[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-Id: <20180516054021.5328-1-ravi.bangoria@linux.ibm.com>
Date: Wed, 16 May 2018 11:10:21 +0530
From: Ravi Bangoria <ravi.bangoria@...ux.ibm.com>
To: mpe@...erman.id.au
Cc: benh@...nel.crashing.org, paulus@...ba.org,
naveen.n.rao@...ux.vnet.ibm.com, anton@...ba.org,
cyrilbur@...il.com, matthew.brown.dev@...il.com,
sandipan@...ux.vnet.ibm.com, linuxppc-dev@...ts.ozlabs.org,
linux-kernel@...r.kernel.org,
Ravi Bangoria <ravi.bangoria@...ux.ibm.com>
Subject: [RFC] powerpc/emulate_step: Fix kernel crash when VSX is not present
emulate_step() is not checking runtime VSX feature flag before
emulating an instruction. This can cause kernel oops when kernel
is compiled with CONFIG_VSX=y but running on machine where VSX is
not supported or disabled. Ex, while running emulate_step tests on
P6 machine:
...
emulate_step_test: lvx : PASS
emulate_step_test: stvx : PASS
Oops: Exception in kernel mode, sig: 4 [#1]
NIP [c000000000095c24] .load_vsrn+0x28/0x54
LR [c000000000094bdc] .emulate_loadstore+0x167c/0x17b0
Call Trace:
[c0000000494c3770] [40fe240c7ae147ae] 0x40fe240c7ae147ae (unreliable)
[c0000000494c37f0] [c000000000094bdc] .emulate_loadstore+0x167c/0x17b0
[c0000000494c3900] [c000000000094f6c] .emulate_step+0x25c/0x5bc
[c0000000494c39b0] [c000000000ec90dc] .test_lxvd2x_stxvd2x+0x64/0x154
[c0000000494c3b90] [c000000000ec9204] .test_emulate_step+0x38/0x4c
[c0000000494c3c00] [c00000000000de2c] .do_one_initcall+0x5c/0x2c0
[c0000000494c3cd0] [c000000000eb4bf8] .kernel_init_freeable+0x314/0x4cc
[c0000000494c3db0] [c00000000000e1e4] .kernel_init+0x24/0x160
[c0000000494c3e30] [c00000000000bc24] .ret_from_kernel_thread+0x58/0xb4
With fix:
...
emulate_step_test: stvx : PASS
emulate_step_test: lxvd2x : FAIL
emulate_step_test: stxvd2x : FAIL
Fixes: https://github.com/linuxppc/linux/issues/148
Reported-by: Michael Ellerman <mpe@...erman.id.au>
Signed-off-by: Ravi Bangoria <ravi.bangoria@...ux.ibm.com>
---
Note: VSX was first introduced on P6. But there are many VSX instructions
which were introduced in later versions. So ideally, analyse_instr()
should check for ISA version along with opcode.
arch/powerpc/lib/sstep.c | 10 ++++++++++
1 file changed, 10 insertions(+)
diff --git a/arch/powerpc/lib/sstep.c b/arch/powerpc/lib/sstep.c
index 34d68f1b1b40..470942f9c567 100644
--- a/arch/powerpc/lib/sstep.c
+++ b/arch/powerpc/lib/sstep.c
@@ -1173,6 +1173,7 @@ int analyse_instr(struct instruction_op *op, const struct pt_regs *regs,
unsigned long int val, val2;
unsigned int mb, me, sh;
long ival;
+ int type;
op->type = COMPUTE;
@@ -2544,6 +2545,15 @@ int analyse_instr(struct instruction_op *op, const struct pt_regs *regs,
#endif /* __powerpc64__ */
}
+
+#ifdef CONFIG_VSX
+ type = op->type & INSTR_TYPE_MASK;
+ if ((type == LOAD_VSX || type == STORE_VSX) &&
+ !cpu_has_feature(CPU_FTR_VSX)) {
+ return -1;
+ }
+#endif /* CONFIG_VSX */
+
return 0;
logical_done:
--
2.16.2
Powered by blists - more mailing lists