Update pnacl/DEPS to bring in merge to r172453

Update pnacl/DEPS to bring in merge to r172453

There was one MIPS test failure introduced, I haven't made an attempt to debug it.

R=jvoung@chromium.org,eliben@chromium.org
BUG= https://code.google.com/p/nativeclient/issues/detail?id=3178
BUG= https://code.google.com/p/nativeclient/issues/detail?id=2275

Committed: https://src.chromium.org/viewvc/native_client?view=rev&revision=10714

[Derek Schuff, 2013-02-05 00:14:05]

the current LLVM downtime is causing spurious build failures, and the recent mac
buildbot xcode change is causing the new mac test failure.

[jvoung (off chromium), 2013-02-05 17:01:55]

On 2013/02/05 00:14:05, Derek Schuff wrote:
> the current LLVM downtime is causing spurious build failures, and the recent
mac
> buildbot xcode change is causing the new mac test failure.

Looking at:
http://git.chromium.org/gitweb/?p=native_client/pnacl-llvm.git;a=commitdiff;h...


There is a:

-   /// getBundleSize - Return the size (in bytes) of code bundling units
-   /// for this target. If 0, bundling is disabled. This is used exclusively
-   /// for Native Client.
-   virtual unsigned getBundleSize() const {
-     return 0;
-   }
- 

in MCAsmBackend.h, but there are still


 " +  unsigned getBundleSize() const { return 16; } "

localmods under ARMAsmBackend... is that still necessary?  Is it supposed to be
replaced by EmitBundleAlignMode(4) ?

[Derek Schuff, 2013-02-05 17:12:12]

Ah, you are right, that's not necessary.

On 2013/02/05 17:01:55, jvoung (cr) wrote:
> On 2013/02/05 00:14:05, Derek Schuff wrote:
> > the current LLVM downtime is causing spurious build failures, and the recent
> mac
> > buildbot xcode change is causing the new mac test failure.
> 
> Looking at:
>
http://git.chromium.org/gitweb/?p=native_client/pnacl-llvm.git;a=commitdiff;h...
> 
> 
> There is a:
> 
> -   /// getBundleSize - Return the size (in bytes) of code bundling units
> -   /// for this target. If 0, bundling is disabled. This is used exclusively
> -   /// for Native Client.
> -   virtual unsigned getBundleSize() const {
> -     return 0;
> -   }
> - 
> 
> in MCAsmBackend.h, but there are still
> 
> 
>  " +  unsigned getBundleSize() const { return 16; } "
> 
> localmods under ARMAsmBackend... is that still necessary?  Is it supposed to
be
> replaced by EmitBundleAlignMode(4) ?

[jvoung (off chromium), 2013-02-05 17:14:32]

otherwise LGTM

On 2013/02/05 17:12:12, Derek Schuff wrote:
> Ah, you are right, that's not necessary.
> 
> On 2013/02/05 17:01:55, jvoung (cr) wrote:
> > On 2013/02/05 00:14:05, Derek Schuff wrote:
> > > the current LLVM downtime is causing spurious build failures, and the
recent
> > mac
> > > buildbot xcode change is causing the new mac test failure.
> > 
> > Looking at:
> >
>
http://git.chromium.org/gitweb/?p=native_client/pnacl-llvm.git;a=commitdiff;h...
> > 
> > 
> > There is a:
> > 
> > -   /// getBundleSize - Return the size (in bytes) of code bundling units
> > -   /// for this target. If 0, bundling is disabled. This is used
exclusively
> > -   /// for Native Client.
> > -   virtual unsigned getBundleSize() const {
> > -     return 0;
> > -   }
> > - 
> > 
> > in MCAsmBackend.h, but there are still
> > 
> > 
> >  " +  unsigned getBundleSize() const { return 16; } "
> > 
> > localmods under ARMAsmBackend... is that still necessary?  Is it supposed to
> be
> > replaced by EmitBundleAlignMode(4) ?

[petarj, 2013-02-07 00:15:45]

@Derek

The failing test for MIPS comes from different value for Data alignment
factor in .eh_frame section.

And this leads to local modification in getDataAlignmentFactor in
lib/MC/MCDwarf.cpp.

[eliben, 2013-02-07 00:25:15]

On 2013/02/07 00:15:45, petarj wrote:
> @Derek
> 
> The failing test for MIPS comes from different value for Data alignment
> factor in .eh_frame section.
> 
> And this leads to local modification in getDataAlignmentFactor in
> lib/MC/MCDwarf.cpp.

This is something that should be fixed for x86 with the x32 changes. How is MIPS
affected?

[petarj, 2013-02-07 01:58:38]

On 2013/02/07 00:25:15, eliben wrote:

The test (MC/Mips/ef_frame.ll) expects different value in Data alignment
factor.