From: Peter Zijlstra <peterz@infradead.org>
To: mingo@kernel.org, will@kernel.org, tglx@linutronix.de
Cc: x86@kernel.org, linux-kernel@vger.kernel.org,
a.darwish@linutronix.de, rostedt@goodmis.org,
bigeasy@linutronix.de, peterz@infradead.org
Subject: [PATCH v2 6/6] x86/entry: Fix NMI vs IRQ state tracking
Date: Thu, 28 May 2020 16:05:41 +0200 [thread overview]
Message-ID: <20200528140947.035164777@infradead.org> (raw)
In-Reply-To: 20200528140535.206916549@infradead.org
While the nmi_enter() users did
trace_hardirqs_{off_prepare,on_finish}() there was no matching
lockdep_hardirqs_*() calls to complete the picture.
Introduce idtentry_{enter,exit}_nmi() to enable proper IRQ state
tracking across the NMIs.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
---
arch/x86/entry/common.c | 42 ++++++++++++++++++++++++++++++++++++----
arch/x86/include/asm/idtentry.h | 3 ++
arch/x86/kernel/nmi.c | 9 +++-----
arch/x86/kernel/traps.c | 20 ++++---------------
include/linux/hardirq.h | 28 ++++++++++++++++++--------
5 files changed, 69 insertions(+), 33 deletions(-)
--- a/arch/x86/entry/common.c
+++ b/arch/x86/entry/common.c
@@ -550,7 +550,7 @@ SYSCALL_DEFINE0(ni_syscall)
* The return value must be fed into the rcu_exit argument of
* idtentry_exit_cond_rcu().
*/
-bool noinstr idtentry_enter_cond_rcu(struct pt_regs *regs)
+noinstr bool idtentry_enter_cond_rcu(struct pt_regs *regs)
{
if (user_mode(regs)) {
enter_from_user_mode();
@@ -619,7 +619,7 @@ static void idtentry_exit_cond_resched(s
* Counterpart to idtentry_enter_cond_rcu(). The return value of the entry
* function must be fed into the @rcu_exit argument.
*/
-void noinstr idtentry_exit_cond_rcu(struct pt_regs *regs, bool rcu_exit)
+noinstr void idtentry_exit_cond_rcu(struct pt_regs *regs, bool rcu_exit)
{
lockdep_assert_irqs_disabled();
@@ -663,7 +663,7 @@ void noinstr idtentry_exit_cond_rcu(stru
* Invokes enter_from_user_mode() to establish the proper context for
* NOHZ_FULL. Otherwise scheduling on exit would not be possible.
*/
-void noinstr idtentry_enter_user(struct pt_regs *regs)
+noinstr void idtentry_enter_user(struct pt_regs *regs)
{
enter_from_user_mode();
}
@@ -680,13 +680,47 @@ void noinstr idtentry_enter_user(struct
*
* Counterpart to idtentry_enter_user().
