preempt_count rework -v3
Status from this message: Queued for 3.13, thank you!
Get series with b4
b4 am [email protected]Versions newest first
Patches in v1 Applied
| # | Subject and files | Size | Replies | Applied |
|---|---|---|---|---|
| 1/11 | x86: Use asm goto to implement better modify_and_test() functionsarch/ arch/ arch/ +2 more | +58 | 4 replies | |
| 2/11 | sched, rcu: Make RCU use resched_cpu()kernel/ kernel/ | +16 | 5 replies | |
| 3/11 | sched: Remove {set,clear}_need_reschedinclude/ | +13 | 0 replies | |
| 4/11 | sched, idle: Fix the idle polling state logicarch/ drivers/ drivers/ +3 more | +91 | 0 replies | |
| 5/11 | sched: Introduce preempt_count accessor functionsinclude/ init/ kernel/ +3 more | +24 | 0 replies | |
| 6/11 | sched: Add NEED_RESCHED to the preempt_countinclude/ include/ kernel/ +2 more | +60 | 0 replies | |
| 7/11 | sched, arch: Create asm/preempt.harch/ arch/ arch/ +29 more | +75 | 0 replies | |
| 8/11 | sched: Create more preempt_count accessorsinclude/ include/ kernel/ | +18 | 0 replies | |
| 9/11 | sched: Extract the basic add/sub preempt_count modifiersarch/ arch/ include/ +7 more | +112 | 0 replies | |
| 10/11 | sched, x86: Provide a per-cpu preempt_count implementationarch/ arch/ arch/ +6 more | +128 | 0 replies | |
| 11/11 | sched, x86: Optimize the preempt_schedule() callarch/ arch/ arch/ +3 more | +100 | 1 reply |
Reviews trailers per patch
None of the 11 patches has a Reviewed-by or Acked-by of their own; the series has one for all of them.
| Patch | David S. Miller | Geert Uytterhoeven | Tony Luck |
|---|---|---|---|
| Series | A-b | A-b | T-b |
| 1/11 | |||
| 2/11 | |||
| 3/11 | |||
| 4/11 | |||
| 5/11 | |||
| 6/11 | |||
| 7/11 | |||
| 8/11 | |||
| 9/11 | |||
| 10/11 | |||
| 11/11 |
Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes. Scroll the table sideways for all 3 reviewers.
Cover letter [PATCH 00/11] preempt_count rework -v3
These patches optimize preempt_enable by firstly folding the preempt and need_resched tests into one -- this should work for all architectures. And secondly by providing per-arch preempt_count implementations; with x86 using per-cpu preempt_count for fastest access.
These patches have been boot tested on CONFIG_PREEMPT=y x86_64 and survive building a x86_64-defconfig kernel.
text data bss filename
11387014 1454776 1187840 defconfig-build/vmlinux.before 11352294 1454776 1187840 defconfig-build/vmlinux.after
void kick_process(struct task_struct *p)
{
int cpu;
preempt_disable();
cpu = task_cpu(p);
if ((cpu != smp_processor_id()) && task_curr(p))
smp_send_reschedule(cpu);
preempt_enable();
}before:
ffffffff81073f00 <kick_process>:
