Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 20 Sep 2008 08:44:11 -0700 (PDT)
From:      kg5nm <kg5nm@comcast.net>
To:        freebsd-questions@freebsd.org
Subject:   Re: irq19 uhci interrupts taking ~100% of one core?
Message-ID:  <19586581.post@talk.nabble.com>
In-Reply-To: <a174d0600809072254j1ad277e8jdfb638f7e1b71a11@mail.gmail.com>
References:  <a174d0600809071607g308a91c0qbc5c114e40c6146e@mail.gmail.com> <a174d0600809072254j1ad277e8jdfb638f7e1b71a11@mail.gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help

I think this is a SATA driver problem on amd64 builds.  I have the same
problem (3 usable cores due to this!).  I'm too am running a custom kernel
but this was also happening on GENERIC.

If you temporarily disable all USB drivers, you should see the SATA ATAPI
driver pop up on 'vmstat -i' instead of UHCI.  IRQ 19 is a multi-use IRQ and
vmstat reports one IRQ grabber.

An additional clue: I unplugged 3 of my 4 SATA drives and the 'IRQ shower'
actually increased about 25%!

I'm considering posting this to the -CURRENT folks to see if this is being
worked, it's pretty painful to lose 25% of my quad-core just to this
interrupt shower; FYI - I'm calling it a 'shower' since it's not bad enough
for the kernel to flag it as a storm.   :)

- Gary (kg5nm@comcast.net)


Scott Gasch wrote:
> 
> Replying to my own question with more data.
> Previously I had been running my own kernel; I was curious if the problem
> would reproduce with a GENERIC kernel.  It does but the symptoms are
> slightly different.  The same irq is firing too often and consuming nearly
> all of one core.  But the driver that is associated with the interrupt is
> different -- it's fwohci0:
> 
> interrupt                          total       rate
> irq6: fdc0                             1          0
> irq17: mskc0 dc0                  313242         13
> irq18: skc0 uhci2*             124475451       5540
> irq19: fwohci0+++              957875379      42638
> irq23: uhci3 ehci1                  1145          0
> cpu0: timer                     44458513       1979
> cpu1: timer                     44448875       1978
> cpu3: timer                     43393901       1931
> cpu2: timer                     43393921       1931
> Total                         1258360428      56014
> 
> 
> This makes me start to wonder if this is not a problem with irq19 (the
> PIC?)
> and not one particular device / driver.  I'm not sure how to make dig
> deeper
> here, any help greatly appreciated.
> 
> Thx,
> Scott
> 
> 
> On Sun, Sep 7, 2008 at 4:07 PM, Scott Gasch <scott.gasch@gmail.com> wrote:
> 
>> Hi,
>> I'm running freebsd 7.0-RELEASE-p4 on a 4-core amd64 box.  nearly 100% of
>> 1
>> cpu is constantly being used handling irq19: uhci4 interrupts.  This
>> seems
>> to happen both with and without any USB devices plugged in:
>>
>> vmstat -i
>> interrupt                          total       rate
>> irq1: atkbd0                           5          0
>> irq6: fdc0                             1          0
>> irq17: mskc0 dc0                 1180547         18
>> irq18: skc0 uhci2*             163250699       2512
>> irq19: uhci4++                3187989508      49072
>> irq23: uhci3 ehci1                    31          0
>> cpu0: timer                    129208570       1988
>> cpu1: timer                    129208457       1988
>> cpu2: timer                    125750147       1935
>> cpu3: timer                    125750122       1935
>> Total                         3862338087      59452
>>
>> dmesg
>> uhci4: <UHCI (generic) USB controller> port 0xa400-0xa41f irq 19 at
>> device
>> 29.1
>> on pci0
>> uhci4: [GIANT-LOCKED]
>> uhci4: [ITHREAD]
>>
>> Any idea what's going on here and/or how to fix this?
>>
>> Thx,
>> Scott
>>
>>
> _______________________________________________
> freebsd-questions@freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-questions
> To unsubscribe, send any mail to
> "freebsd-questions-unsubscribe@freebsd.org"
> 
> 

-- 
View this message in context: http://www.nabble.com/irq19-uhci-interrupts-taking-%7E100--of-one-core--tp19363669p19586581.html
Sent from the freebsd-questions mailing list archive at Nabble.com.




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19586581.post>