GHC unnecessarily sign/zero-extends C call arguments

Ticket created following discussion in https://ghc.haskell.org/trac/ghc/ticket/7684#comment:58

It seems the zero/sign-extension that we're doing in the NCG for arguments less than a word in size is unnecessary, since the ABI doesn't specify any extension. There's some clarification of the spec in this thread: http://gcc.gnu.org/ml/gcc/2013-01/msg00448.html

Trac metadata

Trac field	Value
Version	7.6.3
Type	Bug
TypeOfFailure	OtherFailure
Priority	normal
Resolution	Unresolved
Component	Compiler (NCG)
Test case
Differential revisions
BlockedBy
Related
Blocking
CC	simonmar, tibbe
Operating system
Architecture

mentioned in issue #7684 (closed)

changed milestone to %7.10.1

changed weight to 5

added NCG backend Tbug Trac import labels

Note this exchange in the linked email thread above. I'm not sure if Richard (the editor) is saying that zero extending in the caller is an optimization or not extending in the caller is an optimization (which makes more sense to me):

How? You aren't allowed to access the bits outside the specified argument type (which must match on caller and callee side), so you can't observe them, so it's not required to specify their content.

OK, thanks. It's clear now.

The problem is that LLVM assumes that values are extended at a call. GCC does that, but libffi doesn't. So, calls via libffi to LLVM don't work correctly.

It's an optimization to do so to avoid partial register stalls.

My reading of it was that extending in the caller is not strictly necessary but is done as an optimisation to avoid partial register stalls. Maybe that means we ought to be doing it too - that's not clear to me. I'm sure it's not *always* an optimisation.

removed milestone

Moving to 7.12.1 milestone; if you feel this is an error and should be addressed sooner, please move it back to the 7.10.1 milestone.

changed milestone to %8.0.1

Milestone renamed

added Pnormal label