Creating 2**32 tasks is impossible due to futex pid limits and wasteful
anyway. Nobody has done it.
Bring nr_running() into 32-bit world to save on REX prefixes.
Signed-off-by: Alexey Dobriyan <[email protected]>
---
fs/proc/loadavg.c | 2 +-
fs/proc/stat.c | 2 +-
include/linux/sched/stat.h | 2 +-
kernel/sched/core.c | 4 ++--
4 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/fs/proc/loadavg.c b/fs/proc/loadavg.c
index 8468baee951d..f32878d9a39f 100644
--- a/fs/proc/loadavg.c
+++ b/fs/proc/loadavg.c
@@ -16,7 +16,7 @@ static int loadavg_proc_show(struct seq_file *m, void *v)
get_avenrun(avnrun, FIXED_1/200, 0);
- seq_printf(m, "%lu.%02lu %lu.%02lu %lu.%02lu %ld/%d %d\n",
+ seq_printf(m, "%lu.%02lu %lu.%02lu %lu.%02lu %u/%d %d\n",
LOAD_INT(avnrun[0]), LOAD_FRAC(avnrun[0]),
LOAD_INT(avnrun[1]), LOAD_FRAC(avnrun[1]),
LOAD_INT(avnrun[2]), LOAD_FRAC(avnrun[2]),
diff --git a/fs/proc/stat.c b/fs/proc/stat.c
index f25e8531fd27..941605de7f9a 100644
--- a/fs/proc/stat.c
+++ b/fs/proc/stat.c
@@ -200,7 +200,7 @@ static int show_stat(struct seq_file *p, void *v)
"\nctxt %llu\n"
"btime %llu\n"
"processes %lu\n"
- "procs_running %lu\n"
+ "procs_running %u\n"
"procs_blocked %lu\n",
nr_context_switches(),
(unsigned long long)boottime.tv_sec,
diff --git a/include/linux/sched/stat.h b/include/linux/sched/stat.h
index 568286411b43..f742229091ec 100644
--- a/include/linux/sched/stat.h
+++ b/include/linux/sched/stat.h
@@ -16,7 +16,7 @@ extern unsigned long total_forks;
extern int nr_threads;
DECLARE_PER_CPU(unsigned long, process_counts);
extern int nr_processes(void);
-extern unsigned long nr_running(void);
+extern unsigned int nr_running(void);
extern bool single_task_running(void);
extern unsigned long nr_iowait(void);
extern unsigned long nr_iowait_cpu(int cpu);
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 98191218d891..713ea35cb995 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -4331,9 +4331,9 @@ context_switch(struct rq *rq, struct task_struct *prev,
* externally visible scheduler statistics: current number of runnable
* threads, total number of context switches performed since bootup.
*/
-unsigned long nr_running(void)
+unsigned int nr_running(void)
{
- unsigned long i, sum = 0;
+ unsigned int i, sum = 0;
for_each_online_cpu(i)
sum += cpu_rq(i)->nr_running;
--
2.30.2
Make
struct dl_rq::dl_nr_migratory
struct dl_rq::dl_nr_running
struct rt_rq::rt_nr_boosted
struct rt_rq::rt_nr_migratory
struct rt_rq::rt_nr_total
struct rq::nr_uninterruptible
32-bit.
If total number of tasks can't exceed 2**32 (and less due to futex pid
limits), then per-runqueue counters can't as well.
This patchset has been sponsored by REX Prefix Eradication Society.
Signed-off-by: Alexey Dobriyan <[email protected]>
---
kernel/sched/loadavg.c | 2 +-
kernel/sched/sched.h | 12 ++++++------
2 files changed, 7 insertions(+), 7 deletions(-)
diff --git a/kernel/sched/loadavg.c b/kernel/sched/loadavg.c
index d2a655643a02..aef8072cfebe 100644
--- a/kernel/sched/loadavg.c
+++ b/kernel/sched/loadavg.c
@@ -81,7 +81,7 @@ long calc_load_fold_active(struct rq *this_rq, long adjust)
long nr_active, delta = 0;
nr_active = this_rq->nr_running - adjust;
- nr_active += (long)this_rq->nr_uninterruptible;
+ nr_active += (int)this_rq->nr_uninterruptible;
if (nr_active != this_rq->calc_load_active) {
delta = nr_active - this_rq->calc_load_active;
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index 10a1522b1e30..730c81a54ed1 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -622,8 +622,8 @@ struct rt_rq {
} highest_prio;
#endif
#ifdef CONFIG_SMP
- unsigned long rt_nr_migratory;
- unsigned long rt_nr_total;
+ unsigned int rt_nr_migratory;
+ unsigned int rt_nr_total;
int overloaded;
struct plist_head pushable_tasks;
@@ -637,7 +637,7 @@ struct rt_rq {
raw_spinlock_t rt_runtime_lock;
#ifdef CONFIG_RT_GROUP_SCHED
- unsigned long rt_nr_boosted;
+ unsigned int rt_nr_boosted;
