT:        float* MemRef_out_addr_0_lcssa; // Element size 8
----------------
The base address "out" is synthesizable (from a scop parameter) and therefore does not cause accesses. In current trunk there is still an access generated if ModelReadOnlyScalars==true and it is used by a PHI node. I am assuming this is not intentional.

There are multiple instances of this phenomenon in the tests.


http://reviews.llvm.org/D13762





More information about the llvm-commits mailing list