Messages in this thread | | | Date | Thu, 7 May 2015 12:23:51 +0200 | From | Ingo Molnar <> | Subject | Re: [PATCH] x86: skip delays during SMP initialization similar to Xen |
| |
* Len Brown <lenb@kernel.org> wrote:
> > So I really like this, as it nicely side-steps the 'when should we > > do the legacy delays' issue by flagging on x2apic support. > > > > If anyone has objections, please holler. > > We should have no delays even for many processors that lack x2apic.
Yes, agreed, but this looks like a good (and safe) first step: especially systems that have large CPU counts tend to have x2apic support - and that is where such a change matters most.
I'm not against doing a zero delay approach either, if it can be triggered robustly.
Here's what the boot time looks like on a 120 CPUs system, with the patch applied:
[ 0.558947] x86: Booting SMP configuration: [ 0.563375] .... node #0, CPUs: #1 #2 #3 #4 #5 #6 #7 #8 #9 #10 #11 #12 #13 #14 [ 0.644851] .... node #1, CPUs: #15 #16 #17 #18 #19 #20 #21 #22 #23 #24 #25 #26 #27 #28 #29 [ 0.830474] .... node #2, CPUs: #30 #31 #32 #33 #34 #35 #36 #37 #38 #39 #40 #41 #42 #43 #44 [ 1.016357] .... node #3, CPUs: #45 #46 #47 #48 #49 #50 #51 #52 #53 #54 #55 #56 #57 #58 #59 [ 1.202342] .... node #0, CPUs: #60 #61 #62 #63 #64 #65 #66 #67 #68 #69 #70 #71 #72 #73 #74 [ 1.283864] .... node #1, CPUs: #75 #76 #77 #78 #79 #80 #81 #82 #83 #84 #85 #86 #87 #88 #89 [ 1.397131] .... node #2, CPUs: #90 #91 #92 #93 #94 #95 #96 #97 #98 #99 #100 #101 #102 #103 #104 [ 1.510417] .... node #3, CPUs: #105 #106 #107 #108 #109 #110 #111 #112 #113 #114 #115 #116 #117 #118 #119 [ 1.620967] x86: Booted up 4 nodes, 120 CPUs [ 1.625928] smpboot: Total of 120 processors activated (672866.16 BogoMIPS)
1.1 seconds to boot 120 CPUs, 10.8 seconds to hit init, that's an entirely reasonable runtime I think.
It was 20+ seconds before that, 10+ seconds for the SMP bootup sequence.
Thanks,
Ingo
| |