GitLab

All users construct their Keeps in one go.

In particular create Keeps for mode_T nodes in one go instead of adding one Proj to the Keep after another.

The missing keeps walker does this for us.

new_ir_node() already does, now init_node_attr() does, too.

It was only used in one place, where it is easily replaced.

be_Return is no more.

There is no legitimate reason for an outside user to change the op of a node.
Even internally it should be used less, i.e. only for Deleted, Id and Tuple.

The nodes in question already have this op.
This is some legacy when there were Conv_I2I and Conv_I2I8Bit.

amd64, ia32: Encode the length of the constraint list in the type instead of relying on matching magic numbers.

Remove the unused OVERFLOW_SATURE case and switch the type to a simple
wrap_on_overflow bool.

This first version should perform similar to the support in the ia32
backend, but needs more work to add amd64 specific immediate types.

This replaces ia32_ProduceVal/arm_EmptyReg. You should avoid using this
node in general, but there are some cases where you need it to fixup
nodes that are not register pressure faithful.

The common_transform stuff was meant to be reused by the PBQP code
selection which is not maintained anymore.

The only user should be an arch_add_irn_flags() instead.
It was harmless though, because no other flags were set.