[LLVMdev] difficulting matching i64 subtract immediate due to isel normalization of sub -> add

[David Goodwin]

Hi,

For some 64-bit immediates, ARM can generate a two instruction  
sequence. For example, for the following code:

; 734439407618 = 0x000000ab00000002
define i64 @f1(i64 %a) {
     %tmp = sub i64 %a, 734439407618
     ret i64 %tmp
}

We should be able to generate:

	subs r0, r0, #1
	sbc r1, r1, #171   @ 0xab


But instead we get:

$ llvm-as < %s | llc -march=thumb -mattr=+thumb2

	mvn r2, #1
	mvn r3, #171
	adds r0, r0, r2
	adc r1, r1, r3

(mvn is "move not").

We have code selector patterns for sube and subc that do the right  
thing, but they are not used because isel is converting the "sub i64  
a, i" into "add i64 a, -i". This leads to the attached DAG that is  
impossible to write a pattern for because the addc is producing both a  
carry and a register output, and the adde is consuming the carry and  
producing another output.

I suppose I can write code to match this... but is there a better way?

David

-------------- next part --------------
A non-text attachment was scrubbed...
Name: dag.tiff
Type: image/tiff
Size: 54108 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20090629/8cfc3d53/attachment.tiff>
-------------- next part --------------