Added far return instructions (that is, returns to

[oota-llvm.git] / lib / Target / README.txt

diff --git a/lib/Target/README.txt b/lib/Target/README.txt
index be7bb3a53ecce451eaf65880b3fa5065d23f7b78..9dd2b365c03ee8fd18e07add7865ab4735c250c6 100644 (file)
--- a/lib/Target/README.txt
+++ b/lib/Target/README.txt
@@ -197,13 +197,6 @@ _bar:   addic r3,r3,-1
 
 //===---------------------------------------------------------------------===//
 
-Legalize should lower ctlz like this:
-  ctlz(x) = popcnt((x-1) & ~x)
-
-on targets that have popcnt but not ctlz.  itanium, what else?
-
-//===---------------------------------------------------------------------===//
-
 quantum_sigma_x in 462.libquantum contains the following loop:
 
       for(i=0; i<reg->size; i++)
@@ -335,11 +328,6 @@ when it is declared U32.
 
 //===---------------------------------------------------------------------===//
 
-Promote for i32 bswap can use i64 bswap + shr.  Useful on targets with 64-bit
-regs and bswap, like itanium.
-
-//===---------------------------------------------------------------------===//
-
 LSR should know what GPR types a target has.  This code:
 
 volatile short X, Y; // globals
@@ -1123,16 +1111,6 @@ optimized with "clang -emit-llvm-bc | opt -std-compile-opts".
 
 //===---------------------------------------------------------------------===//
 
-We would like to do the following transform in the instcombiner:
-
-  -X/C -> X/-C
-
-However, this isn't valid if (-X) overflows. We can implement this when we
-have the concept of a "C signed subtraction" operator that which is undefined
-on overflow.
-
-//===---------------------------------------------------------------------===//
-
 This was noticed in the entryblock for grokdeclarator in 403.gcc:
 
         %tmp = icmp eq i32 %decl_context, 4          
@@ -1675,6 +1653,7 @@ entry:
 
 Instcombine should be able to optimize away the loads (and thus the globals).
 
+See also PR4973
 
 //===---------------------------------------------------------------------===//