From 116ac378bb3ff844df333e7609e7604651a0db9d Mon Sep 17 00:00:00 2001 From: Nicholas Piggin Date: Fri, 8 May 2020 14:34:03 +1000 Subject: powerpc/64s: machine check interrupt update NMI accounting machine_check_early() is taken as an NMI, so nmi_enter() is used there. machine_check_exception() is no longer taken as an NMI (it's invoked via irq_work in the case a machine check hits in kernel mode), so remove the nmi_enter() from that case. In NMI context, hash faults don't try to refill the hash table, which can lead to crashes accessing non-pinned kernel pages. System reset still has this potential problem. Signed-off-by: Nicholas Piggin [mpe: Drop change in show_regs() which breaks Book3E] Signed-off-by: Michael Ellerman Link: https://lore.kernel.org/r/20200508043408.886394-12-npiggin@gmail.com --- arch/powerpc/kernel/mce.c | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'arch/powerpc/kernel/mce.c') diff --git a/arch/powerpc/kernel/mce.c b/arch/powerpc/kernel/mce.c index 8077b5fb18a7..be7e3f92a7b5 100644 --- a/arch/powerpc/kernel/mce.c +++ b/arch/powerpc/kernel/mce.c @@ -574,6 +574,9 @@ EXPORT_SYMBOL_GPL(machine_check_print_event_info); long machine_check_early(struct pt_regs *regs) { long handled = 0; + bool nested = in_nmi(); + if (!nested) + nmi_enter(); hv_nmi_check_nonrecoverable(regs); @@ -582,6 +585,10 @@ long machine_check_early(struct pt_regs *regs) */ if (ppc_md.machine_check_early) handled = ppc_md.machine_check_early(regs); + + if (!nested) + nmi_exit(); + return handled; } -- cgit v1.2.3