Skip ABI verification for tests of __builtin_{frame,return}_address.

Skip ABI verification for tests of __builtin_{frame,return}_address.

Skip in scons and the gcc torture test suite.

Also skip ABI verification for the LLVM math intrinsics,
bitmanip intrinsics, and atomic intrinsics in scons for now.
Haven't hunted those down in gcc torture test suite yet.

BUG=https://code.google.com/p/nativeclient/issues/detail?id=3378
R=mseaborn@chromium.org

Committed: https://src.chromium.org/viewvc/native_client?view=rev&revision=11329

[Mark Seaborn, 2013-05-09 21:26:37]

https://codereview.chromium.org/14619022/diff/9005/tests/toolchain/nacl.scons
File tests/toolchain/nacl.scons (right):

https://codereview.chromium.org/14619022/diff/9005/tests/toolchain/nacl.scons...
tests/toolchain/nacl.scons:56: # (src, exit_status, golden_file, stable_bitcode,
cond),
This is getting hard to read -- too many fields.

Maybe replace with function calls for readability?

def AddIntrinsicTest(src, exit_status, golden_file, stable_bitcode, cond=True):
  ...

AddIntrinsicTest('setlongjmp.c', '55', None, stable_bitcode=True)
...

https://codereview.chromium.org/14619022/diff/9005/tests/toolchain/nacl.scons...
tests/toolchain/nacl.scons:80: if stable_bitcode:
How about this to reduce duplication:
if stable_bitcode:
  test_env = env
else:
  test_env = nonstable_env
nexe = env.ComponentProgram(...)

or
test_env = {True: env, False: nonstable_env}[stable_bitcode]

if you prefer :-)

[eliben, 2013-05-09 21:33:47]

On 2013/05/09 21:26:37, Mark Seaborn wrote:
> https://codereview.chromium.org/14619022/diff/9005/tests/toolchain/nacl.scons
> File tests/toolchain/nacl.scons (right):
> 
>
https://codereview.chromium.org/14619022/diff/9005/tests/toolchain/nacl.scons...
> tests/toolchain/nacl.scons:56: # (src, exit_status, golden_file,
stable_bitcode,
> cond),
> This is getting hard to read -- too many fields.
> 
> Maybe replace with function calls for readability?
> 
> def AddIntrinsicTest(src, exit_status, golden_file, stable_bitcode,
cond=True):
>   ...
> 
> AddIntrinsicTest('setlongjmp.c', '55', None, stable_bitcode=True)
> ...
> 
>
https://codereview.chromium.org/14619022/diff/9005/tests/toolchain/nacl.scons...
> tests/toolchain/nacl.scons:80: if stable_bitcode:
> How about this to reduce duplication:
> if stable_bitcode:
>   test_env = env
> else:
>   test_env = nonstable_env
> nexe = env.ComponentProgram(...)
> 
> or
> test_env = {True: env, False: nonstable_env}[stable_bitcode]
> 
> if you prefer :-)

If we're golfing then:

test_env = env if stable_bitcode else nontstable_env

:-)

[jvoung (off chromium), 2013-05-09 23:31:42]

https://codereview.chromium.org/14619022/diff/9005/tests/toolchain/nacl.scons
File tests/toolchain/nacl.scons (right):

https://codereview.chromium.org/14619022/diff/9005/tests/toolchain/nacl.scons...
tests/toolchain/nacl.scons:56: # (src, exit_status, golden_file, stable_bitcode,
cond),
On 2013/05/09 21:26:37, Mark Seaborn wrote:
> This is getting hard to read -- too many fields.
> 
> Maybe replace with function calls for readability?
> 
> def AddIntrinsicTest(src, exit_status, golden_file, stable_bitcode,
cond=True):
>   ...
> 
> AddIntrinsicTest('setlongjmp.c', '55', None, stable_bitcode=True)
> ...

Good idea!  Done.

https://codereview.chromium.org/14619022/diff/9005/tests/toolchain/nacl.scons...
tests/toolchain/nacl.scons:80: if stable_bitcode:
On 2013/05/09 21:26:37, Mark Seaborn wrote:
> How about this to reduce duplication:
> if stable_bitcode:
>   test_env = env
> else:
>   test_env = nonstable_env
> nexe = env.ComponentProgram(...)
> 
> or
> test_env = {True: env, False: nonstable_env}[stable_bitcode]
> 
> if you prefer :-)

Reduced duplication.

[Mark Seaborn, 2013-05-09 23:39:11]

LGTM

https://codereview.chromium.org/14619022/diff/1008/tests/toolchain/nacl.scons
File tests/toolchain/nacl.scons (right):

https://codereview.chromium.org/14619022/diff/1008/tests/toolchain/nacl.scons...
tests/toolchain/nacl.scons:49: nonstable_env = env.Clone()
Need to move this line out of the conditional otherwise you'll get a NameError
below in the non-bitcode case.

https://codereview.chromium.org/14619022/diff/1008/tests/toolchain/nacl.scons...
tests/toolchain/nacl.scons:80: has_golden_file=True, cond=env.Bit('bitcode'))
You could put this in an "if" instead of using the "cond" arg.

https://codereview.chromium.org/14619022/diff/1008/tools/toolchain_tester/kno...
File tools/toolchain_tester/known_failures_pnacl.txt (right):

https://codereview.chromium.org/14619022/diff/1008/tools/toolchain_tester/kno...
tools/toolchain_tester/known_failures_pnacl.txt:183: # This uses
__builtin_return_address(1)
'This uses' -> '20010122-1.c uses', otherwise this appears to refer to the whole
list.

https://codereview.chromium.org/14619022/diff/1008/tools/toolchain_tester/kno...
tools/toolchain_tester/known_failures_pnacl.txt:184: # Segfaults because of an
uninit frame pointer (must force fp on?).
'20010122-1.c segfaults...'

[jvoung (off chromium), 2013-05-09 23:48:25]

https://codereview.chromium.org/14619022/diff/1008/tests/toolchain/nacl.scons
File tests/toolchain/nacl.scons (right):

https://codereview.chromium.org/14619022/diff/1008/tests/toolchain/nacl.scons...
tests/toolchain/nacl.scons:49: nonstable_env = env.Clone()
On 2013/05/09 23:39:11, Mark Seaborn wrote:
> Need to move this line out of the conditional otherwise you'll get a NameError
> below in the non-bitcode case.

Oops, how about set nonstable_env = env in the non-bitcode case?  Don't really
need to pass --pnacl-disable... to gcc.

https://codereview.chromium.org/14619022/diff/1008/tests/toolchain/nacl.scons...
tests/toolchain/nacl.scons:80: has_golden_file=True, cond=env.Bit('bitcode'))
On 2013/05/09 23:39:11, Mark Seaborn wrote:
> You could put this in an "if" instead of using the "cond" arg.

Done.

[jvoung (off chromium), 2013-05-10 15:23:18]

Committed patchset #9 manually as r11329 (presubmit successful).