struct rq *rq;
struct task_group *tg;
@@ -654,7 +654,7 @@ struct dl_rq {
/* runqueue is an rbtree, ordered by deadline */
struct rb_root_cached root;
- unsigned long dl_nr_running;
+ unsigned int dl_nr_running;
#ifdef CONFIG_SMP
/*
@@ -668,7 +668,7 @@ struct dl_rq {
u64 next;
} earliest_dl;
- unsigned long dl_nr_migratory;
+ unsigned int dl_nr_migratory;
int overloaded;
/*
@@ -946,7 +946,7 @@ struct rq {
* one CPU and if it got migrated afterwards it may decrease
* it on another CPU. Always updated under the runqueue lock:
*/
- unsigned long nr_uninterruptible;
+ unsigned int nr_uninterruptible;
struct task_struct __rcu *curr;
struct task_struct *idle;
--
2.30.2
The following commit has been merged into the sched/core branch of tip:
Commit-ID: e6fe3f422be128b7d65de607f6ae67bedc55f0ca
Gitweb: https://git.kernel.org/tip/e6fe3f422be128b7d65de607f6ae67bedc55f0ca
Author: Alexey Dobriyan <[email protected]>
AuthorDate: Thu, 22 Apr 2021 23:02:28 +03:00
Committer: Ingo Molnar <[email protected]>
CommitterDate: Wed, 12 May 2021 21:34:17 +02:00
sched: Make multiple runqueue task counters 32-bit
Make:
struct dl_rq::dl_nr_migratory
struct dl_rq::dl_nr_running
struct rt_rq::rt_nr_boosted
struct rt_rq::rt_nr_migratory
struct rt_rq::rt_nr_total
struct rq::nr_uninterruptible
32-bit.
If total number of tasks can't exceed 2**32 (and less due to futex pid
limits), then per-runqueue counters can't as well.
This patchset has been sponsored by REX Prefix Eradication Society.
Signed-off-by: Alexey Dobriyan <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
---
kernel/sched/loadavg.c | 2 +-
kernel/sched/sched.h | 12 ++++++------
2 files changed, 7 insertions(+), 7 deletions(-)
diff --git a/kernel/sched/loadavg.c b/kernel/sched/loadavg.c
index 1c79896..954b229 100644
--- a/kernel/sched/loadavg.c
+++ b/kernel/sched/loadavg.c
@@ -81,7 +81,7 @@ long calc_load_fold_active(struct rq *this_rq, long adjust)
long nr_active, delta = 0;
nr_active = this_rq->nr_running - adjust;
- nr_active += (long)this_rq->nr_uninterruptible;
+ nr_active += (int)this_rq->nr_uninterruptible;
if (nr_active != this_rq->calc_load_active) {
delta = nr_active - this_rq->calc_load_active;
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index 904c52b..8f0194c 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -636,8 +636,8 @@ struct rt_rq {
} highest_prio;
#endif
#ifdef CONFIG_SMP
- unsigned long rt_nr_migratory;
- unsigned long rt_nr_total;
+ unsigned int rt_nr_migratory;
+ unsigned int rt_nr_total;
int overloaded;
struct plist_head pushable_tasks;
@@ -651,7 +651,7 @@ struct rt_rq {
raw_spinlock_t rt_runtime_lock;
#ifdef CONFIG_RT_GROUP_SCHED
- unsigned long rt_nr_boosted;
+ unsigned int rt_nr_boosted;
struct rq *rq;
struct task_group *tg;
@@ -668,7 +668,7 @@ struct dl_rq {
/* runqueue is an rbtree, ordered by deadline */
struct rb_root_cached root;
- unsigned long dl_nr_running;
+ unsigned int dl_nr_running;
#ifdef CONFIG_SMP
/*
@@ -682,7 +682,7 @@ struct dl_rq {
u64 next;
} earliest_dl;
- unsigned long dl_nr_migratory;
+ unsigned int dl_nr_migratory;
int overloaded;
/*
@@ -960,7 +960,7 @@ struct rq {
* one CPU and if it got migrated afterwards it may decrease
* it on another CPU. Always updated under the runqueue lock:
*/
- unsigned long nr_uninterruptible;
+ unsigned int nr_uninterruptible;
struct task_struct __rcu *curr;
struct task_struct *idle;
The following commit has been merged into the sched/core branch of tip:
Commit-ID: 01aee8fd7fb23049e2b52abadbe1f7b5e94a52d2
Gitweb: https://git.kernel.org/tip/01aee8fd7fb23049e2b52abadbe1f7b5e94a52d2
Author: Alexey Dobriyan <[email protected]>
AuthorDate: Thu, 22 Apr 2021 23:02:25 +03:00
Committer: Ingo Molnar <[email protected]>
CommitterDate: Wed, 12 May 2021 21:34:14 +02:00
sched: Make nr_running() return 32-bit value
Creating 2**32 tasks is impossible due to futex pid limits and wasteful
anyway. Nobody has done it.
