[PATCH] tracing: Have all levels of checks prevent recursion

[Steven Rostedt]

On Fri, 15 Oct 2021 13:35:04 -0400
Steven Rostedt <rostedt at goodmis.org> wrote:

> On Fri, 15 Oct 2021 18:17:02 +0200
> Peter Zijlstra <peterz at infradead.org> wrote:
> 
> > On Fri, Oct 15, 2021 at 11:00:35AM -0400, Steven Rostedt wrote:  
> > > From: "Steven Rostedt (VMware)" <rostedt at goodmis.org>
> > > 
> > > While writing an email explaining the "bit = 0" logic for a discussion on    
> >   
> > >  	bit = trace_get_context_bit() + start;    
> > 
> > While there, you were also going to update that function to match/use
> > get_recursion_context(). Because your version is still branch hell.  
> 
> That would probably be a separate patch. This is just a fix to a design
> flaw, changing the context tests would be performance enhancement.
> 
> Thanks for the reminder (as it was on my todo list to update that code).

Something like this:

From: "Steven Rostedt (VMware)" <rostedt at goodmis.org>
Subject: [PATCH] tracing: Reuse logic from perf's get_recursion_context()

Instead of having branches that adds noise to the branch prediction, use
the addition logic to set the bit for the level of interrupt context that
the state is currently in. This copies the logic from perf's
get_recursion_context() function.

Link: https://lore.kernel.org/all/20211015161702.GF174703@worktop.programming.kicks-ass.net/

Suggested-by: Peter Zijlstra <peterz at infradead.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt at goodmis.org>
---
 include/linux/trace_recursion.h | 11 ++++++-----
 kernel/trace/ring_buffer.c      | 12 ++++++------
 2 files changed, 12 insertions(+), 11 deletions(-)

diff --git a/include/linux/trace_recursion.h b/include/linux/trace_recursion.h
index 168fdf07419a..41f5bfd9e93f 100644
--- a/include/linux/trace_recursion.h
+++ b/include/linux/trace_recursion.h
@@ -119,12 +119,13 @@ enum {
 static __always_inline int trace_get_context_bit(void)
 {
 	unsigned long pc = preempt_count();
+	unsigned char bit = 0;
 
-	if (!(pc & (NMI_MASK | HARDIRQ_MASK | SOFTIRQ_OFFSET)))
-		return TRACE_CTX_NORMAL;
-	else
-		return pc & NMI_MASK ? TRACE_CTX_NMI :
-			pc & HARDIRQ_MASK ? TRACE_CTX_IRQ : TRACE_CTX_SOFTIRQ;
+	bit += !!(pc & (NMI_MASK));
+	bit += !!(pc & (NMI_MASK | HARDIRQ_MASK));
+	bit += !!(pc & (NMI_MASK | HARDIRQ_MASK | SOFTIRQ_OFFSET));
+
+	return TRACE_CTX_NORMAL - bit;
 }
 
 #ifdef CONFIG_FTRACE_RECORD_RECURSION
diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
index c5a3fbf19617..15d4380006e3 100644
--- a/kernel/trace/ring_buffer.c
+++ b/kernel/trace/ring_buffer.c
@@ -3168,13 +3168,13 @@ trace_recursive_lock(struct ring_buffer_per_cpu *cpu_buffer)
 {
 	unsigned int val = cpu_buffer->current_context;
 	unsigned long pc = preempt_count();
-	int bit;
+	int bit = 0;
 
-	if (!(pc & (NMI_MASK | HARDIRQ_MASK | SOFTIRQ_OFFSET)))
-		bit = RB_CTX_NORMAL;
-	else
-		bit = pc & NMI_MASK ? RB_CTX_NMI :
-			pc & HARDIRQ_MASK ? RB_CTX_IRQ : RB_CTX_SOFTIRQ;
+	bit += !!(pc & (NMI_MASK));
+	bit += !!(pc & (NMI_MASK | HARDIRQ_MASK));
+	bit += !!(pc & (NMI_MASK | HARDIRQ_MASK | SOFTIRQ_OFFSET));
+
+	bit = RB_CTX_NORMAL - bit;
 
 	if (unlikely(val & (1 << (bit + cpu_buffer->nest)))) {
 		/*
-- 
2.31.1