Re: [PATCH] x86/oprofile/ppro: Do not use __this_cpu* accessors in preemptible context

[Borislav Petkov]

On Wed, Nov 08, 2017 at 11:48:28AM +0100, Thomas Gleixner wrote:
> AFAICT, ppro_setup_ctrs() has the same issue

Here's an updated patch:

---
From: Borislav Petkov <bp@xxxxxxx>
Date: Tue, 7 Nov 2017 18:20:50 +0100
Subject: [PATCH] x86/oprofile/ppro: Do not use __this_cpu* accessors in preemptible context

The warning below says it all. And we should not have used those
accessors in the first place: it is PPro so no mixed silicon revisions
and thus can simply use boot_cpu_data.

  BUG: using __this_cpu_read() in preemptible [00000000] code: swapper/0/1
  caller is __this_cpu_preempt_check
  CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.14.0-rc8 #4
  Call Trace:
   dump_stack
   check_preemption_disabled
   ? do_early_param
   __this_cpu_preempt_check
   arch_perfmon_init
   op_nmi_init
   ? alloc_pci_root_info
   oprofile_arch_init
   oprofile_init
   do_one_initcall
   ...


Fix the other callsite in ppro_setup_ctrs() too, while at it.

Reported-and-tested-by: Fengguang Wu <fengguang.wu@xxxxxxxxx>
Signed-off-by: Borislav Petkov <bp@xxxxxxx>
Fix-creation-mandated-by: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
Cc: Robert Richter <rric@xxxxxxxxxx>
Cc: x86@xxxxxxxxxx
---
 arch/x86/oprofile/op_model_ppro.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/arch/x86/oprofile/op_model_ppro.c b/arch/x86/oprofile/op_model_ppro.c
index 350f7096baac..488b2efb24b9 100644
--- a/arch/x86/oprofile/op_model_ppro.c
+++ b/arch/x86/oprofile/op_model_ppro.c
@@ -84,8 +84,8 @@ static void ppro_setup_ctrs(struct op_x86_model_spec const *model,
 		 * counter width:
 		 */
 		if (!(eax.split.version_id == 0 &&
-			__this_cpu_read(cpu_info.x86) == 6 &&
-				__this_cpu_read(cpu_info.x86_model) == 15)) {
+		      boot_cpu_data.x86 == 6 &&
+		      boot_cpu_data.x86_model == 15)) {
 
 			if (counter_width < eax.split.bit_width)
 				counter_width = eax.split.bit_width;
@@ -212,8 +212,9 @@ static void arch_perfmon_setup_counters(void)
 	eax.full = cpuid_eax(0xa);
 
 	/* Workaround for BIOS bugs in 6/15. Taken from perfmon2 */
-	if (eax.split.version_id == 0 && __this_cpu_read(cpu_info.x86) == 6 &&
-		__this_cpu_read(cpu_info.x86_model) == 15) {
+	if (eax.split.version_id == 0 &&
+	    boot_cpu_data.x86 == 6 &&
+	    boot_cpu_data.x86_model == 15) {
 		eax.split.version_id = 2;
 		eax.split.num_counters = 2;
 		eax.split.bit_width = 40;
-- 
2.13.0

SUSE Linux GmbH, GF: Felix ImendÃrffer, Jane Smithard, Graham Norton, HRB 21284 (AG NÃrnberg)
--