*/
-void noinstr idtentry_exit_user(struct pt_regs *regs)
+noinstr void idtentry_exit_user(struct pt_regs *regs)
{
lockdep_assert_irqs_disabled();
prepare_exit_to_usermode(regs);
}
+noinstr bool idtentry_enter_nmi(struct pt_regs *regs)
+{
+ bool irq_state = lockdep_hardirqs_enabled();
+
+ __nmi_enter();
+ lockdep_hardirqs_off(CALLER_ADDR0);
+ lockdep_hardirq_enter();
+ rcu_nmi_enter();
+
+ instrumentation_begin();
+ trace_hardirqs_off_finish();
+ ftrace_nmi_enter();
+ instrumentation_end();
+
+ return irq_state;
+}
+
+noinstr void idtentry_exit_nmi(struct pt_regs *regs, bool restore)
+{
+ instrumentation_begin();
+ ftrace_nmi_exit();
+ if (restore) {
+ trace_hardirqs_on_prepare();
+ lockdep_hardirqs_on_prepare(CALLER_ADDR0);
+ }
+ instrumentation_end();
+
+ rcu_nmi_exit();
+ lockdep_hardirq_exit();
+ if (restore)
+ lockdep_hardirqs_on(CALLER_ADDR0);
+ __nmi_exit();
+}
+
#ifdef CONFIG_XEN_PV
#ifndef CONFIG_PREEMPTION
/*
--- a/arch/x86/include/asm/idtentry.h
+++ b/arch/x86/include/asm/idtentry.h
@@ -16,6 +16,9 @@ void idtentry_exit_user(struct pt_regs *
bool idtentry_enter_cond_rcu(struct pt_regs *regs);
void idtentry_exit_cond_rcu(struct pt_regs *regs, bool rcu_exit);
+bool idtentry_enter_nmi(struct pt_regs *regs);
+void idtentry_exit_nmi(struct pt_regs *regs, bool irq_state);
+
/**
* DECLARE_IDTENTRY - Declare functions for simple IDT entry points
* No error code pushed by hardware
--- a/arch/x86/kernel/nmi.c
+++ b/arch/x86/kernel/nmi.c
@@ -330,7 +330,6 @@ static noinstr void default_do_nmi(struc
__this_cpu_write(last_nmi_rip, regs->ip);
instrumentation_begin();
- trace_hardirqs_off_finish();
handled = nmi_handle(NMI_LOCAL, regs);
__this_cpu_add(nmi_stats.normal, handled);
@@ -417,8 +416,6 @@ static noinstr void default_do_nmi(struc
unknown_nmi_error(reason, regs);
out:
- if (regs->flags & X86_EFLAGS_IF)
- trace_hardirqs_on_prepare();
instrumentation_end();
}
@@ -511,6 +508,8 @@ static noinstr bool is_debug_stack(unsig
DEFINE_IDTENTRY_NMI(exc_nmi)
{
+ bool irq_state;
+
if (IS_ENABLED(CONFIG_SMP) && cpu_is_offline(smp_processor_id()))
return;
@@ -535,14 +534,14 @@ DEFINE_IDTENTRY_NMI(exc_nmi)
}
#endif
- nmi_enter();
+ irq_state = idtentry_enter_nmi(regs);
inc_irq_stat(__nmi_count);
if (!ignore_nmis)
default_do_nmi(regs);
- nmi_exit();
+ idtentry_exit_nmi(regs, irq_state);
#ifdef CONFIG_X86_64
if (unlikely(this_cpu_read(update_debug_stack))) {
--- a/arch/x86/kernel/traps.c
+++ b/arch/x86/kernel/traps.c
@@ -387,7 +387,7 @@ DEFINE_IDTENTRY_DF(exc_double_fault)
}
#endif
- nmi_enter();
+ idtentry_enter_nmi(regs);
instrumentation_begin();
notify_die(DIE_TRAP, str, regs, error_code, X86_TRAP_DF, SIGSEGV);
@@ -632,15 +632,12 @@ DEFINE_IDTENTRY_RAW(exc_int3)
instrumentation_end();
idtentry_exit_user(regs);
} else {
- nmi_enter();
+ bool irq_state = idtentry_enter_nmi(regs);
instrumentation_begin();
- trace_hardirqs_off_finish();
if (!do_int3(regs))
die("int3", regs, 0);
- if (regs->flags & X86_EFLAGS_IF)
- trace_hardirqs_on_prepare();
instrumentation_end();
- nmi_exit();
+ idtentry_exit_nmi(regs, irq_state);
}
}
@@ -852,10 +849,7 @@ static void noinstr handle_debug(struct
static __always_inline void exc_debug_kernel(struct pt_regs *regs,
unsigned long dr6)
{
- nmi_enter();
- instrumentation_begin();
- trace_hardirqs_off_finish();
- instrumentation_end();
+ bool irq_state = idtentry_enter_nmi(regs);
/*
* The SDM says "The processor clears the BTF flag when it
@@ -878,11 +872,7 @@ static __always_inline void exc_debug_ke
if (dr6)
handle_debug(regs, dr6, false);
- instrumentation_begin();
- if (regs->flags & X86_EFLAGS_IF)
- trace_hardirqs_on_prepare();
- instrumentation_end();
- nmi_exit();
+ idtentry_exit_nmi(regs, irq_state);
}
static __always_inline void exc_debug_user(struct pt_regs *regs,
--- a/include/linux/hardirq.h
+++ b/include/linux/hardirq.h
@@ -111,32 +111,42 @@ extern void rcu_nmi_exit(void);
/*
* nmi_enter() can nest up to 15 times; see NMI_BITS.
