[PATCH] Verify clang TargetInfo DataLayouts against LLVM

[Alp Toker]

On 31/12/2013 18:34, Rafael Espíndola wrote:
> On 29 December 2013 23:53, Alp Toker <alp at nuanti.com> wrote:
>> Hi Rafael,
>>
>> This is all that's left to enable opportunistic checking at clang IRGen time
>> where an LLVM TargetMachine is available.
>>
>>   include/clang/CodeGen/BackendUtil.h |    4 ++--
>>   lib/CodeGen/BackendUtil.cpp         |   11 ++++++++---
>>   lib/CodeGen/CodeGenAction.cpp       |    3 ++-
>>   3 files changed, 12 insertions(+), 6 deletions(-)
>>
>> Re-posting the updated patch as I lost track of the original thread, and
>> CC'ing in llvm-commits.
>>
>> You mentioned there's a remaining inconsistent ARM DataLayout that needs to
>> get fixed before we switch this on -- was it determined if it's the LLVM or
>> clang string that's at fault?
> Not so much at fault. The issue is that llvm uses only the triple,
> where clangs uses an attribute for the abi. I will ping that patch and
> CC you.

Got it, thanks.

Guessing clang also needs an update following your LLVM r198287 yesterday:

error: frontend target description 
'e-i64:64-f80:128-s:64-n8:16:32:64-S128' does not match backend data 
layout 'e-i64:64-f80:128-n8:16:32:64-S128'

>
> Sorry for disappearing. I am spending the holidays with my parents, so
> reading just a bit of email.

No rush, enjoy the break Rafael :-)

Alp.


>
> Cheers,
> Rafael

-- 
http://www.nuanti.com
the browser experts