Use bool for boolean flag in printk_once()
PATCH 1/1 Use bool for boolean flag in printk_once()
Using the type bool (instead of int) for the __print_once flag in the printk_once() macro matches the intent of the code better, and allows the compiler to generate smaller code; eg a typical callsite with gcc 4.3.3 on i386:
add/remove: 0/0 grow/shrink: 0/2 up/down: 0/-6 (-6)
function old new delta
static.__print_once 4 1 -3get_cpu_vendor 146 143 -3
Saving 6 bytes of object size per callsite by slightly improving the readability of the source seems like a win to me.
--- include/linux/kernel.h | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-)
include/linux/kernel.h +2 -2
@@ -249,10 +249,10 @@ extern bool printk_timed_ratelimit(unsigned long *caller_jiffies,* Print a one-time message (analogous to WARN_ONCE() et al): */ #define printk_once(x...) ({ \ - static int __print_once = 1; \ + static bool __print_once = true; \ \ if (__print_once) { \ - __print_once = 0; \ + __print_once = false; \ printk(x); \ } \ })
Quoted 25 lines · On Thu, 13 Aug 2009 13:48:26 -0700 Roland Dreier <[email protected]> wrote:
Using the type bool (instead of int) for the
__print_onceflag in theprintk_once()macro matches the intent of the code better, and allows the compiler to generate smaller code; eg a typical callsite with gcc 4.3.3 on i386:add/remove: 0/0 grow/shrink: 0/2 up/down: 0/-6 (-6)
function old new delta
static.__print_once 4 1 -3get_cpu_vendor146 143 -3Saving 6 bytes of object size per callsite by slightly improving the readability of the source seems like a win to me.
Signed-off-by Roland Dreier <[email protected]>--- include/linux/kernel.h | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-)include/linux/kernel.h +0
-0@@ -249,10 +249,10 @@ extern bool printk_timed_ratelimit(unsigned long *caller_jiffies,* Print a one-time message (analogous to WARN_ONCE() et al): */
include/linux/kernel.h extern bool printk_timed_ratelimit(unsigned long *caller_jiffies, · line 258
Quoted 10 lines
include/linux/kernel.h +2
-2#define printk_once(x...) ({ \ - static int __print_once = 1; \ + static bool __print_once = true; \ \ if (__print_once) { \ - __print_once = 0; \ + __print_once = false; \ printk(x); \ } \ })
hm, OK, in trace_recursive_lock() we get:
cmpl $0, __print_once.28104(%rip) #, __print_once je .L719 #,
changed to
cmpb $0, __print_once.28104(%rip) # __print_once je .L719 #,
so the compiler used a byte for the bool.
Interestingly that bool is in section `data'. I bet printk_once() should use __read_mostly.
Also, that bool still uses four bytes of storage:
text data bss dec hex filename 22527 1445 6392 30364 769c kernel/trace/ring_buffer.o 22524 1445 6392 30361 7699 kernel/trace/ring_buffer.o (patched)
I wonder if there's some way in which we could/should cause all these little random __read_mostly bytes to get packed together somewhere.
Quoted 41 lines · Andrew Morton
Quoted 25 lines · On Thu, 13 Aug 2009 13:48:26 -0700 Roland Dreier <[email protected]> wrote:
Using the type bool (instead of int) for the
__print_onceflag in theprintk_once()macro matches the intent of the code better, and allows the compiler to generate smaller code; eg a typical callsite with gcc 4.3.3 on i386:add/remove: 0/0 grow/shrink: 0/2 up/down: 0/-6 (-6)
function old new delta
static.__print_once 4 1 -3get_cpu_vendor146 143 -3Saving 6 bytes of object size per callsite by slightly improving the readability of the source seems like a win to me.
Signed-off-by Roland Dreier <[email protected]>--- include/linux/kernel.h | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-)include/linux/kernel.h +0
-0@@ -249,10 +249,10 @@ extern bool printk_timed_ratelimit(unsigned long *caller_jiffies,* Print a one-time message (analogous to WARN_ONCE() et al): */ include/linux/kernel.h extern bool printk_timed_ratelimit(unsigned long *caller_jiffies, · line 258
Quoted 10 lines
include/linux/kernel.h +2
-2#define printk_once(x...) ({ \ - static int __print_once = 1; \ + static bool __print_once = true; \ \ if (__print_once) { \ - __print_once = 0; \ + __print_once = false; \ printk(x); \ } \ })hm, OK, in
trace_recursive_lock()we get:
Quoted 7 lines
cmpl $0, __print_once.28104(%rip) #, __print_once je .L719 #,changed to
cmpb $0, __print_once.28104(%rip) # __print_once je .L719 #,
The test should be inverted so that it's not initialized.
I submitted this once:
http://lkml.indiana.edu/hypermail/linux/kernel/0905.2/02636.html
static bool printed;
if (!printed) {
printed = true;
printk...
}