Bring nr_running() into 32-bit world to save on REX prefixes.
Signed-off-by: Alexey Dobriyan <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
---
fs/proc/loadavg.c | 2 +-
fs/proc/stat.c | 2 +-
include/linux/sched/stat.h | 2 +-
kernel/sched/core.c | 4 ++--
4 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/fs/proc/loadavg.c b/fs/proc/loadavg.c
index 8468bae..f32878d 100644
--- a/fs/proc/loadavg.c
+++ b/fs/proc/loadavg.c
@@ -16,7 +16,7 @@ static int loadavg_proc_show(struct seq_file *m, void *v)
get_avenrun(avnrun, FIXED_1/200, 0);
- seq_printf(m, "%lu.%02lu %lu.%02lu %lu.%02lu %ld/%d %d\n",
+ seq_printf(m, "%lu.%02lu %lu.%02lu %lu.%02lu %u/%d %d\n",
LOAD_INT(avnrun[0]), LOAD_FRAC(avnrun[0]),
LOAD_INT(avnrun[1]), LOAD_FRAC(avnrun[1]),
LOAD_INT(avnrun[2]), LOAD_FRAC(avnrun[2]),
diff --git a/fs/proc/stat.c b/fs/proc/stat.c
index f25e853..941605d 100644
--- a/fs/proc/stat.c
+++ b/fs/proc/stat.c
@@ -200,7 +200,7 @@ static int show_stat(struct seq_file *p, void *v)
"\nctxt %llu\n"
"btime %llu\n"
"processes %lu\n"
- "procs_running %lu\n"
+ "procs_running %u\n"
"procs_blocked %lu\n",
nr_context_switches(),
(unsigned long long)boottime.tv_sec,
diff --git a/include/linux/sched/stat.h b/include/linux/sched/stat.h
index 939c3ec..73606b3 100644
--- a/include/linux/sched/stat.h
+++ b/include/linux/sched/stat.h
@@ -17,7 +17,7 @@ extern unsigned long total_forks;
extern int nr_threads;
DECLARE_PER_CPU(unsigned long, process_counts);
extern int nr_processes(void);
-extern unsigned long nr_running(void);
+extern unsigned int nr_running(void);
extern bool single_task_running(void);
extern unsigned long nr_iowait(void);
extern unsigned long nr_iowait_cpu(int cpu);
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index ac8882d..2c6cdb0 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -4692,9 +4692,9 @@ context_switch(struct rq *rq, struct task_struct *prev,
* externally visible scheduler statistics: current number of runnable
* threads, total number of context switches performed since bootup.
*/
-unsigned long nr_running(void)
+unsigned int nr_running(void)
{
- unsigned long i, sum = 0;
+ unsigned int i, sum = 0;
for_each_online_cpu(i)
sum += cpu_rq(i)->nr_running;
Alexey,
On Thu, Apr 22 2021 at 23:02, Alexey Dobriyan wrote:
> Creating 2**32 tasks is impossible due to futex pid limits and wasteful
> anyway. Nobody has done it.
>
this whole pile lacks useful numbers. What's the actual benefit of that
churn?
Just with the default config for one of my reference machines:
text data bss dec hex filename
16679864 6627950 1671296 24979110 17d26a6 ../build/vmlinux-before
16679894 6627950 1671296 24979140 17d26c4 ../build/vmlinux-after
------------------------------------------------------------------------
+30
I'm truly impressed by the massive savings of this change and I'm even
more impressed by the justification:
> Bring nr_running() into 32-bit world to save on REX prefixes.
Aside of the obvious useless churn, REX prefixes are universaly true for
all architectures, right? There is a world outside x86 ...
Thanks,
tglx
* Alexey Dobriyan <[email protected]> wrote:
> Make
>
> struct dl_rq::dl_nr_migratory
> struct dl_rq::dl_nr_running
>
> struct rt_rq::rt_nr_boosted
> struct rt_rq::rt_nr_migratory
> struct rt_rq::rt_nr_total
>
> struct rq::nr_uninterruptible
>
> 32-bit.
>
> If total number of tasks can't exceed 2**32 (and less due to futex pid
> limits), then per-runqueue counters can't as well.
Applied to tip:sched/core, thanks!
There was a bit of a conflict with recent changes in
drivers/cpuidle/governors/menu.c, but I fixed it up, hopefully correctly.
> This patchset has been sponsored by REX Prefix Eradication Society.
A patchset with such an impeccable recommendation letter is impossible to
resist.
Thanks,
Ingo
On Thu, May 13, 2021 at 01:58:16AM +0200, Thomas Gleixner wrote:
> Alexey,
>
> On Thu, Apr 22 2021 at 23:02, Alexey Dobriyan wrote:
> > Creating 2**32 tasks is impossible due to futex pid limits and wasteful
> > anyway. Nobody has done it.
