| Message ID | 20170405062611.GB26143@fergus.ozlabs.ibm.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
diff --git a/arch/powerpc/kvm/book3s_hv_rmhandlers.S b/arch/powerpc/kvm/book3s_hv_rmhandlers.S index 6cadda5fa7d4..35d8d718e34b 100644 --- a/arch/powerpc/kvm/book3s_hv_rmhandlers.S +++ b/arch/powerpc/kvm/book3s_hv_rmhandlers.S @@ -1240,7 +1240,11 @@ END_FTR_SECTION_IFSET(CPU_FTR_HAS_PPR) cmpwi r12, BOOK3S_INTERRUPT_H_DOORBELL bne 3f BEGIN_FTR_SECTION +BEGIN_FTR_SECTION_NESTED(98) PPC_MSGSYNC +FTR_SECTION_ELSE_NESTED(98) + .long (31<<26) + (755<<1) /* darn r0,0 */ +ALT_FTR_SECTION_END_NESTED_IFCLR(CPU_FTR_POWER9_DD1, 98) lwsync END_FTR_SECTION_IFSET(CPU_FTR_ARCH_300) lbz r0, HSTATE_HOST_IPI(r13)
On POWER9 DD1 the msgsync instruction doesn't work correctly, but the new 'darn' (deliver a random number) instruction happens to do an equivalent function as a side effect. Signed-off-by: Paul Mackerras <paulus@ozlabs.org> --- arch/powerpc/kvm/book3s_hv_rmhandlers.S | 4 ++++ 1 file changed, 4 insertions(+)