On Thu, Dec 20, 2012 at 1:06 AM, Benyei, Guy <[email protected]> wrote:
> Hi Eli,
>
> From the LLVM Language Reference Manual:
>
> " When constructing the data layout for a given target, LLVM starts with a 
> default set of specifications which are then (possibly) overridden by the 
> specifications in the datalayout keyword. The default specifications are 
> given in this list:
>
> E - big endian..."
>
> It seems to me, that the default big endian is taken even if I don't define a 
> data layout, and the rest of the data is still relevant.
> I'm not sure how to handle this, from the SPIR point of view it's possible to 
> completely ignore the issue by declaring that the default endian in SPIR is 
> neither little nor big, and this IR is still strong enough to work. Another 
> solution would be to introduce a new notation for "unknown" endian.
>
> Any other ideas on this issue?

>From later in that section of LangRef: "If you would like to generate
IR that does not embed this target-specific detail into the IR, then
you don’t have to specify the string."

-Eli

_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

Reply via email to