> >
>
> this whole pile lacks useful numbers. What's the actual benefit of that
> churn?
The long term goal is to use 32-bit data more. People will see it in
core kernel and copy everywhere elase.
> Just with the default config for one of my reference machines:
>
> text data bss dec hex filename
> 16679864 6627950 1671296 24979110 17d26a6 ../build/vmlinux-before
> 16679894 6627950 1671296 24979140 17d26c4 ../build/vmlinux-after
> ------------------------------------------------------------------------
> +30
>
> I'm truly impressed by the massive savings of this change and I'm even
> more impressed by the justification:
>
> > Bring nr_running() into 32-bit world to save on REX prefixes.
I collected numbers initially but then stopped because noone cared and
they can be config and arch dependent.
> Aside of the obvious useless churn,
oh... Sometimes I think churn is the whole point.
> REX prefixes are universaly true for
> all architectures, right? There is a world outside x86 ...
In general, 32-bitness is preferred for code generation.
32-bit RISCs naturally prefers 32-bit.
64-bit RISCs don't care because they remember 32-bit roots and
have necessary 32-bit fixed width(!) instructions.
x86_64 is the only arch where going 64-bit generally adds more bytes
to the instruction stream.
Effects can be smudged by compilers of course, in this case, percpu
stuff. That "unsigned int i" is a mistake. Proper diff looks like this:
-ffffffff811115fa: 8b 44 18 04 mov eax,DWORD PTR [rax+rbx*1+0x4]
-ffffffff811115fe: 49 01 c4 add r12,rax
+ffffffff811115fa: 44 03 64 18 04 add r12d,DWORD PTR [rax+rbx*1+0x4]
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -4348,9 +4348,10 @@ context_switch(struct rq *rq, struct task_struct *prev,
* externally visible scheduler statistics: current number of runnable
* threads, total number of context switches performed since bootup.
*/
-unsigned long nr_running(void)
+unsigned int nr_running(void)
{
- unsigned long i, sum = 0;
+ unsigned int sum = 0;
+ unsigned long i;
for_each_online_cpu(i)
sum += cpu_rq(i)->nr_running;
* Thomas Gleixner <[email protected]> wrote:
> Alexey,
>
> On Thu, Apr 22 2021 at 23:02, Alexey Dobriyan wrote:
> > Creating 2**32 tasks is impossible due to futex pid limits and wasteful
> > anyway. Nobody has done it.
> >
>
> this whole pile lacks useful numbers. What's the actual benefit of that
> churn?
I applied the 4 patch series from Alexey Dobriyan because they offer four
distinct technical advantages to the scheduler code:
- Shorter instructions generated in nr_running(), nr_iowait(),
nr_iowait_cpu() due to losing the REX prefix.
- Shorter instructions generated by users of these 3 functions as well.
- Tighter data packing and structure size reduction in 'struct rt_rq',
'struct dl_rq' and 'struct rq', due to 8-byte 'long' fields shrinking to
4-byte 'int' based fields.
- Together these 4 patches clean up all derivative uses of the
rq::nr_running base type, which is already 'unsigned int' and that's
pretty fundamental given our PID ABI limits. Having type mismatch where
we use 64-bit data types for certain APIs while 32-bit data types for
others is inconsistent crap I wouldn't accept if it was submitted as new
code.
As to the numbers:
The data structure size improvements are IMO obvious, and they are also
measurable, here's the before/after Pahole comparison of 'struct rt_rq':
--- pahole.struct.rt_rq.before 2021-05-13 11:40:53.207077908 +0200
+++ pahole.struct.rt_rq.after 2021-05-13 11:41:42.257385897 +0200
@@ -7,22 +7,22 @@ struct rt_rq {
int curr; /* 1624 4 */
int next; /* 1628 4 */
} highest_prio; /* 1624 8 */
- long unsigned int rt_nr_migratory; /* 1632 8 */
- long unsigned int rt_nr_total; /* 1640 8 */
- int overloaded; /* 1648 4 */
+ unsigned int rt_nr_migratory; /* 1632 4 */
+ unsigned int rt_nr_total; /* 1636 4 */
+ int overloaded; /* 1640 4 */
/* XXX 4 bytes hole, try to pack */
- struct plist_head pushable_tasks; /* 1656 16 */
- /* --- cacheline 26 boundary (1664 bytes) was 8 bytes ago --- */
- int rt_queued; /* 1672 4 */
- int rt_throttled; /* 1676 4 */
- u64 rt_time; /* 1680 8 */
- u64 rt_runtime; /* 1688 8 */
- raw_spinlock_t rt_runtime_lock; /* 1696 4 */
+ struct plist_head pushable_tasks; /* 1648 16 */
+ /* --- cacheline 26 boundary (1664 bytes) --- */
+ int rt_queued; /* 1664 4 */
+ int rt_throttled; /* 1668 4 */
+ u64 rt_time; /* 1672 8 */
+ u64 rt_runtime; /* 1680 8 */
+ raw_spinlock_t rt_runtime_lock; /* 1688 4 */
- /* size: 1704, cachelines: 27, members: 13 */
- /* sum members: 1696, holes: 1, sum holes: 4 */
+ /* size: 1696, cachelines: 27, members: 13 */
+ /* sum members: 1688, holes: 1, sum holes: 4 */
/* padding: 4 */
- /* last cacheline: 40 bytes */
+ /* last cacheline: 32 bytes */
'struct rt_rq' got shrunk from 1704 bytes to 1696 bytes, an 8 byte
reduction.
