From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To: linux-kernel@vger.kernel.org
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
stable@vger.kernel.org, Steven Rostedt <rostedt@goodmis.org>
Subject: [ 16/46] tracing: Fix stack tracer with fentry use
Date: Mon, 6 May 2013 15:58:22 -0700 [thread overview]
Message-ID: <20130506225822.382333258@linuxfoundation.org> (raw)
In-Reply-To: <20130506225820.628893970@linuxfoundation.org>
3.0-stable review patch. If anyone has any objections, please let me know.
------------------
From: "Steven Rostedt (Red Hat)" <rostedt@goodmis.org>
commit d4ecbfc49b4b1d4b597fb5ba9e4fa25d62f105c5 upstream.
When gcc 4.6 on x86 is used, the function tracer will use the new
option -mfentry which does a call to "fentry" at every function
instead of "mcount". The significance of this is that fentry is
called as the first operation of the function instead of the mcount
usage of being called after the stack.
This causes the stack tracer to show some bogus results for the size
of the last function traced, as well as showing "ftrace_call" instead
of the function. This is due to the stack frame not being set up
by the function that is about to be traced.
# cat stack_trace
Depth Size Location (48 entries)
----- ---- --------
0) 4824 216 ftrace_call+0x5/0x2f
1) 4608 112 ____cache_alloc+0xb7/0x22d
2) 4496 80 kmem_cache_alloc+0x63/0x12f
The 216 size for ftrace_call includes both the ftrace_call stack
(which includes the saving of registers it does), as well as the
stack size of the parent.
To fix this, if CC_USING_FENTRY is defined, then the stack_tracer
will reserve the first item in stack_dump_trace[] array when
calling save_stack_trace(), and it will fill it in with the parent ip.
Then the code will look for the parent pointer on the stack and
give the real size of the parent's stack pointer:
# cat stack_trace
Depth Size Location (14 entries)
----- ---- --------
0) 2640 48 update_group_power+0x26/0x187
1) 2592 224 update_sd_lb_stats+0x2a5/0x4ac
2) 2368 160 find_busiest_group+0x31/0x1f1
3) 2208 256 load_balance+0xd9/0x662
I'm Cc'ing stable, although it's not urgent, as it only shows bogus
size for item #0, the rest of the trace is legit. It should still be
corrected in previous stable releases.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
kernel/trace/trace_stack.c | 33 +++++++++++++++++++++++++++++----
1 file changed, 29 insertions(+), 4 deletions(-)
--- a/kernel/trace/trace_stack.c
+++ b/kernel/trace/trace_stack.c
@@ -17,13 +17,27 @@
#define STACK_TRACE_ENTRIES 500
+/*
+ * If fentry is used, then the function being traced will
+ * jump to fentry directly before it sets up its stack frame.
+ * We need to ignore that one and record the parent. Since
+ * the stack frame for the traced function wasn't set up yet,
+ * the stack_trace wont see the parent. That needs to be added
+ * manually to stack_dump_trace[] as the first element.
+ */
+#ifdef CC_USING_FENTRY
+# define add_func 1
+#else
+# define add_func 0
+#endif
+
static unsigned long stack_dump_trace[STACK_TRACE_ENTRIES+1] =
{ [0 ... (STACK_TRACE_ENTRIES)] = ULONG_MAX };
static unsigned stack_dump_index[STACK_TRACE_ENTRIES];
static struct stack_trace max_stack_trace = {
- .max_entries = STACK_TRACE_ENTRIES,
- .entries = stack_dump_trace,
+ .max_entries = STACK_TRACE_ENTRIES - add_func,
+ .entries = &stack_dump_trace[add_func],
};
static unsigned long max_stack_size;
@@ -38,7 +52,7 @@ int stack_tracer_enabled;
static int last_stack_tracer_enabled;
static inline void
-check_stack(unsigned long *stack)
+check_stack(unsigned long ip, unsigned long *stack)
{
unsigned long this_size, flags;
unsigned long *p, *top, *start;
@@ -69,6 +83,17 @@ check_stack(unsigned long *stack)
save_stack_trace(&max_stack_trace);
/*
+ * When fentry is used, the traced function does not get
+ * its stack frame set up, and we lose the parent.
