[cfe-commits] [PATCH] Use custom ABIInfo for le32/PNaCl argument codegen

[Derek Schuff]

There's a document at
https://sites.google.com/a/chromium.org/dev/nativeclient/pnacl/bitcode-abi
but all it says currently about calling conventions right now is
basically what it says in the tests; i.e. pass structs byval and don't
split or simplify them, 2 inreg params, etc. This patch makes the
behavior close to the previous behavior (up until before r161022 when
inreg handling was moved from CodeGenModule::ConstructAttributeList
into the target-specific ABIInfos). We are actually reviewing this
part of the spec to better specify it and may make some changes. But
this patch is a start.

Updated patch attached, thanks.
-Derek

On Thu, Aug 30, 2012 at 3:14 PM, Eli Friedman <eli.friedman at gmail.com> wrote:
> On Thu, Aug 30, 2012 at 2:05 PM, Derek Schuff <dschuff at google.com> wrote:
>> This patch uses a new ABIInfo implementation specific to the le32
>> target, rather than falling back to DefaultABIInfo. Its behavior is
>> basically the same, but it also allows the regparm argument attribute.
>> It also includes basic tests for argument codegen and attributes.
>> Please review, thanks.
>
> Do you have some sort of ABI spec you're writing this against?  Some
> of the regparm code seems a little weird, but I can't say it's wrong
> without anything to compare against.
>
> +// PR7025
>
> Please update the comments in the tests.
>
> -Eli
-------------- next part --------------
A non-text attachment was scrubbed...
Name: le32-args-2.patch
Type: application/octet-stream
Size: 7122 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20120831/0ef3e293/attachment.obj>