powerpc/64s: Disable STRICT_KERNEL_RWX

[ Upstream commit 8659a0e0efdd975c73355dbc033f79ba3b31e82c ]

Several strange crashes have been eventually traced back to
STRICT_KERNEL_RWX and its interaction with code patching.

Various paths in our ftrace, kprobes and other patching code need to
be hardened against patching failures, otherwise we can end up running
with partially/incorrectly patched ftrace paths, kprobes or jump
labels, which can then cause strange crashes.

Although fixes for those are in development, they're not -rc material.

There also seem to be problems with the underlying strict RWX logic,
which needs further debugging.

So for now disable STRICT_KERNEL_RWX on 64-bit to prevent people from
enabling the option and tripping over the bugs.

Fixes: 1e0fc9d1eb2b ("powerpc/Kconfig: Enable STRICT_KERNEL_RWX for some configs")
Cc: stable@vger.kernel.org # v4.13+
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200520133605.972649-1-mpe@ellerman.id.au
Signed-off-by: Sasha Levin <sashal@kernel.org>

diff --git a/arch/powerpc/Kconfig b/arch/powerpc/Kconfig
index b74c3a68c0ad30a1082bf8e068162506dddb68ae..679e1e3c1695390699b7ffae7d7a5d964ff0ed62 100644 (file)
--- a/arch/powerpc/Kconfig
+++ b/arch/powerpc/Kconfig
@@ -141,7 +141,7 @@ config PPC
        select ARCH_HAS_GCOV_PROFILE_ALL
        select ARCH_HAS_SCALED_CPUTIME          if VIRT_CPU_ACCOUNTING_NATIVE
        select ARCH_HAS_SG_CHAIN
-       select ARCH_HAS_STRICT_KERNEL_RWX       if ((PPC_BOOK3S_64 || PPC32) && !HIBERNATION)
+       select ARCH_HAS_STRICT_KERNEL_RWX       if (PPC32 && !HIBERNATION)
        select ARCH_HAS_TICK_BROADCAST          if GENERIC_CLOCKEVENTS_BROADCAST
        select ARCH_HAS_UBSAN_SANITIZE_ALL
        select ARCH_HAS_ZONE_DEVICE             if PPC_BOOK3S_64