ppc: build test mutators against the ppc64 register namespace - #275
ppc: build test mutators against the ppc64 register namespace#275bbiiggppiigg wants to merge 1 commit into
Conversation
The PowerPC instruction decoder and process-control stack move entirely to the
ppc64 register namespace (ppc32 is unsupported; its enum value Arch_ppc32 and
its register namespace are removed). Two test sources still referenced them and
no longer compile:
- instruction/power_decode.C decoded with Arch_ppc32 and compared results
against ppc32:: register ASTs. Switch both to Arch_ppc64 / ppc64:: so the
expected registers match what the decoder now emits.
- proccontrol/pc_irpc_asm.h carried an Arch_ppc32 iRPC code-generation case
referencing the removed enumerator. Drop it (ppc32 is unsupported).
Depends on the Dyninst change that removes Arch_ppc32 and emits ppc64-namespace
registers from the PowerPC decoder; build and run this on ppc64 together with it.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
Are these changes reflected in dyninst/dyninst#2344? |
No. I thought that belongs to when we drop the ppc32 support. |
We're dropping support for ppc32? I've looked at this, and many Linux distros dropped support within the last decade but there are some that still support little* endian 32 (e.g., RHEL 9). Big endian is ppc64 only, as far as I can tell for at least the last 8-10 year. |
I don't know if that is happening or not, but as per |
2 participants
The PowerPC instruction decoder and process-control stack move entirely to the ppc64 register namespace (ppc32 is unsupported; its enum value Arch_ppc32 and its register namespace are removed). Two test sources still referenced them and no longer compile:
Depends on the Dyninst change that removes Arch_ppc32 and emits ppc64-namespace registers from the PowerPC decoder; build and run this on ppc64 together with it.