On Thu, 7 Dec 2017, at 12:58 AM, Tanu Kaskinen wrote:
> On Mon, 2017-12-04 at 19:44 +0530, Arun Raghavan wrote:
> > On Wed, 22 Nov 2017, at 08:44 PM, Tanu Kaskinen wrote:
> > > The get_cpuid() function in cpu-x86.c was buggy on x86-64. When building
> > > without optimizations, the homegrown assembly code overwrote the
> > > beginning of the function argument list on the stack. That happened to
> > > work fine on regular x86-64, but caused crashing with the x32 ABI.
> > >
> > > At least GCC and clang provide cpuid.h, which has the __get_cpuid()
> > > function that can be used instead of the homegrown assembly.
> >
> > Just as a sanity check -- does this introduce any compiler version
> > dependencies? The clang header seems to have been updated relatively
> > recently for SSE 4.1/4.2 (2016), but the gcc header seems quite old, so
> > probably okay.
>
> I haven't checked the history of the compilers, but if there are any
> compilers in use that don't support __get_cpuid(), they will hopefully
> just skip the cpuid parts since the code is guarded by HAVE_CPUID_H.
>
> > > @@ -3663,11 +3666,13 @@ bool pa_running_in_vm(void) {
> >
> > I wonder if the use-case for this is still valid (not being able to run
> > tsched in a VM).
>
> I don't know, and I don't plan to find out... If you want to test
> tsched in all the VMs, be my guest :)
>
> I suppose the patch is okay to push?
Yes, please go ahead.
Cheers,
Arun