'struct dl_rq' shrunk by 8 bytes:
- /* size: 104, cachelines: 2, members: 10 */
- /* sum members: 100, holes: 1, sum holes: 4 */
- /* last cacheline: 40 bytes */
+ /* size: 96, cachelines: 2, members: 10 */
+ /* sum members: 92, holes: 1, sum holes: 4 */
+ /* last cacheline: 32 bytes */
'struct rq', which embedds both rt_rq and dl_rq, got 20 bytes smaller:
- /* sum members: 2967, holes: 10, sum holes: 137 */
+ /* sum members: 2947, holes: 11, sum holes: 157 */
Side note: there's now 20 bytes more new padding holes which could possibly
be coalesced a bit more by reordering members sensibly - resulting in even
more data footprint savings. (But those should be separate changes and only
for fields we truly care about from a performance POV.)
As to code generation improvements:
> Just with the default config for one of my reference machines:
>
> text data bss dec hex filename
> 16679864 6627950 1671296 24979110 17d26a6 ../build/vmlinux-before
> 16679894 6627950 1671296 24979140 17d26c4 ../build/vmlinux-after
> ------------------------------------------------------------------------
> +30
Using '/usr/bin/size' to compare before/after generated code is the wrong
way to measure code generation improvements for smaller changes due to
default alignment creating a reserve of 'padding' bytes at the end of most
functions. You have to look at the low level generated assembly directly.
For example, the nr_iowait_cpu() commit, with before/after generated code
of the callers of the function:
9745516841a5: ("sched: Make nr_iowait() return 32-bit value")
ffffffffxxxxxxxx: e8 49 8e fb ff call ffffffffxxxxxxxx <nr_iowait_cpu>
- ffffffffxxxxxxxx: 48 85 c0 test %rax,%rax
ffffffffxxxxxxxx: e8 64 8e fb ff call ffffffffxxxxxxxx <nr_iowait_cpu>
+ ffffffffxxxxxxxx: 85 c0 test %eax,%eax
Note how the 'test %rax,%rax' lost the 0x48 64-bit REX prefix and the
generated code got one byte shorter from "48 85 c0" to "85 c0".
( Note, my asm generation scripts filter out some of the noise to make it
easier to diff generated asm, hence the ffffffffxxxxxxxx placeholder. )
The nr_iowait() function itself got shorter by two bytes as well, due to:
Before:
ffffffffxxxxxxxx <nr_iowait>:
ffffffffxxxxxxxx: 41 54 push %r12
ffffffffxxxxxxxx: 48 c7 c7 ff ff ff ff mov $0xffffffffxxxxxxxx,%rdi
ffffffffxxxxxxxx: 45 31 e4 xor %r12d,%r12d
ffffffffxxxxxxxx: 55 push %rbp
ffffffffxxxxxxxx: 8b 2d 01 ea 5d 01 mov 0x15dea01(%rip),%ebp # ffffffffxxxxxxxx <nr_cpu_ids>
ffffffffxxxxxxxx: 53 push %rbx
ffffffffxxxxxxxx: 48 c7 c3 c0 95 02 00 mov $0x295c0,%rbx
ffffffffxxxxxxxx: eb 17 jmp ffffffffxxxxxxxx <nr_iowait+0x34>
ffffffffxxxxxxxx: 48 98 cltq
ffffffffxxxxxxxx: 48 8b 14 c5 a0 c6 2e mov -0x7dd13960(,%rax,8),%rdx
ffffffffxxxxxxxx: 82
ffffffffxxxxxxxx: 48 01 da add %rbx,%rdx
