On Fri, Nov 04, 2016 at 04:09:37PM +0000, Maciej W. Rozycki wrote: > On Fri, 4 Nov 2016, Guenter Roeck wrote: > > > > As above, unless absolutely critical to have floating point code then > > > the vDSO should just avoid all FP related issues and build soft-float. > > > > FWIW, my logic was quite simple: The rest of the kernel builds with > > -msoft-float, thus vDSO should do the same. Of course, I don't know the > > entire context, so there may well be a reason to handle it differently > > than the rest of the kernel. > > VDSO is not a part of the kernel, it's user mode code, made visible in > the user virtual memory, and implicitly loaded along the rest of the DSOs > on program startup by the dynamic loader (ld.so). It has to be PIC for > that reason, too, causing all the hassle we had with making it such that > it does not need a GOT. > > > Anyway, isn't the kernel supposed to not use floating point operations > > in the first place ? Is this different for vDSO ? > > This code is executed in the user mode so while floating-point code may > not be needed there, not at least right now, there's actually nothing > which should stop us from from adding some should such a need arise. > Just for my understanding - so the code is compiled with the kernel and part of the kernel source but executed in user mode ? If you ever add real floating point code, doesn't that also mean that you'll have to implement the necessary linker helper functions or wrappers (such as the wrappers needed for 64-bit integer divide operations in 32 bit code) ? Thanks, Guenter
