From: Steven Rostedt <rostedt@goodmis.org>
To: Borislav Petkov <bp@alien8.de>
Cc: LKML <linux-kernel@vger.kernel.org>,
Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
Ingo Molnar <mingo@kernel.org>,
Thomas Gleixner <tglx@linutronix.de>,
Linus Torvalds <torvalds@linux-foundation.org>,
Peter Zijlstra <peterz@infradead.org>
Subject: [PATCH] lockdep: Quiet gcc about dangerous __builtin_return_address() operations
Date: Thu, 8 Sep 2016 12:34:33 -0400 [thread overview]
Message-ID: <20160908123433.2cf93b96@gandalf.local.home> (raw)
[
Boris, does this quiet gcc for you?
I haven't fully tested this yet, as I still don't have a compiler
that does the warning.
]
Gcc's new warnings about __builtin_return_address(n) operations with
n > 0 is popping up around the kernel. The operation is dangerous, and
the warning is "good to know". But there's instances that we use
__builtin_return_address(n) with n > 0 and are aware of the issues,
and work around them. And its used mostly for tracing and debugging. In
these cases, the warning becomes a distraction and is not helpful.
To get better lock issue traces, a function like get_lock_parent_ip()
uses __builtin_return_address() to find the caller of the lock, and
skip over the internal callers of the lock itself. Currently it is only
used in the kernel/ directory and only if certain configs are enabled.
Create a new config called CONFIG_USING_GET_LOCK_PARENT_IP that gets
selected when another config relies on get_lock_parent_ip(), and this
will now enable the function get_lock_parent_ip(), otherwise it wont be
defined. It will also disable the frame-address warnings from gcc in
the kernel directory.
Reported-by: Borislav Petkov <bp@alien8.de>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
---
diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h
index 1e2b316d6693..7862c59ac725 100644
--- a/include/linux/ftrace.h
+++ b/include/linux/ftrace.h
@@ -714,6 +714,7 @@ static inline void __ftrace_enabled_restore(int enabled)
#define CALLER_ADDR5 ((unsigned long)ftrace_return_address(5))
#define CALLER_ADDR6 ((unsigned long)ftrace_return_address(6))
+#ifdef CONFIG_USING_GET_LOCK_PARENT_IP
static inline unsigned long get_lock_parent_ip(void)
{
unsigned long addr = CALLER_ADDR0;
@@ -725,6 +726,7 @@ static inline unsigned long get_lock_parent_ip(void)
return addr;
return CALLER_ADDR2;
}
+#endif
#ifdef CONFIG_IRQSOFF_TRACER
extern void time_hardirqs_on(unsigned long a0, unsigned long a1);
diff --git a/kernel/Makefile b/kernel/Makefile
index e2ec54e2b952..bff8214bf5f6 100644
--- a/kernel/Makefile
+++ b/kernel/Makefile
@@ -11,6 +11,13 @@ obj-y = fork.o exec_domain.o panic.o \
notifier.o ksysfs.o cred.o reboot.o \
async.o range.o smpboot.o
+# Tracing may do some dangerous __builtin_return_address() operations
+# We know they are dangerous, we don't need gcc telling us that.
+ifdef CONFIG_USING_GET_LOCK_PARENT_IP
+FRAME_CFLAGS := $(call cc-disable-warning,frame-address)
+KBUILD_CFLAGS += $(FRAME_CFLAGS)
+endif
+
obj-$(CONFIG_MULTIUSER) += groups.o
ifdef CONFIG_FUNCTION_TRACER
diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig
index 72c07c2ffd79..f02a02fdc3ed 100644
--- a/kernel/trace/Kconfig
+++ b/kernel/trace/Kconfig
@@ -197,6 +197,7 @@ config PREEMPT_TRACER
select RING_BUFFER_ALLOW_SWAP
select TRACER_SNAPSHOT
select TRACER_SNAPSHOT_PER_CPU_SWAP
+ select USING_GET_LOCK_PARENT_IP
help
This option measures the time spent in preemption-off critical
sections, with microsecond accuracy.
diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index 2307d7c89dac..9a2bd7af64df 100644
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -977,6 +977,7 @@ config TIMER_STATS
config DEBUG_PREEMPT
bool "Debug preemptible kernel"
depends on DEBUG_KERNEL && PREEMPT && TRACE_IRQFLAGS_SUPPORT
+ select USING_GET_LOCK_PARENT_IP
default y
help
If you say Y here then the kernel will use a debug variant of the
@@ -1159,8 +1160,17 @@ config LOCK_TORTURE_TEST
endmenu # lock debugging
+config USING_GET_LOCK_PARENT_IP
+ bool
+ help
+ Enables the use of the function get_lock_parent_ip() that
+ will use __builtin_return_address(n) with n > 0 causing
+ some gcc warnings. When this is selected, those warnings
+ will be suppressed.
+
config TRACE_IRQFLAGS
bool
+ select USING_GET_LOCK_PARENT_IP
help
Enables hooks to interrupt enabling and disabling for
either tracing or lock debugging.
next reply other threads:[~2016-09-08 16:34 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-09-08 16:34 Steven Rostedt [this message]
2016-09-08 18:33 ` Borislav Petkov
2016-09-08 18:38 ` Steven Rostedt
2016-09-15 13:54 ` Steven Rostedt
2016-09-15 14:08 ` Peter Zijlstra
2016-09-15 15:17 ` Steven Rostedt
2016-10-06 13:35 ` [tip:locking/urgent] locking/lockdep: Quiet GCC " tip-bot for Steven Rostedt
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=20160908123433.2cf93b96@gandalf.local.home \
--to=rostedt@goodmis.org \
--cc=bigeasy@linutronix.de \
--cc=bp@alien8.de \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@kernel.org \
--cc=peterz@infradead.org \
--cc=tglx@linutronix.de \
--cc=torvalds@linux-foundation.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
all inboxes | Powered by JetHome®