Skip to contentSkip to conversation

2.6.28-rc4-mmotm1110 - you gotta be kidding me...

DISCUSSION [email protected] 4 participants 4 messages Nov 11, 2008

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

Somebody's been hitting the phunky pharmaceuticals in the last 4 days, because this ball-of-joy snuck into linux-next.patch sometime between -mmotm1106 and --mmotm1110.

Seen in a 'make silentallconfig'

Single-depth WCHAN output (SCHED_NO_NO_OMIT_FRAME_POINTER) [Y/n/?] (NEW) ?

Calculate simpler /proc/<PID>/wchan values. If this option is disabled then wchan values will recurse back to the caller function. This provides more accurate wchan values, at the expense of slightly more scheduling overhead.

If in doubt, say "Y".

So if I say 'y', is that a request to disable it, or enable it? And what exactly do I get if I vote *against* 'more accurate wchan values'? Do I get everybody having the same wchan pointing somewhere in the scheduler code, because that's where __builtin_return_address() points?

And please - a triple negative in the Kconfig variable name? This has gotta be a winner for poor taste in variable naming...

#1 lore raw
Quoted 4 lines · On Mon, 10 Nov 2008 21:55:37 -0500 [email protected] wrote:

Somebody's been hitting the phunky pharmaceuticals in the last 4 days, because this ball-of-joy snuck into linux-next.patch sometime between -mmotm1106 and --mmotm1110.

Quoted 8 lines

Seen in a 'make silentallconfig'

Single-depth WCHAN output (SCHED_NO_NO_OMIT_FRAME_POINTER) [Y/n/?] (NEW) ?

Calculate simpler /proc/<PID>/wchan values. If this option is disabled then wchan values will recurse back to the caller function. This provides more accurate wchan values, at the expense of slightly more scheduling overhead.

I got lost here.

Quoted 10 lines

If in doubt, say "Y".

So if I say 'y', is that a request to disable it, or enable it? And what exactly do I get if I vote *against* 'more accurate wchan values'? Do I get everybody having the same wchan pointing somewhere in the scheduler code, because that's where __builtin_return_address() points?

And please - a triple negative in the Kconfig variable name? This has gotta be a winner for poor taste in variable naming...

Even if that is all sorted out, how the heck is anyone to decide whether or not they need this thing?

Also, if we really really are so wishy-washy indecisive that we need to make the function optional, it should if at all possible be made runtime-configurable, not compile-time.

#2 lore raw
Quoted 29 lines · On Mon, 10 Nov 2008 19:37:13 -0800 Andrew Morton <[email protected]>…
Quoted 4 lines · On Mon, 10 Nov 2008 21:55:37 -0500 [email protected] wrote:

Somebody's been hitting the phunky pharmaceuticals in the last 4 days, because this ball-of-joy snuck into linux-next.patch sometime between -mmotm1106 and --mmotm1110.

Quoted 8 lines

Seen in a 'make silentallconfig'

Single-depth WCHAN output (SCHED_NO_NO_OMIT_FRAME_POINTER) [Y/n/?] (NEW) ?

Calculate simpler /proc/<PID>/wchan values. If this option is disabled then wchan values will recurse back to the caller function. This provides more accurate wchan values, at the expense of slightly more scheduling overhead.

I got lost here.

Quoted 10 lines

If in doubt, say "Y".

So if I say 'y', is that a request to disable it, or enable it? And what exactly do I get if I vote *against* 'more accurate wchan values'? Do I get everybody having the same wchan pointing somewhere in the scheduler code, because that's where __builtin_return_address() points?

And please - a triple negative in the Kconfig variable name? This has gotta be a winner for poor taste in variable naming...

Quoted 6 lines

Even if that is all sorted out, how the heck is anyone to decide whether or not they need this thing?

Also, if we really really are so wishy-washy indecisive that we need to make the function optional, it should if at all possible be made runtime-configurable, not compile-time.

This came from commit a87d091434ed2a34d647979ab12084139ee1fe41 ("x86,
sched: enable wchan config menu item on 64-bit"). We have had CONFIG_SCHED_NO_NO_OMIT_FRAME_POINTER for some time on various architectures, this commit just made it available on x86_64 (by changing its dependency from X86_32 to X86).

