GitLab

All other structs holding a register class call it cls, too.

Just do the check in the corresponding implementation of arch_perform_memory_operand().

this also required a rewrite of the perform_memory_operand logic which
should be cleaner now.

As they were just a single flag the name should reflect that. I did not
change arch_irn_flags_none, as that is not about a single flag.

It has been empty for nearly all systems. People who used to put stuff
in config.h can still create a config.h on their own and inject
-include (gcc) or /FI (msvc) into their CPPFLAGS.

be: Do not dump the input node, only the slot, for "should be same" and "must be different" constraints.

The node in question is shown just a few lines above and repeating it here just clutters up the info box.

When arch_perform_memory_operand() returned a positive result before, there really should be a callback to actually perform the action.

Just state the copyright and the project the files belong to. You can
then find the COPYING file for the whole project.

bearch: When dumping backend info, do not print an error, when a node has no in_reqs, if its arity is 0.

bemain: Centrally call be_gas_end_compilation_unit() and be_emit_exit() instead of doing it per backend.

Also remove the now write-only attribute main_env from struct arch_env_t.

bemain: Centrally call be_emit_init() and be_gas_begin_compilation_unit() instead of doing it per backend.

bearch: Dump the output requirement and the assigned register in the same line for better readability.

make it more robust and remove unnecessary req check

This also corrects dumping schedule edges.

arch_register_type_virtual is sufficient.

They must match, even for joker registers.

bearch: Assert, instead of test, that the register requirement is not arch_register_req_type_none in arch_reg_is_allocatable().

If it is, something is very wrong.

so external users can set their own opcode functions.