KVM: PPC: Book3S HV: trace_tlbie must not be called in realmode
authorNicholas Piggin <npiggin@gmail.com>
Thu, 5 Apr 2018 17:56:30 +0000 (03:56 +1000)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 19 Apr 2018 06:56:17 +0000 (08:56 +0200)
commit42b53a13ecac396d74c126c562b18ae8c6a99341
tree80469990b1a821c7f0dae603d5c16b76997cee2d
parent5661d43b03c5b3dde9262a7bbd417991eba0723f
KVM: PPC: Book3S HV: trace_tlbie must not be called in realmode

commit 19ce7909ed11c49f7eddf59e7f49cd3062bf83d5 upstream.

This crashes with a "Bad real address for load" attempting to load
from the vmalloc region in realmode (faulting address is in DAR).

  Oops: Bad interrupt in KVM entry/exit code, sig: 6 [#1]
  LE SMP NR_CPUS=2048 NUMA PowerNV
  CPU: 53 PID: 6582 Comm: qemu-system-ppc Not tainted 4.16.0-01530-g43d1859f0994
  NIP:  c0000000000155ac LR: c0000000000c2430 CTR: c000000000015580
  REGS: c000000fff76dd80 TRAP: 0200   Not tainted  (4.16.0-01530-g43d1859f0994)
  MSR:  9000000000201003 <SF,HV,ME,RI,LE>  CR: 48082222  XER: 00000000
  CFAR: 0000000102900ef0 DAR: d00017fffd941a28 DSISR: 00000040 SOFTE: 3
  NIP [c0000000000155ac] perf_trace_tlbie+0x2c/0x1a0
  LR [c0000000000c2430] do_tlbies+0x230/0x2f0

I suspect the reason is the per-cpu data is not in the linear chunk.
This could be restored if that was able to be fixed, but for now,
just remove the tracepoints.

Fixes: 0428491cba92 ("powerpc/mm: Trace tlbie(l) instructions")
Cc: stable@vger.kernel.org # v4.13+
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
arch/powerpc/kvm/book3s_hv_rm_mmu.c