4.9-stable review patch. If anyone has any objections, please let me know.
------------------
From: Ravi Bangoria ravi.bangoria@linux.vnet.ibm.com
commit f41d84dddc66b164ac16acf3f584c276146f1c48 upstream.
It's theoretically possible that branch instructions recorded in BHRB (Branch History Rolling Buffer) entries have already been unmapped before they are processed by the kernel. Hence, trying to dereference such memory location will result in a crash. eg:
Unable to handle kernel paging request for data at address 0xd000000019c41764 Faulting instruction address: 0xc000000000084a14 NIP [c000000000084a14] branch_target+0x4/0x70 LR [c0000000000eb828] record_and_restart+0x568/0x5c0 Call Trace: [c0000000000eb3b4] record_and_restart+0xf4/0x5c0 (unreliable) [c0000000000ec378] perf_event_interrupt+0x298/0x460 [c000000000027964] performance_monitor_exception+0x54/0x70 [c000000000009ba4] performance_monitor_common+0x114/0x120
Fix it by deferefencing the addresses safely.
Fixes: 691231846ceb ("powerpc/perf: Fix setting of "to" addresses for BHRB") Suggested-by: Naveen N. Rao naveen.n.rao@linux.vnet.ibm.com Signed-off-by: Ravi Bangoria ravi.bangoria@linux.vnet.ibm.com Reviewed-by: Naveen N. Rao naveen.n.rao@linux.vnet.ibm.com [mpe: Use probe_kernel_read() which is clearer, tweak change log] Signed-off-by: Michael Ellerman mpe@ellerman.id.au Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org
--- arch/powerpc/perf/core-book3s.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)
--- a/arch/powerpc/perf/core-book3s.c +++ b/arch/powerpc/perf/core-book3s.c @@ -401,8 +401,12 @@ static __u64 power_pmu_bhrb_to(u64 addr) int ret; __u64 target;
- if (is_kernel_addr(addr)) - return branch_target((unsigned int *)addr); + if (is_kernel_addr(addr)) { + if (probe_kernel_read(&instr, (void *)addr, sizeof(instr))) + return 0; + + return branch_target(&instr); + }
/* Userspace: need copy instruction here then translate it */ pagefault_disable();