+ * Add that one in manally. We set up save_stack_trace()
+ * to not touch the first element in this case.
+ */
+ if (add_func) {
+ stack_dump_trace[0] = ip;
+ max_stack_trace.nr_entries++;
+ }
+
+ /*
* Now find where in the stack these are.
*/
i = 0;
@@ -124,7 +149,7 @@ stack_trace_call(unsigned long ip, unsig
if (per_cpu(trace_active, cpu)++ != 0)
goto out;
- check_stack(&stack);
+ check_stack(parent_ip, &stack);
out:
per_cpu(trace_active, cpu)--;
next prev parent reply other threads:[~2013-05-06 22:58 UTC|newest]
Thread overview: 49+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-05-06 22:58 [ 00/46] 3.0.77-stable review Greg Kroah-Hartman
2013-05-06 22:58 ` [ 01/46] powerpc: Add isync to copy_and_flush Greg Kroah-Hartman
2013-05-06 22:58 ` [ 02/46] powerpc/spufs: Initialise inode->i_ino in spufs_new_inode() Greg Kroah-Hartman
2013-05-06 22:58 ` [ 03/46] USB: serial: option: Added support Olivetti Olicard 145 Greg Kroah-Hartman
2013-05-06 22:58 ` [ 04/46] USB: option: add a D-Link DWM-156 variant Greg Kroah-Hartman
2013-05-06 22:58 ` [ 05/46] usb/misc/appledisplay: Add 24" LED Cinema display Greg Kroah-Hartman
2013-05-06 22:58 ` [ 06/46] USB: add ftdi_sio USB ID for GDM Boost V1.x Greg Kroah-Hartman
2013-05-06 22:58 ` [ 07/46] USB: ftdi_sio: correct ST Micro Connect Lite PIDs Greg Kroah-Hartman
2013-05-06 22:58 ` [ 08/46] usbfs: Always allow ctrl requests with USB_RECIP_ENDPOINT on the ctrl ep Greg Kroah-Hartman
2013-05-06 22:58 ` [ 09/46] usb-storage: CY7C68300A chips do not support Cypress ATACB Greg Kroah-Hartman
2013-05-06 22:58 ` [ 10/46] s390/memory hotplug: prevent offline of active memory increments Greg Kroah-Hartman
2013-05-06 22:58 ` [ 11/46] xen/time: Fix kasprintf splat when allocating timer%d IRQ line Greg Kroah-Hartman
2013-05-06 22:58 ` [ 12/46] serial_core.c: add put_device() after device_find_child() Greg Kroah-Hartman
2013-05-06 22:58 ` [ 13/46] tty: fix up atime/mtime mess, take three Greg Kroah-Hartman
2013-05-06 22:58 ` [ 14/46] fbcon: when font is freed, clear also vc_font.data Greg Kroah-Hartman
2013-05-06 22:58 ` [ 15/46] tracing: Use stack of calling function for stack tracer Greg Kroah-Hartman
2013-05-06 22:58 ` Greg Kroah-Hartman [this message]
2013-05-06 22:58 ` [ 17/46] tracing: Remove most or all of stack tracer stack size from stack_max_size Greg Kroah-Hartman
2013-05-06 22:58 ` [ 18/46] tracing: Fix off-by-one on allocating stat->pages Greg Kroah-Hartman
2013-05-06 22:58 ` [ 19/46] tracing: Check return value of tracing_init_dentry() Greg Kroah-Hartman
2013-05-06 22:58 ` [ 20/46] tracing: Reset ftrace_graph_filter_enabled if count is zero Greg Kroah-Hartman
2013-05-06 22:58 ` [ 21/46] i2c: xiic: must always write 16-bit words to TX_FIFO Greg Kroah-Hartman
2013-05-06 22:58 ` [ 22/46] Fix initialization of CMCI/CMCP interrupts Greg Kroah-Hartman
