How to debug excessive TMR and IPI interrupts?

Discussion in 'OS X Mavericks (10.9)' started by RoyG, Nov 5, 2014.

  1. RoyG macrumors newbie

    Jul 24, 2011
    I am having a problem with my mid 2010 MPB that overnight went into super slow performance. The two last things that happened that I know of prior to this problem is someone bumping the headphones plug so the system crashed. (Weird thing, really.) The other thing was to install Silverlight to watch some video on demand. I don't think Silverlight has anything to do with the problem.

    I ran a full sysdiagnose, noticing in 'powermetrics.txt' that TMR interrupts seems a bit high. Same with IPI. I am guessing that TMR is TiMeR, and IPI is InterProcessing Interrupts. None of which is helpful without knowing exactly what is causing such a high amount of interrupts:

    **** Interrupt distribution ****

    CPU 0:
    Vector 0x46(SMC): 0.97 interrupts/sec
    Vector 0x51(OHC1/EHC1/LUS2/LUS0): 48.26 interrupts/sec
    Vector 0x52(OHC2/SATA/LSI0/Z000): 36.68 interrupts/sec
    Vector 0x71(HDEF): 22.20 interrupts/sec
    Vector 0x75(IGPU): 283.77 interrupts/sec
    Vector 0x76(ARPT): 356.16 interrupts/sec
    Vector 0xdd(TMR): 25696.28 interrupts/sec
    Vector 0xde(IPI): 1467.09 interrupts/sec
    CPU 1:
    Vector 0xdd(TMR): 28957.66 interrupts/sec
    Vector 0xde(IPI): 1686.19 interrupts/sec

    How do I debug further into this problem?

    Unrelated to this, I also notice an exaggerated amount of entries in the system log like these:

    Nov 2 16:02:54 x-mbp[219]: tcp_connection_destination_prepare_complete 82 connectx to <ipv6 address>#443 failed: 65 - No route to host
    Nov 2 16:02:54 x-mbp[219]: tcp_connection_destination_prepare_complete 81 connectx to <ipv6 address>#443 failed: 65 - No route to host
    Nov 2 16:02:54 x-mbp[219]: tcp_connection_handle_destination_prepare_complete 82 failed to connect
    Nov 2 16:02:54 x-mbp[219]: tcp_connection_handle_destination_prepare_complete 81 failed to connect
    Nov 2 16:02:54 x-mbp[219]: tcp_connection_handle_destination_prepare_complete 82 failed to connect

    Check the time on these... There are several hundreds of them generated each second. Constantly. That could make a lot of pressure on the system too, I suppose.

    Any help appreciated.

    Thank you.
  2. RoyG, Nov 6, 2014
    Last edited: Nov 6, 2014

    RoyG thread starter macrumors newbie

    Jul 24, 2011
    Ok, a follow-up:

    I decided to upgrade to Yosemite. The interrupt levels are now at something that to me seem more normal. Still a bit high, but I don't think it is the cause of these problems. TMR is now around 500-700, and IPI right around 1000.

    The problem is, though, that installing Yosemite did not help for the performance issues. What I have tried to do now, is to run the first batch of cleanup articles in an app called ONYX. I also ran the clear font cache thing there. I tried making a new login account. Reset Safari. (As well as you can do in version 8.) RAM test ok. Turned off login items. No extensions at all in Safari. No 3rd party kexts. No smart errors, no file system errors. Fixed permissions. Installed a smcFanControl to try to speed the fan to max in case it is a thermal issue, but on this mbp 13" mid 2010 model it didn't spin up anything. It did show that the temperature at "exhaust" was around 74 degrees at medium-heavy load, so that seemed normal.

    I also checked the log around "Previous shutdown reason: 3"-messages, and there were nothing obvious to find. (Shutdown reason 3 is forced/unexpected shutdown.)

    There are still a lot of the previously mentioned log entries. Also there are a lot of these for various applications:

    Nov 6 00:38:14 x-MBP WindowServer[155]: common_reenable_update: UI updates were finally reenabled by application "Activity Monitor" after 1.43 seconds (server forcibly re-enabled them after 1.00 seconds)

    Plus XPC and Open Directory error messages, but not as frequent.

    Also several:

    Nov 6 01:14:57 x-MBP kernel[0]: PM response took 746 ms (363,

    Nov 6 18:55:35 x-MBP[1] ([425]): Service exited with abnormal code: 1

    All in all, it seems WebKit has a serious issue, but how does that affect a system in this way??

    Video and audio in any app stutters because of the performance issues too.


    Oh I discovered this too in WebKits disfavour:

    Nov 6 00:00:17 x-MBP kernel[0]: process[1306] thread 79954 caught burning CPU! It used more than 50% CPU (Actual recent usage: 63%) over 180 seconds. thread lifetime cpu usage 123.104076 seconds, (116.047339 user, 7.056737 system) ledger info: balance: 90004798422 credit: 120232897853 debit: 30228099431 limit: 90000000000 (50%) period: 180000000000 time since last refill (ns): 142085130895

    Seems these belong to the previous log too:

    Nov 6 00:00:17 x-MBP[1] ([1318]): Endpoint has been activated through legacy launch(3) APIs. Please switch to XPC or bootstrap_check_in():
    Nov 6 00:00:17 x-MBP ReportCrash[1318]: Invoking spindump for pid=1306 thread=79954 percent_cpu=63 duration=143 because of excessive cpu utilization
    Nov 6 00:00:21 x-MBP spindump[1266]: Saved cpu_resource.diag report for version 10600 (10600.1.25) to /Library/Logs/DiagnosticReports/

Share This Page