*/
-#define nmi_enter() \
+#define __nmi_enter() \
do { \
+ lockdep_off(); \
arch_nmi_enter(); \
printk_nmi_enter(); \
- lockdep_off(); \
BUG_ON(in_nmi() == NMI_MASK); \
__preempt_count_add(NMI_OFFSET + HARDIRQ_OFFSET); \
- rcu_nmi_enter(); \
+ } while (0)
+
+#define nmi_enter() \
+ do { \
+ __nmi_enter(); \
lockdep_hardirq_enter(); \
+ rcu_nmi_enter(); \
instrumentation_begin(); \
ftrace_nmi_enter(); \
instrumentation_end(); \
} while (0)
+#define __nmi_exit() \
+ do { \
+ BUG_ON(!in_nmi()); \
+ __preempt_count_sub(NMI_OFFSET + HARDIRQ_OFFSET); \
+ printk_nmi_exit(); \
+ arch_nmi_exit(); \
+ lockdep_on(); \
+ } while (0)
+
#define nmi_exit() \
do { \
instrumentation_begin(); \
ftrace_nmi_exit(); \
instrumentation_end(); \
- lockdep_hardirq_exit(); \
rcu_nmi_exit(); \
- BUG_ON(!in_nmi()); \
- __preempt_count_sub(NMI_OFFSET + HARDIRQ_OFFSET); \
- lockdep_on(); \
- printk_nmi_exit(); \
- arch_nmi_exit(); \
+ lockdep_hardirq_exit(); \
+ __nmi_exit(); \
} while (0)
#endif /* LINUX_HARDIRQ_H */
prev parent reply other threads:[~2020-05-28 14:17 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-05-28 14:05 [PATCH v2 0/6] x86/entry,lockdep: Improve IRQ state tracking Peter Zijlstra
2020-05-28 14:05 ` [PATCH v2 1/6] x86/entry: Clarify irq_{enter,exit}_rcu() Peter Zijlstra
2020-05-28 14:05 ` [PATCH v2 2/6] x86/entry: Rename trace_hardirqs_off_prepare() Peter Zijlstra
2020-05-28 14:05 ` [PATCH v2 3/6] lockdep: Change hardirq{s_enabled,_context} to per-cpu variables Peter Zijlstra
2020-05-29 16:16 ` Peter Zijlstra
2020-05-29 19:38 ` Peter Zijlstra
2020-05-28 14:05 ` [PATCH v2 4/6] lockdep: Remove lockdep_hardirq{s_enabled,_context}() argument Peter Zijlstra
2020-05-28 14:05 ` [PATCH v2 5/6] lockdep: Prepare for NMI IRQ state tracking Peter Zijlstra
2020-05-28 14:05 ` Peter Zijlstra [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200528140947.035164777@infradead.org \
--to=peterz@infradead.org \
--cc=a.darwish@linutronix.de \
--cc=bigeasy@linutronix.de \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@kernel.org \
--cc=rostedt@goodmis.org \
--cc=tglx@linutronix.de \
--cc=will@kernel.org \
--cc=x86@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).