Hi,
I have a couple of AMD64 machines with onboard sound devices running
and they all have one thing in common : flaky sound when one is
doing some heavy disk I/O on the SATA disks. Interesting enough
/proc/interrupts shows that libata is using the same IRQ as the sound
devices :
model name : AMD Athlon(tm) 64 X2 Dual Core Processor 6000+
stepping : 3
cpu MHz : 3014.679
cache size : 1024 KB
# cat /proc/interrupts
CPU0 CPU1
0: 55869 154467838 IO-APIC-edge timer
8: 0 0 IO-APIC-edge rtc
9: 0 0 IO-APIC-level acpi
16: 1453 499686 IO-APIC-level libata, NVidia CK8S
17: 0 0 IO-APIC-level ehci_hcd:usb1
18: 27633 69978987 IO-APIC-level eth0
NMI: 10718 12112
LOC: 154533874 154533852
ERR: 0
MIS: 0
model name : AMD Opteron(tm) Processor 246
stepping : 10
cpu MHz : 1991.355
cache size : 1024 KB
# cat /proc/interrupts
CPU0 CPU1
0: 18108 19856365 IO-APIC-edge timer
1: 16 7915 IO-APIC-edge i8042
3: 0 56 IO-APIC-edge serial
4: 0 56 IO-APIC-edge serial
8: 0 0 IO-APIC-edge rtc
9: 0 0 IO-APIC-level acpi
12: 346 62269 IO-APIC-edge i8042
14: 0 190 IO-APIC-edge ide0
15: 1 189 IO-APIC-edge ide1
16: 827 358917 IO-APIC-level libata, AMD AMD8111
17: 0 18 IO-APIC-level ohci_hcd:usb1, ohci_hcd:usb2
18: 555 231470 IO-APIC-level HiSax, nvidia
19: 15179 3489796 IO-APIC-level eth0
NMI: 211 511
LOC: 19871720 19871628
ERR: 0
MIS: 0
Kernel is 2.6.15,
Preemption Model is (Preemptible Kernel (Low-Latency Desktop)), Memory
model is (Flat Memory), [*] Preempt The Big Kernel Lock is switched on.
Timer frequency is (1000 HZ) or (100 HZ), but changing this value is
of no influence. One of the machines is a genuine dual Opteron machine
but i'm rather disappointed with the NUMA capabilities of
the 2.6.15 kernel. It does do NUMA, but thats all it does, it doesn't
add anything compared to a SMP kernel with NUMA switched off.
Does one really need the PREEMPT_RT approach to get rock solid
sound as described here ? :
"A realtime preemption overview"
by Paul McKenney, August 10, 2005
http://lwn.net/Articles/146861/
How about giving your sound device a proper seperate IRQ number?
At least libata should like eth0 have its own kernel resources.
Regards,
Robert
PS. please also cc: to my email address.
--
Robert M. Stockmann - RHCE
Network Engineer - UNIX/Linux Specialist
crashrecovery.org [email protected]
Robert M. Stockmann wrote:
> Hi,
>
> I have a couple of AMD64 machines with onboard sound devices running
> and they all have one thing in common : flaky sound when one is
> doing some heavy disk I/O on the SATA disks. Interesting enough
> /proc/interrupts shows that libata is using the same IRQ as the sound
> devices :
>
> model name : AMD Athlon(tm) 64 X2 Dual Core Processor 6000+
> stepping : 3
> cpu MHz : 3014.679
> cache size : 1024 KB
>
> # cat /proc/interrupts
> CPU0 CPU1
> 0: 55869 154467838 IO-APIC-edge timer
> 8: 0 0 IO-APIC-edge rtc
> 9: 0 0 IO-APIC-level acpi
> 16: 1453 499686 IO-APIC-level libata, NVidia CK8S
> 17: 0 0 IO-APIC-level ehci_hcd:usb1
> 18: 27633 69978987 IO-APIC-level eth0
> NMI: 10718 12112
> LOC: 154533874 154533852
> ERR: 0
> MIS: 0
What is the behavior of a recent kernel?
Is your SATA interface set to legacy mode in BIOS?
Jeff
On Wed, 27 Aug 2008, Jeff Garzik wrote:
> Date: Wed, 27 Aug 2008 20:59:35 -0400
> From: Jeff Garzik <[email protected]>
> To: Robert M. Stockmann <[email protected]>
> Cc: [email protected]
> Subject: Re: libata, Sound on same IRQ : flaky sound
>
> Robert M. Stockmann wrote:
> > Hi,
> >
> > I have a couple of AMD64 machines with onboard sound devices running
> > and they all have one thing in common : flaky sound when one is
> > doing some heavy disk I/O on the SATA disks. Interesting enough
> > /proc/interrupts shows that libata is using the same IRQ as the sound
> > devices :
> >
> > model name : AMD Athlon(tm) 64 X2 Dual Core Processor 6000+
> > stepping : 3
> > cpu MHz : 3014.679
> > cache size : 1024 KB
> >
> > # cat /proc/interrupts
> > CPU0 CPU1
> > 0: 55869 154467838 IO-APIC-edge timer
> > 8: 0 0 IO-APIC-edge rtc
> > 9: 0 0 IO-APIC-level acpi
> > 16: 1453 499686 IO-APIC-level libata, NVidia CK8S
> > 17: 0 0 IO-APIC-level ehci_hcd:usb1
> > 18: 27633 69978987 IO-APIC-level eth0
> > NMI: 10718 12112
> > LOC: 154533874 154533852
> > ERR: 0
> > MIS: 0
>
> What is the behavior of a recent kernel?
>
> Is your SATA interface set to legacy mode in BIOS?
