This should be a low level function used to interpret an MCInstrDesc that only depends on the hwmode. It should not depend on other dynamic context like the parent function. In general more ABI properties like this should be expressed directly in the instruction definitions, so introduce new TCRETURN pseudos to use with the special case register classes (e.g. in a better future the callee saved registers would always be encoded directly in a mask on the return instruction). This will help unify X86 onto a pending replacement mechanism for getPointerRegClass.
Fixes regression casued by #156817.
These are identified by misc-include-cleaner. I've filtered out those that break builds. Also, I'm staying away from llvm-config.h, config.h, and Compiler.h, which likely cause platform- or compiler-specific build failures.
The first commit is identical to
69bec0afbb8f2aa0021d18ea38768360b16583a9.
The second commit fixes the instruction verification failures by
replacing the erroneous instruction with a trap after the error is
reported and adds `-verify-machineinstrs` to the tests added in the
original PR to catch the issue sooner.
After that change, all tests pass with both
`LLVM_ENABLE_EXPENSIVE_CHECKS={On,Off}`.
cc @RKSimon @e-kud @phoebewang @arsenm as reviewers on the original PR
Reverts llvm/llvm-project#123872 - this is breaking on EXPENSIVE_CHECKS builds Co-authored-by: Abhishek Kaushik <abhishek.kaushik@intel.com>
The assertion previously did not work correctly because the operand was being truncated to an `int` prior to comparison. Change the assertion into a a reported error as suggested in https://github.com/llvm/llvm-project/pull/101840#issuecomment-2304992425 by @arsenm Finally, fix the lowering on 64-bit targets so that offsets larger than 32-bit are correctly addressed and add tests for various reported issues.
This fixes the x86 backend to properly reserve `rbp` when the `frame-pointer=reserved` option is used. Currently, this option is ignored. Disassembly of the new test case before: ```s pushq %rbp pushq %rbx pushq %rax movl %esi, %ebx movl %edi, %ebp callq bar@PLT movl %ebp, %edi movl %ebx, %esi callq bar@PLT addq $8, %rsp popq %rbx popq %rbp retq ``` ...and after (`r14` is used as scratch space in place of `rbp`): ```s pushq %r14 pushq %rbx pushq %rax movl %esi, %ebx movl %edi, %r14d callq bar@PLT movl %r14d, %edi movl %ebx, %esi callq bar@PLT addq $8, %rsp popq %rbx popq %r14 retq ``` Fixes #117178.
Without newlines at the end, git diff would display: No newline at end of file