GitLab

noone was using it anyway

switch with double-word input leads to panic, but still better than
obscure verifier errors.

Reusing the same dummy struct as a hidden parameter for multiple
calls is not always correct (see opt/structcall2.c for an example).
For now, just use multiple dummy structs.

This means floatingpoint calculation in a firm graph must be performed
in the mode specified (and may not be performed in a wider mode as
before).

You can still attach NoMem if no memory input is present.
This makes it more consistent with the Call node and fixes failing
get_ASM_mem in case of no memory input.

Let list_for_each_entry(), list_for_each_entry_reverse() and list_for_each_entry_safe() declare their iterator variables.

Remove now redundant (and sometimes wrong) location info from the panic messages.

This makes it consistent with other firm analyse functions. Also removes
the unnecessary return value.

graph_properties are things that a transformation may destroy but which
can be easily recalculated.

graph_constraints are set after lowering phases have been running and
forbid certain constructs that no (later) transformation is allowed to construct

First it is unclear what is a good way to deconstruct a program into extended
basic blocks, also none of our optimizations works on extended basic
blocks.

It was a member of ir_prog before but not correctly handled.

The opcode list was a member of irprog before which wasn't really
handled consistently. Also make sure opcodes are properly freed at
ir_finish().

unknown_type, code_type, none_type, unknown_entity reference are hold in
the irprog now. This makes handling more consistent since now all types
and entities are equally part of irprog.

public functions are already documented in their headers, no need to
repeat the comment at the definition site.

It was a strange feature only inspired by the lazyness of some frontend
developer (=me)