On 03.04.2025 21:21, Andrew Cooper wrote: > On 03/04/2025 7:23 pm, dm...@proton.me wrote: >> From: Denis Mukhin <dmuk...@ford.com> >> >> The new toolchain baseline knows the rdrand/rdseed instructions, >> no need to carry the workaround in the code. >> >> Resolves: https://gitlab.com/xen-project/xen/-/work_items/208 >> Signed-off-by: Denis Mukhin <dmuk...@ford.com> >> --- >> xen/arch/x86/arch.mk | 2 -- >> xen/arch/x86/x86_emulate/0fc7.c | 15 +++++---------- > > Huh... > > I was expecting a hunk in xen/arch/x86/include/asm/random.h but lookin > at it, I see that didn't even get the HAVE_AS_RDRAND treatment in the > first place. Also, the + constraint, rather than =, looks suspicious.
Might that have been for a similar reason as the one you recently added a XOR ahead of POPCNT for? IOW us wanting the clearing of the output to happen irrespective of whether the if() condition is true? Jan