Signature · 3 lines
Cheers,
Stephen Rothwell                    [email protected]
http://www.canb.auug.org.au/~sfr/
#3 lore raw
Quoted 5 lines · Stephen Rothwell

This came from commit a87d091434ed2a34d647979ab12084139ee1fe41 ("x86, sched: enable wchan config menu item on 64-bit"). We have had CONFIG_SCHED_NO_NO_OMIT_FRAME_POINTER for some time on various architectures, this commit just made it available on x86_64 (by changing its dependency from X86_32 to X86).

correct. The SCHED_NO_NO name comes from ancient history. Nevertheless i just renamed it to CONFIG_SCHED_OMIT_FRAME_POINTER in tip/sched/core, and pushed out a new auto-sched-next with that rename in place. We can get rid of it completely once Ken's /proc/<pid>/stack hits upstream. (as the superior replacement for wchan)

Ingo

----------------->

Quoted 1 line

From ae1e9130bfb9ad55eb97ec3fb17a122b7a118f98 Mon Sep 17 00:00:00 2001

From: Ingo Molnar <[email protected]>
Date: Tue, 11 Nov 2008 09:05:16 +0100
Subject: [PATCH] sched: rename SCHED_NO_NO_OMIT_FRAME_POINTER => SCHED_OMIT_FRAME_POINTER

Impact: cleanup, change .config option name

We had this ugly config name for a long time for hysteric raisons. Rename it to a saner name.

We still cannot get rid of it completely, until /proc/<pid>/stack usage replaces WCHAN usage for good.

We'll be able to do that in the v2.6.29/v2.6.30 timeframe.

---
 arch/ia64/Kconfig         |    2 +-
 arch/m32r/Kconfig         |    2 +-
 arch/mips/Kconfig         |    2 +-
 arch/powerpc/Kconfig      |    2 +-
 arch/x86/Kconfig          |    2 +-
 include/asm-m32r/system.h |    2 +-
 kernel/Makefile           |    2 +-
 7 files changed, 7 insertions(+), 7 deletions(-)
arch/ia64/Kconfig +1 -1
index 27eec71..59d1278 100644
@@ -99,7 +99,7 @@ config GENERIC_IOMAP
 	bool
 	default y
 
-config SCHED_NO_NO_OMIT_FRAME_POINTER
+config SCHED_OMIT_FRAME_POINTER
 	bool
 	default y
 
arch/m32r/Kconfig +1 -1
index dbaed4a..29047d5 100644
@@ -273,7 +273,7 @@ config GENERIC_CALIBRATE_DELAY
 	bool
 	default y
 
-config SCHED_NO_NO_OMIT_FRAME_POINTER
+config SCHED_OMIT_FRAME_POINTER
         bool
         default y
 
arch/mips/Kconfig +1 -1
index f4af967..a5255e7 100644
@@ -653,7 +653,7 @@ config GENERIC_CMOS_UPDATE
 	bool
 	default y
 
-config SCHED_NO_NO_OMIT_FRAME_POINTER
+config SCHED_OMIT_FRAME_POINTER
 	bool
 	default y
 
arch/powerpc/Kconfig +1 -1
index 525c13a..adb23ea 100644
@@ -141,7 +141,7 @@ config GENERIC_NVRAM
 	bool
 	default y if PPC32
 
-config SCHED_NO_NO_OMIT_FRAME_POINTER
+config SCHED_OMIT_FRAME_POINTER
 	bool
 	default y
 
arch/x86/Kconfig +1 -1
index 1d5550d..74db682 100644
@@ -364,7 +364,7 @@ config X86_RDC321X
 	  as R-8610-(G).
 	  If you don't have one of these chips, you should say N here.
 
-config SCHED_NO_NO_OMIT_FRAME_POINTER
+config SCHED_OMIT_FRAME_POINTER
 	def_bool y
 	prompt "Single-depth WCHAN output"
 	depends on X86
include/asm-m32r/system.h +1 -1
index 70a57c8..c980f5b 100644
@@ -23,7 +23,7 @@
  */
 
 #if defined(CONFIG_FRAME_POINTER) || \
-	!defined(CONFIG_SCHED_NO_NO_OMIT_FRAME_POINTER)
+	!defined(CONFIG_SCHED_OMIT_FRAME_POINTER)
 #define M32R_PUSH_FP "	push fp\n"
 #define M32R_POP_FP  "	pop  fp\n"
 #else
kernel/Makefile +1 -1
index e1af039..46e67a3 100644
@@ -91,7 +91,7 @@ obj-$(CONFIG_FUNCTION_TRACER) += trace/
 obj-$(CONFIG_TRACING) += trace/
 obj-$(CONFIG_SMP) += sched_cpupri.o
 
-ifneq ($(CONFIG_SCHED_NO_NO_OMIT_FRAME_POINTER),y)
+ifneq ($(CONFIG_SCHED_OMIT_FRAME_POINTER),y)
 # According to Alan Modra <[email protected]>, the -fno-omit-frame-pointer is
 # needed for x86 only.  Why this used to be enabled for all architectures is beyond
 # me.  I suspect most platforms don't need this, but until we know that for sure