- buffer values are never modified so we can use char instead of unsigned char.
Sorry, I meant to accept this before it landed.
2: On AIX, the assembler does not need the assembly file contains the dwarf sections length info in the dwarf section header(if the dwarf section has header.) Instead, the assembler will insert the calculated length into dwarf sections header of the final object according to DWARF type. Namely AIX assembler will insert 4 bytes in each section header for DWARF32 and 12 bytes for DWARF64.
Looks like we get a lot of happiness from using these attributes. ;-)
I had a look at it before and the general direction also looked okay to me too, which I was happy to see confirmed. Therefore I am happy to accept this, even though I haven't checked all intrinsics because this is a lot of changes. So perhaps wait a day with committing in case anyone wants to have one more look.
Eh you beat me to it. Had a similar patch I've been measuring and testing. Your's is more complete though :)
Sorry about the last change, that was the wrong patch.
LGTM from a change point of view. I can't really add to the discussion about how/when to switch from 0.92 to 0.93 so I'm not saying "merge right away" :)
Makes sense to me. I think we'll have to handle "misaligned" vectors at some point too, since the DataLayout can only represent the alignment of vectors based on their overall size.
(Some ARM-aware ppl should review/accept)