2013-05-06 22:58 ` [ 23/46] PCI / ACPI: Dont query OSC support with all possible controls Greg Kroah-Hartman
2013-05-06 22:58 ` [ 24/46] Wrong asm register contraints in the futex implementation Greg Kroah-Hartman
2013-05-06 22:58 ` [ 25/46] Wrong asm register contraints in the kvm implementation Greg Kroah-Hartman
2013-05-06 22:58 ` [ 26/46] fs/fscache/stats.c: fix memory leak Greg Kroah-Hartman
2013-05-06 22:58 ` [ 27/46] ALSA: usb-audio: disable autopm for MIDI devices Greg Kroah-Hartman
2013-05-06 22:58 ` [ 28/46] ALSA: usb-audio: Fix autopm error during probing Greg Kroah-Hartman
2013-05-06 22:58 ` [ 29/46] ASoC: max98088: Fix logging of hardware revision Greg Kroah-Hartman
2013-05-06 22:58 ` [ 30/46] hrtimer: Fix ktime_add_ns() overflow on 32bit architectures Greg Kroah-Hartman
2013-05-06 22:58 ` [ 31/46] hrtimer: Add expiry time overflow check in hrtimer_interrupt Greg Kroah-Hartman
2013-05-06 22:58 ` [ 32/46] drivers/rtc/rtc-cmos.c: dont disable hpet emulation on suspend Greg Kroah-Hartman
2013-05-06 22:58 ` [ 33/46] cgroup: fix an off-by-one bug which may trigger BUG_ON() Greg Kroah-Hartman
2013-05-06 22:58 ` [ 34/46] clockevents: Set dummy handler on CPU_DEAD shutdown Greg Kroah-Hartman
2013-05-06 22:58 ` [ 35/46] LOCKD: Ensure that nlmclnt_block resets block->b_status after a server reboot Greg Kroah-Hartman
2013-05-06 22:58 ` [ 36/46] NFSv4: Handle NFS4ERR_DELAY and NFS4ERR_GRACE in nfs4_open_delegation_recall Greg Kroah-Hartman
2013-05-06 22:58 ` [ 37/46] nfsd4: dont close read-write opens too soon Greg Kroah-Hartman
2013-05-06 22:58 ` [ 38/46] nfsd: Decode and send 64bit time values Greg Kroah-Hartman
2013-05-06 22:58 ` [ 39/46] wireless: regulatory: fix channel disabling race condition Greg Kroah-Hartman
2013-05-06 22:58 ` [ 40/46] ipc: sysv shared memory limited to 8TiB Greg Kroah-Hartman
2013-05-06 22:58 ` [ 41/46] ext4: fix Kconfig documentation for CONFIG_EXT4_DEBUG Greg Kroah-Hartman
2013-05-06 22:58 ` [ 42/46] x86: Eliminate irq_mis_count counted in arch_irq_stat Greg Kroah-Hartman
2013-05-06 22:58 ` [ 43/46] mmc: core: Fix bit width test failing on old eMMC cards Greg Kroah-Hartman
2013-05-06 22:58 ` [ 44/46] mfd: adp5520: Restore mode bits on resume Greg Kroah-Hartman
2013-05-06 22:58 ` [ 45/46] s390: move dummy io_remap_pfn_range() to asm/pgtable.h Greg Kroah-Hartman
2013-05-06 22:58 ` [ 46/46] Revert :can: sja1000: fix handling on dt properties on little endian systems" Greg Kroah-Hartman
[not found] ` <CAKocOOM7rXbFbD73ZZ2zmS6SGgaytmWd-36uV+G2OVO88Zb=Og@mail.gmail.com>
2013-05-07 19:14 ` [ 00/46] 3.0.77-stable review Shuah Khan
2013-05-07 19:26 ` gregkh
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=20130506225822.382333258@linuxfoundation.org \
--to=gregkh@linuxfoundation.org \
--cc=linux-kernel@vger.kernel.org \
--cc=rostedt@goodmis.org \
--cc=stable@vger.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