This is the mail archive of the mailing list for the glibc project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] fix BZ 18116 - build failure on ppc64le: setcontext.S uses power6 mtfsf when not supported

On Wed, 2015-03-18 at 10:52 -0600, Martin Sebor wrote:
> >> You mean around each of the instances, as in the attached patch,
> >> as opposed to around the whole #if/#else block?  Would you mind
> >> explaining in what way you think it's more correct than the first
> >> patch (a single pair of .machine directives around the whole
> >> conditional block)?
> >>
> > I mean around each instance of mtfsf  0xff,fp0,1,0
> >
> > What you suggested is only correct and safe for IBM POWER hardware
> > Power6 and later. It is not safe or correct for powerpc64 chips from
> > other manufacturer that implement a subset of the PowerISA-2.05 (or
> > later).
> I suspect you've misread the changes in the first patch.
> The first and the third patches are equivalent and after
> preprocessing will result in the assembler seeing the
> exact same code, including directives, regardless of what
> the target cpu is.

The 3rd patch dated March 16th is good. 

My response on March 17th was to answer your questions in the text.

Sorry I did not respond earlier but our IMAP server has  been down for
the last 3 days.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]