On 24.06.2019 21:51, maya%netbsd.org@localhost wrote: > I didn't put any of my own input, I tried to be as close to "revert the > commit" as possible, but might've made some inattentive errors. > > To clarify, amd64 32bit debugging isn't my highest priority, I am OK if > this waits past branching, I'm just after MIPS debugging, and it runs a > compat32 userland. It's not happy after this patch either, but I don't > know if it ever was. > We really ought to test compat32 more systematically. I was thinking about going for rebuilding src/tests with -m32 and installing it into /usr/tests32 whenever it is applicable in a port (so, applicable for at least: sparc64, amd64, evbarm64, evbppc64 and maybe evbmips64 ?). Full ptrace(2) testing against compat32 with pure ptrace(2) is more tricky as we might be unaware whether there is trash in registers, without writing at least a minimal featured debugger. I intend to test this through GDB/LLDB in ATF. For start, adding /usr/tests32 with ptrace(2) tests will make the bar higher for quality of the ptrace(2) APIs.
Attachment:
signature.asc
Description: OpenPGP digital signature