From: Eric Botcazou <ebotca...@adacore.com> Date: Tue, 13 Nov 2012 22:50:49 +0100
>> Thanks for finding this, that's definitely incorrect behavior. I bet there >> is some unintended override triggered by sparc4 selection, and I'll go and >> fix that soon. > > You're welcome. That's the reason why I needed to go the ASM_ARCH way, the > straightforward approach would have put the -32/-64 first. Right. And meanwhile I found the problem, there is this code block in the Sparc option parser of GAS that goes: case OPTION_XARCH: #ifdef OBJ_ELF if (strncmp (arg, "v9", 2) != 0) md_parse_option (OPTION_32, NULL); else md_parse_option (OPTION_64, NULL); #endif /* Fall through. */ And that's where the unexpected size override is happening. That test simply needs to be adjusted and I'll try to sort that out tonight. > And I managed to miss a substitution in the previous patch, so please drop it > and use the attached one instead. Thanks for this, I was just starting to work on integrating our work together and debugging the result.