ffffffffxxxxxxxx: 48 63 82 98 09 00 00 movslq 0x998(%rdx),%rax
ffffffffxxxxxxxx: 49 01 c4 add %rax,%r12
ffffffffxxxxxxxx: 48 c7 c6 18 72 67 82 mov $0xffffffffxxxxxxxx,%rsi
ffffffffxxxxxxxx: e8 80 46 39 00 call ffffffffxxxxxxxx <cpumask_next>
ffffffffxxxxxxxx: 89 c7 mov %eax,%edi
ffffffffxxxxxxxx: 39 e8 cmp %ebp,%eax
ffffffffxxxxxxxx: 72 d7 jb ffffffffxxxxxxxx <nr_iowait+0x1d>
ffffffffxxxxxxxx: 4c 89 e0 mov %r12,%rax
ffffffffxxxxxxxx: 5b pop %rbx
ffffffffxxxxxxxx: 5d pop %rbp
ffffffffxxxxxxxx: 41 5c pop %r12
ffffffffxxxxxxxx: c3 ret
After:
ffffffffxxxxxxxx <nr_iowait>:
ffffffffxxxxxxxx: 41 54 push %r12
ffffffffxxxxxxxx: bf ff ff ff ff mov $0xffffffff,%edi
ffffffffxxxxxxxx: 45 31 e4 xor %r12d,%r12d
ffffffffxxxxxxxx: 55 push %rbp
ffffffffxxxxxxxx: 8b 2d 03 ea 5d 01 mov 0x15dea03(%rip),%ebp # ffffffffxxxxxxxx <nr_cpu_ids>
ffffffffxxxxxxxx: 53 push %rbx
ffffffffxxxxxxxx: 48 c7 c3 c0 95 02 00 mov $0x295c0,%rbx
ffffffffxxxxxxxx: eb 17 jmp ffffffffxxxxxxxx <nr_iowait+0x32>
ffffffffxxxxxxxx: 48 63 c7 movslq %edi,%rax
ffffffffxxxxxxxx: 48 8b 14 c5 a0 c6 2e mov -0x7dd13960(,%rax,8),%rdx
ffffffffxxxxxxxx: 82
ffffffffxxxxxxxx: 48 01 da add %rbx,%rdx
ffffffffxxxxxxxx: 8b 82 98 09 00 00 mov 0x998(%rdx),%eax
ffffffffxxxxxxxx: 41 01 c4 add %eax,%r12d
ffffffffxxxxxxxx: 48 c7 c6 18 72 67 82 mov $0xffffffffxxxxxxxx,%rsi
ffffffffxxxxxxxx: e8 a2 46 39 00 call ffffffffxxxxxxxx <cpumask_next>
ffffffffxxxxxxxx: 89 c7 mov %eax,%edi
ffffffffxxxxxxxx: 39 c5 cmp %eax,%ebp
ffffffffxxxxxxxx: 77 d7 ja ffffffffxxxxxxxx <nr_iowait+0x1b>
ffffffffxxxxxxxx: 44 89 e0 mov %r12d,%eax
ffffffffxxxxxxxx: 5b pop %rbx
ffffffffxxxxxxxx: 5d pop %rbp
ffffffffxxxxxxxx: 41 5c pop %r12
ffffffffxxxxxxxx: c3 ret
The size of nr_iowait() shrunk from 78 bytes to 76 bytes.
Or the other commit:
01aee8fd7fb2: ("sched: Make nr_running() return 32-bit value")
ffffffffxxxxxxxx: e8 d9 24 e8 ff call ffffffffxxxxxxxx <nr_running>
ffffffffxxxxxxxx: 4c 8b 05 cc 92 70 01 mov 0x17092cc(%rip),%r8 # ffffffffxxxxxxxx <total_forks>
ffffffffxxxxxxxx: 4c 8b 64 24 50 mov 0x50(%rsp),%r12
- ffffffffxxxxxxxx: 48 89 44 24 08 mov %rax,0x8(%rsp)
ffffffffxxxxxxxx: 4c 89 04 24 mov %r8,(%rsp)
ffffffffxxxxxxxx: e8 f9 24 e8 ff call ffffffffxxxxxxxx <nr_running>
ffffffffxxxxxxxx: 4c 8b 05 ed 92 70 01 mov 0x17092ed(%rip),%r8 # ffffffffxxxxxxxx <total_forks>
ffffffffxxxxxxxx: 4c 8b 64 24 50 mov 0x50(%rsp),%r12
+ ffffffffxxxxxxxx: 89 44 24 08 mov %eax,0x8(%rsp)
ffffffffxxxxxxxx: 4c 89 04 24 mov %r8,(%rsp)
Note how "mov %rax,0x8(%rsp)" got shortened by one byte to "mov %eax,0x8(%rsp)":
- ffffffffxxxxxxxx: 48 89 44 24 08 mov %rax,0x8(%rsp)
+ ffffffffxxxxxxxx: 89 44 24 08 mov %eax,0x8(%rsp)
The nr_running() function itself got shorter by 2 bytes, due to shorter
instruction sequences.
