Il 04/09/2014 14:24, Wanpeng Li ha scritto: > On Thu, Sep 04, 2014 at 01:33:10PM +0200, Paolo Bonzini wrote: >> Il 04/09/2014 11:53, Paolo Bonzini ha scritto: >>> Il 03/09/2014 20:25, Chris J Arges ha scritto: >>>> <snip> >>>>> I'm not sure about the reason for the warp, but indeed the offset and >>>>> uptime match (I'll check them against the trace tomorrow) so it's "just" >>>>> that the VM's TSC base is not taken into account correctly. >>>>> >>>>> Can you gather another trace with the problematic patch reverted? >>>>> >>>>> Paolo >>>>> >>>> >>>> Here is the third trace running with 0d3da0d2 reverted from the latest >>>> kvm queue branch 11cc9ea3: >>>> >>>> http://people.canonical.com/~arges/kvm/trace-3.dat.xz >>> >>> Thanks! And---yay!---I reproduced it on another machine. >> >> And my bisection landed on the merge of the timer branch (commit >> e7fda6c4c3c1a7d6996dd75fd84670fa0b5d448f). Here is the log: >> >> $ git bisect bad origin/master >> $ git bisect good v3.16 >> $ git bisect good kvm-3.17-1 # 42cbc04fd3b5e3f9b011bf9fa3ce0b3d1e10b58b >> >> good [ 6929.863545] loaded kvm module (v3.17-rc1-158-g451fd72219dd) >> bad [ 6971.932790] loaded kvm module (for-linus) >> bad [ 7216.073128] loaded kvm module (v3.16-6426-gae045e245542) >> good [ 7286.198948] loaded kvm module (v3.16-3283-g53ee983378ff) >> good [ 7350.534060] loaded kvm module (v3.16-rc7-1668-gaef4f5b6db65) >> good [ 7439.037038] loaded kvm module (v3.16-4006-g91c2ff7708d4) >> good [ 7481.188637] loaded kvm module (v3.16-rc6-450-g7ba3c21c17d0) >> bad [ 7535.292730] loaded kvm module (v3.16-4635-ge7fda6c4c3c1) >> good [ 7589.722691] loaded kvm module (v3.16-rc5-110-g9b0fd802e8c0) >> good [ 7630.286418] loaded kvm module (v3.16-4467-ged5c41d30ef2) >> good [ 7712.470986] loaded kvm module (v3.16-rc1-35-g885d078bfe92) >> good [ 7763.443626] loaded kvm module (v3.16-rc1-381-g1b0733837a9b) >> good [ 7825.497414] loaded kvm module (v3.16-rc5-116-g7806f60e1d20) >> good [ 7893.174056] loaded kvm module (v3.16-rc1-384-gc6f1224573c3) >> >> This means that: >> >> - Tomasz's patch (commit 0d3da0d26e3c3515997c99451ce3b0ad1a69a36c) is >> fine, it just enables the (wrong) master clock more often >> >> - the failure is within that branch. >> >> I then cherry-picked Tomasz's patch during a new bisection, and landed >> on one of my original suspects: >> >> commit cbcf2dd3b3d4d990610259e8d878fc8dc1f17d80 >> Author: Thomas Gleixner <tglx@xxxxxxxxxxxxx> >> Date: Wed Jul 16 21:04:54 2014 +0000 >> >> x86: kvm: Make kvm_get_time_and_clockread() nanoseconds based >> >> Convert the relevant base data right away to nanoseconds instead of >> doing the conversion on every readout. Reduces text size by 160 >> bytes. >> >> Signed-off-by: Thomas Gleixner <tglx@xxxxxxxxxxxxx> >> Cc: Gleb Natapov <gleb@xxxxxxxxxx> >> Cc: kvm@xxxxxxxxxxxxxxx >> Acked-by: Paolo Bonzini <pbonzini@xxxxxxxxxx> >> Signed-off-by: John Stultz <john.stultz@xxxxxxxxxx> >> > > Yes, I also look into this bad commit recently which lead to guest hang > after live migration or after local save/restore. Thanks for the report! Wanpeng, can you test and/or review the patch I just posted ("[PATCH] KVM: x86: fix kvmclock breakage from timers branch merge")? Paolo -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
