Messages in this thread Patch in this message | | | Date | Thu, 28 May 2015 22:16:35 -0400 | Subject | Re: [PATCH] x86: skip delays during SMP initialization similar to Xen | From | Len Brown <> |
| |
>> this time can be reduced by 7% (113 ms) by deleting announce_cpu()
While the KERN_DEBUG output is fast, accounce_cpu() uses KERN_INFO, which goes to the (serial) console by default.
One would expect it to drain at about 10 bits/byte / 115200 baud = 87us/byte. I measured some vanilla printk's via KERN_INFO on this server, and they ran at 100us/byte. I guess that isn't so far off -- an effective 100,000 baud...
announce_cpu() is actually 8 lines @ 112 bytes/line = 896 bytes, on this box. @ 100usec/byte, that is 89ms -- about double your estimate of 50ms; and closer to my original observation that deleting announce_cpu() altogether saved 113ms.
I suppose somebody at 9600 baud would be less delighted -- 10x slower would be 896 bytes * 10 bits/byte / 9600 = 933ms, yeas, almost a ful lsecond, just for announce_cpu() output.
So I guess the question is if announce_cpu() really needs to be KERN_INFO, or if it can be KERN_DEBUG. I think that KERN_DEBUG is sufficient, though if it were up to me, I'd delete it entirely.
That is because the success case prints this on the console already:
[ 1.959675] x86: Booted up 4 nodes, 120 CPUs [ 1.965202] smpboot: Total of 120 processors activated (671148.72 BogoMIPS)
Success also prints the pretty announce_cpu() list in dmesg, and can still do so if "debug" is on the cmdline...
But most importantly, the error case would trip output in native_cpu_up(), due to this upstream patch:
commit feef1e8ecbadf24f8e6829c935df8683cabae41b Author: Igor Mammedov <imammedo@redhat.com> Date: Thu Jun 5 15:42:44 2014 +0200
x86/smpboot: Log error on secondary CPU wakeup failure at ERR level
If system is running without debug level logging, it will not log error if do_boot_cpu() failed to wakeup AP. It may lead to silent AP bringup failures at boot time. Change message level to KERN_ERR to make error visible to user as it's done on other architectures.
Signed-off-by: Igor Mammedov <imammedo@redhat.com> Acked-by: Toshi Kani <toshi.kani@hp.com> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/1401975765-22328-3-git-send-email-imammedo@redhat.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index 2988f69..ae2fd975 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -918,7 +918,7 @@ int native_cpu_up(unsigned int cpu, struct task_struct *tidle)
err = do_boot_cpu(apicid, cpu, tidle); if (err) { - pr_debug("do_boot_cpu failed %d\n", err); + pr_err("do_boot_cpu failed(%d) to wakeup CPU#%u\n", err, cpu); return -EIO; }
cheers, Len Brown, Intel Open Source Technology Center
| |