The third commit improved code generation too:
8fc2858e572c: ("sched: Make nr_iowait_cpu() return 32-bit value")
ffffffffxxxxxxxx <nr_iowait_cpu>:
ffffffffxxxxxxxx: 48 63 ff movslq %edi,%rdi
ffffffffxxxxxxxx: 48 c7 c0 c0 95 02 00 mov $0x295c0,%rax
ffffffffxxxxxxxx: 48 03 04 fd a0 46 0f add -0x7df0b960(,%rdi,8),%rax
ffffffffxxxxxxxx: 82
- ffffffffxxxxxxxx: 48 63 80 98 09 00 00 movslq 0x998(%rax),%rax
ffffffffxxxxxxxx: c3 ret
ffffffffxxxxxxxx <nr_iowait_cpu>:
ffffffffxxxxxxxx: 48 63 ff movslq %edi,%rdi
ffffffffxxxxxxxx: 48 c7 c0 c0 95 02 00 mov $0x295c0,%rax
ffffffffxxxxxxxx: 48 03 04 fd a0 46 0f add -0x7df0b960(,%rdi,8),%rax
ffffffffxxxxxxxx: 82
- ffffffffxxxxxxxx: 8b 80 98 09 00 00 mov 0x998(%rax),%eax
ffffffffxxxxxxxx: c3 ret
Note how the 'movslq 0x998(%rax),%rax' lost the REX prefix and got one byte
shorter. Call sites got shorter too:
ffffffffxxxxxxxx: e8 e8 73 fa ff call ffffffffxxxxxxxx <nr_iowait_cpu>
- ffffffffxxxxxxxx: 48 85 c0 test %rax,%rax
ffffffffxxxxxxxx: e8 c8 73 fa ff call ffffffffxxxxxxxx <nr_iowait_cpu>
- ffffffffxxxxxxxx: 85 c0 test %eax,%eax
You often won't see these effects in the 'size vmlinux' output, because
function alignment padding reserves usually hide 1-2 byte size improvements
in generated code.
> I'm truly impressed by the massive savings of this change and I'm even
> more impressed by the justification:
>
> > Bring nr_running() into 32-bit world to save on REX prefixes.
>
> Aside of the obvious useless churn, REX prefixes are universaly true for
> all architectures, right? There is a world outside x86 ...
Even architectures that have the same instruction length for 32-bit and
64-bit data types benefit:
- smaller data structures benefit all 64-bit architectures
- the cleaner, more coherent nr_running type definitions are now more
consistently 'int' based, not the previous nonsensical 'int/long' mix
that also made the generated code larger on x86.
More importantly, the maintenance benchmark in these cases is not whether a
change actively helps every architecture we care about - but whether the
change is a *disadvantage* for them - and it isn't here.
Changes that primarily benefit one common architecture, while not others,
are still eligible for upstream merging if they otherwise meet the quality
threshold and don't hurt the other architectures.
TL;DR:
This benefits from this series are small, but are far from 'useless churn',
unless we want to arbitrarily cut off technically valid contributions that
improve generated code, data structure size and code readability, submitted
by a long-time contributor who has contributed over 1,300 patches to the
kernel already, just because we don't think these add up a significant
enough benefit?
No doubt the quality barrier must be and is higher for smaller changes -
but this series IMO passed that barrier.
Anyway, I've Cc:-ed Linus and Greg, if you are advocating for some sort of
cut-off threshold for small but measurable improvements from long-time
contributors, it should probably be clearly specified & documented in
Documentation/SubmittingPatches ...
Thanks,
Ingo
On Thu, May 13, 2021 at 11:58:38AM +0200, Ingo Molnar wrote:
>
> * Thomas Gleixner <[email protected]> wrote:
>
> > Just with the default config for one of my reference machines:
> >
> > text data bss dec hex filename
> > 16679864 6627950 1671296 24979110 17d26a6 ../build/vmlinux-before
> > 16679894 6627950 1671296 24979140 17d26c4 ../build/vmlinux-after
> > ------------------------------------------------------------------------
> > +30
>
> Using '/usr/bin/size' to compare before/after generated code is the wrong
> way to measure code generation improvements for smaller changes due to
> default alignment creating a reserve of 'padding' bytes at the end of most
> functions. You have to look at the low level generated assembly directly.
This is bloat-o-meter output with Fedora 33 .config:
This is how they look like, something gets bigger but total is smaller
(otherwise why would I send it). Apparently something got one 1 byte too
many and pushed padding.
add/remove: 0/0 grow/shrink: 6/21 up/down: 18/-42 (-24)
Function old new delta
calc_load_fold_active 50 56 +6
calc_load_nohz_start 100 103 +3
calc_load_nohz_remote 85 88 +3
calc_global_load_tick 86 89 +3
pull_dl_task 901 903 +2
switched_from_dl 613 614 +1
update_rt_migration 165 164 -1
update_dl_migration 141 140 -1
ttwu_do_activate 181 180 -1
tick_nohz_idle_exit 225 224 -1
tick_irq_enter 227 226 -1
print_rt_rq.cold 238 237 -1
print_rt_rq 413 412 -1
nr_iowait_cpu 31 30 -1
init_rt_rq 143 142 -1
show_stat 1745 1743 -2
nr_running 75 73 -2
nr_iowait 83 81 -2
get_cpu_iowait_time_us 260 258 -2
get_cpu_idle_time_us 260 258 -2
find_lock_later_rq 507 505 -2
enqueue_task_rt 777 775 -2
enqueue_task_dl 2461 2459 -2
dequeue_rt_stack 576 574 -2
menu_select 1492 1489 -3
__dequeue_dl_entity 419 414 -5
init_dl_rq 88 81 -7
Total: Before=25729849, After=25729825, chg -0.00%
Ingo,
On Thu, May 13 2021 at 11:58, Ingo Molnar wrote:
> * Thomas Gleixner <[email protected]> wrote:
>
> You often won't see these effects in the 'size vmlinux' output, because
> function alignment padding reserves usually hide 1-2 byte size improvements
> in generated code.
