coresight: tmc-etf: Do not call smp_processor_id from preemptible
authorSuzuki K Poulose <suzuki.poulose@arm.com>
Thu, 20 Jun 2019 22:12:35 +0000 (16:12 -0600)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 29 Jan 2020 09:24:40 +0000 (10:24 +0100)
commit 024c1fd9dbcc1d8a847f1311f999d35783921b7f upstream.

During a perf session we try to allocate buffers on the "node" associated
with the CPU the event is bound to. If it is not bound to a CPU, we
use the current CPU node, using smp_processor_id(). However this is unsafe
in a pre-emptible context and could generate the splats as below :

 BUG: using smp_processor_id() in preemptible [00000000] code: perf/2544
 caller is tmc_alloc_etf_buffer+0x5c/0x60
 CPU: 2 PID: 2544 Comm: perf Not tainted 5.1.0-rc6-147786-g116841e #344
 Hardware name: ARM LTD ARM Juno Development Platform/ARM Juno Development Platform, BIOS EDK II Feb  1 2019
 Call trace:
  dump_backtrace+0x0/0x150
  show_stack+0x14/0x20
  dump_stack+0x9c/0xc4
  debug_smp_processor_id+0x10c/0x110
  tmc_alloc_etf_buffer+0x5c/0x60
  etm_setup_aux+0x1c4/0x230
  rb_alloc_aux+0x1b8/0x2b8
  perf_mmap+0x35c/0x478
  mmap_region+0x34c/0x4f0
  do_mmap+0x2d8/0x418
  vm_mmap_pgoff+0xd0/0xf8
  ksys_mmap_pgoff+0x88/0xf8
  __arm64_sys_mmap+0x28/0x38
  el0_svc_handler+0xd8/0x138
  el0_svc+0x8/0xc

Use NUMA_NO_NODE hint instead of using the current node for events
not bound to CPUs.

Fixes: 2e499bbc1a929ac ("coresight: tmc: implementing TMC-ETF AUX space API")
Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Cc: stable <stable@vger.kernel.org> # 4.7+
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Link: https://lore.kernel.org/r/20190620221237.3536-4-mathieu.poirier@linaro.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/hwtracing/coresight/coresight-tmc-etf.c

index 14df4e34c21cf49857f037f080f729d2c8e2db72..03ce12ad431756e81d11cb84bfed4a6d083bdf05 100644 (file)
@@ -292,9 +292,7 @@ static void *tmc_alloc_etf_buffer(struct coresight_device *csdev, int cpu,
        int node;
        struct cs_buffers *buf;
 
-       if (cpu == -1)
-               cpu = smp_processor_id();
-       node = cpu_to_node(cpu);
+       node = (cpu == -1) ? NUMA_NO_NODE : cpu_to_node(cpu);
 
        /* Allocate memory structure for interaction with Perf */
        buf = kzalloc_node(sizeof(struct cs_buffers), GFP_KERNEL, node);