ffffffff81073f00: 55 push %rbp I ffffffff81073f01: 65 48 8b 04 25 f0 b7 mov %gs:0xb7f0,%rax ffffffff81073f08: 00 00 I ffffffff81073f0a: 83 80 44 e0 ff ff 01 addl $0x1,-0x1fbc(%rax) ffffffff81073f11: 48 89 e5 mov %rsp,%rbp ffffffff81073f14: 48 8b 47 08 mov 0x8(%rdi),%rax ffffffff81073f18: 8b 50 18 mov 0x18(%rax),%edx ffffffff81073f1b: 65 8b 04 25 1c b0 00 mov %gs:0xb01c,%eax ffffffff81073f22: 00 ffffffff81073f23: 39 c2 cmp %eax,%edx ffffffff81073f25: 74 1b je ffffffff81073f42 <kick_process+0x42> ffffffff81073f27: 89 d1 mov %edx,%ecx ffffffff81073f29: 48 c7 c0 40 2c 01 00 mov $0x12c40,%rax ffffffff81073f30: 48 8b 0c cd 00 25 cc mov -0x7e33db00(,%rcx,8),%rcx ffffffff81073f37: 81 ffffffff81073f38: 48 3b bc 08 00 08 00 cmp 0x800(%rax,%rcx,1),%rdi ffffffff81073f3f: 00 ffffffff81073f40: 74 2e je ffffffff81073f70 <kick_process+0x70> D ffffffff81073f42: 65 48 8b 04 25 f0 b7 mov %gs:0xb7f0,%rax ffffffff81073f49: 00 00 D ffffffff81073f4b: 83 a8 44 e0 ff ff 01 subl $0x1,-0x1fbc(%rax) D ffffffff81073f52: 48 8b 80 38 e0 ff ff mov -0x1fc8(%rax),%rax D ffffffff81073f59: a8 08 test $0x8,%al D ffffffff81073f5b: 75 02 jne ffffffff81073f5f <kick_process+0x5f> ffffffff81073f5d: 5d pop %rbp ffffffff81073f5e: c3 retq D ffffffff81073f5f: e8 9c 6f 75 00 callq ffffffff817caf00 <preempt_schedule> ffffffff81073f64: 5d pop %rbp ffffffff81073f65: c3 retq ffffffff81073f66: 66 2e 0f 1f 84 00 00 nopw %cs:0x0(%rax,%rax,1) ffffffff81073f6d: 00 00 00 ffffffff81073f70: 89 d7 mov %edx,%edi ffffffff81073f72: ff 15 a8 99 ba 00 callq *0xba99a8(%rip) # ffffffff81c1d920 <smp_ops+0x20> ffffffff81073f78: eb c8 jmp ffffffff81073f42 <kick_process+0x42> ffffffff81073f7a: 66 0f 1f 44 00 00 nopw 0x0(%rax,%rax,1)
after:
ffffffff8106f5e0 <kick_process>:
ffffffff8106f5e0: 55 push %rbp I ffffffff8106f5e1: 65 ff 04 25 e0 b7 00 incl %gs:0xb7e0 ffffffff8106f5e8: 00 ffffffff8106f5e9: 48 89 e5 mov %rsp,%rbp ffffffff8106f5ec: 48 8b 47 08 mov 0x8(%rdi),%rax ffffffff8106f5f0: 8b 50 18 mov 0x18(%rax),%edx ffffffff8106f5f3: 65 8b 04 25 1c b0 00 mov %gs:0xb01c,%eax ffffffff8106f5fa: 00 ffffffff8106f5fb: 39 c2 cmp %eax,%edx ffffffff8106f5fd: 74 1b je ffffffff8106f61a <kick_process+0x3a> ffffffff8106f5ff: 89 d1 mov %edx,%ecx ffffffff8106f601: 48 c7 c0 40 2c 01 00 mov $0x12c40,%rax ffffffff8106f608: 48 8b 0c cd 00 25 cc mov -0x7e33db00(,%rcx,8),%rcx ffffffff8106f60f: 81 ffffffff8106f610: 48 3b bc 08 00 08 00 cmp 0x800(%rax,%rcx,1),%rdi ffffffff8106f617: 00 ffffffff8106f618: 74 26 je ffffffff8106f640 <kick_process+0x60> D ffffffff8106f61a: 65 ff 0c 25 e0 b7 00 decl %gs:0xb7e0 ffffffff8106f621: 00 D ffffffff8106f622: 74 0c je ffffffff8106f630 <kick_process+0x50> ffffffff8106f624: 5d pop %rbp ffffffff8106f625: c3 retq ffffffff8106f626: 66 2e 0f 1f 84 00 00 nopw %cs:0x0(%rax,%rax,1) ffffffff8106f62d: 00 00 00 D ffffffff8106f630: e8 fb b4 f9 ff callq ffffffff8100ab30 <___preempt_schedule> ffffffff8106f635: 5d pop %rbp ffffffff8106f636: c3 retq ffffffff8106f637: 66 0f 1f 84 00 00 00 nopw 0x0(%rax,%rax,1) ffffffff8106f63e: 00 00 ffffffff8106f640: 89 d7 mov %edx,%edi ffffffff8106f642: ff 15 d8 e2 ba 00 callq *0xbae2d8(%rip) # ffffffff81c1d920 <smp_ops+0x20> ffffffff8106f648: eb d0 jmp ffffffff8106f61a <kick_process+0x3a> ffffffff8106f64a: 66 0f 1f 44 00 00 nopw 0x0(%rax,%rax,1)