I'm not that stupid. And I certainly looked where this comes from.
> More importantly, the maintenance benchmark in these cases is not whether a
> change actively helps every architecture we care about - but whether the
> change is a *disadvantage* for them - and it isn't here.
That's clearly documented in the changelogs of these patches, right?
> Changes that primarily benefit one common architecture, while not others,
> are still eligible for upstream merging if they otherwise meet the quality
> threshold and don't hurt the other architectures.
That has been proven by compile testing the relevant architectures as
documented in the changelog, right?
> TL;DR:
>
> This benefits from this series are small, but are far from 'useless churn',
> unless we want to arbitrarily cut off technically valid contributions that
> improve generated code, data structure size and code readability, submitted
> by a long-time contributor who has contributed over 1,300 patches to the
> kernel already, just because we don't think these add up a significant
> enough benefit?
>
> No doubt the quality barrier must be and is higher for smaller changes -
> but this series IMO passed that barrier.
>
> Anyway, I've Cc:-ed Linus and Greg, if you are advocating for some sort of
> cut-off threshold for small but measurable improvements from long-time
> contributors, it should probably be clearly specified & documented in
> Documentation/SubmittingPatches ...
What I'm arguing about is already documented:
Quantify optimizations and trade-offs. If you claim improvements in
performance, memory consumption, stack footprint, or binary size,
include numbers that back them up.
That series fails to provide any of this and it does not matter whether
this comes from a long time contributor or from a newbie.
Long term contributors are not excempt from documented process. In fact
they should lead by example.
If you as a maintainer put different measures on newbies and long-time
contrinbutors then you pretty much have proven the point the UMN people
tried to make (in the wrong way).
Thanks,
tglx
On Thu, May 13 2021 at 11:58, Ingo Molnar wrote:
> * Thomas Gleixner <[email protected]> wrote:
> As to the numbers:
> - /* size: 1704, cachelines: 27, members: 13 */
> - /* sum members: 1696, holes: 1, sum holes: 4 */
> + /* size: 1696, cachelines: 27, members: 13 */
> + /* sum members: 1688, holes: 1, sum holes: 4 */
> /* padding: 4 */
> - /* last cacheline: 40 bytes */
> + /* last cacheline: 32 bytes */
>
> 'struct rt_rq' got shrunk from 1704 bytes to 1696 bytes, an 8 byte
> reduction.
Amazing and it still occupies 27 cache lines
> ffffffffxxxxxxxx: e8 49 8e fb ff call ffffffffxxxxxxxx <nr_iowait_cpu>
> - ffffffffxxxxxxxx: 48 85 c0 test %rax,%rax
>
> ffffffffxxxxxxxx: e8 64 8e fb ff call ffffffffxxxxxxxx <nr_iowait_cpu>
> + ffffffffxxxxxxxx: 85 c0 test %eax,%eax
>
> Note how the 'test %rax,%rax' lost the 0x48 64-bit REX prefix and the
> generated code got one byte shorter from "48 85 c0" to "85 c0".
Which will surely be noticable big time. None of this truly matters
because once the data is in L1 the REX prefix is just noise.
> ( Note, my asm generation scripts filter out some of the noise to make it
> easier to diff generated asm, hence the ffffffffxxxxxxxx placeholder. )
>
> The nr_iowait() function itself got shorter by two bytes as well, due to:
>
> The size of nr_iowait() shrunk from 78 bytes to 76 bytes.
That's important because nr_iowait() is truly a hotpath function...
> The nr_running() function itself got shorter by 2 bytes, due to shorter
> instruction sequences.
along with nr_running() which both feed /proc/stat. The latter feeds
/proc/loadavg as well.
Point well taken.
But looking at the /proc/stat usage there is obviously way bigger fish
to fry.
seq_printf(...., nr_running(), nr_iowait());
which is outright stupid because both functions iterate over CPUs
instead of doing it once, which definitely would be well measurable on
large machines.
But looking at nr_running() and nr_iowait():
nr_running() walks all CPUs in cpu_online_mask
nr_iowait() walks all CPUs in cpu_possible_mask
The latter is because rq::nr_iowait is not transferred to an online CPU
when a CPU goes offline. Oh well.
That aside:
I'm not against the change per se, but I'm disagreeing with patches
which come with zero information, are clearly focussed on one
architecture and obviously nobody bothered to check whether there is an
impact on others.
Thanks,
tglx