>
the BIOS supports raid0, but the sata disks are running standalone
as /dev/sda /dev/sdb etc.
installing a recent kernel is not that easy, as the required
upgrade of udev will break the initscripts and a lot of installed
services and even the functionality of the KDE desktop. Installing
a recent distro is not desirable as they resemble me only
of a Open Source Vista distro, where it takes months to clean
up the unwanted garbage and bloat inside. Its preposterous that
a upgrade to a recent kernel mandates the installation of a recent
distro. I run Mandrake 10.1 with kernel 2.6.15 and some applications
upgraded like firefox, mplayer and k3b etc.
Robert
--
Robert M. Stockmann - RHCE
Network Engineer - UNIX/Linux Specialist
crashrecovery.org [email protected]
Robert M. Stockmann <stock <at> stokkie.net> writes:
>
> the BIOS supports raid0, but the sata disks are running standalone
> as /dev/sda /dev/sdb etc.
>
> installing a recent kernel is not that easy, as the required
> upgrade of udev will break the initscripts and a lot of installed
> services and even the functionality of the KDE desktop. Installing
> a recent distro is not desirable as they resemble me only
> of a Open Source Vista distro, where it takes months to clean
> up the unwanted garbage and bloat inside. Its preposterous that
> a upgrade to a recent kernel mandates the installation of a recent
> distro. I run Mandrake 10.1 with kernel 2.6.15 and some applications
> upgraded like firefox, mplayer and k3b etc.
>
> Robert
If you don't want bloat AND you want a recent kernel AND lots to fiddle around
and figure out for yourself, install PLD Linux :) No, seriously. Last time I
tried, a usable graphic environment (GNOME that is, dunno about KDE) fitted in
300-500 MiB ON DISK. (without graphical env. fitted in 150MB ON DISK). Try that.
el es
Robert M. Stockmann wrote:
> Hi,
>
> I have a couple of AMD64 machines with onboard sound devices running
> and they all have one thing in common : flaky sound when one is
> doing some heavy disk I/O on the SATA disks. Interesting enough
> /proc/interrupts shows that libata is using the same IRQ as the sound
> devices :
>
> model name : AMD Athlon(tm) 64 X2 Dual Core Processor 6000+
> stepping : 3
> cpu MHz : 3014.679
> cache size : 1024 KB
>
> # cat /proc/interrupts
> CPU0 CPU1
> 0: 55869 154467838 IO-APIC-edge timer
> 8: 0 0 IO-APIC-edge rtc
> 9: 0 0 IO-APIC-level acpi
> 16: 1453 499686 IO-APIC-level libata, NVidia CK8S
> 17: 0 0 IO-APIC-level ehci_hcd:usb1
> 18: 27633 69978987 IO-APIC-level eth0
> NMI: 10718 12112
> LOC: 154533874 154533852
> ERR: 0
> MIS: 0
>
> model name : AMD Opteron(tm) Processor 246
> stepping : 10
> cpu MHz : 1991.355
> cache size : 1024 KB
>
> # cat /proc/interrupts
> CPU0 CPU1
> 0: 18108 19856365 IO-APIC-edge timer
> 1: 16 7915 IO-APIC-edge i8042
> 3: 0 56 IO-APIC-edge serial
> 4: 0 56 IO-APIC-edge serial
> 8: 0 0 IO-APIC-edge rtc
> 9: 0 0 IO-APIC-level acpi
> 12: 346 62269 IO-APIC-edge i8042
> 14: 0 190 IO-APIC-edge ide0
> 15: 1 189 IO-APIC-edge ide1
> 16: 827 358917 IO-APIC-level libata, AMD AMD8111
> 17: 0 18 IO-APIC-level ohci_hcd:usb1, ohci_hcd:usb2
> 18: 555 231470 IO-APIC-level HiSax, nvidia
> 19: 15179 3489796 IO-APIC-level eth0
> NMI: 211 511
> LOC: 19871720 19871628
> ERR: 0
> MIS: 0
>
> Kernel is 2.6.15,
>
> Preemption Model is (Preemptible Kernel (Low-Latency Desktop)), Memory
> model is (Flat Memory), [*] Preempt The Big Kernel Lock is switched on.
> Timer frequency is (1000 HZ) or (100 HZ), but changing this value is
> of no influence. One of the machines is a genuine dual Opteron machine
> but i'm rather disappointed with the NUMA capabilities of
> the 2.6.15 kernel. It does do NUMA, but thats all it does, it doesn't
> add anything compared to a SMP kernel with NUMA switched off.
>
> Does one really need the PREEMPT_RT approach to get rock solid
> sound as described here ? :
>
> "A realtime preemption overview"
> by Paul McKenney, August 10, 2005
> http://lwn.net/Articles/146861/
>
> How about giving your sound device a proper seperate IRQ number?
> At least libata should like eth0 have its own kernel resources.
That's an issue with the way the motherboard IRQ lines are wired.
There's nothing the kernel can do about it.
Normally I wouldn't expect that to make a big difference though..
You'd really have to try a newer kernel first in order to get much help,
though. That's a pretty ancient kernel. Quite likely the situation is
improved in newer versions.
>
> Regards,
>
> Robert
> PS. please also cc: to my email address.
On Wed, Aug 27, 2008 at 6:30 PM, Robert M. Stockmann <[email protected]> wrote:
<snip>
> installing a recent kernel is not that easy, as the required
> upgrade of udev will break the initscripts and a lot of installed
> services and even the functionality of the KDE desktop. Installing
> a recent distro is not desirable as they resemble me only
> of a Open Source Vista distro, where it takes months to clean
> up the unwanted garbage and bloat inside. Its preposterous that
> a upgrade to a recent kernel mandates the installation of a recent
> distro. I run Mandrake 10.1 with kernel 2.6.15 and some applications
> upgraded like firefox, mplayer and k3b etc.
>
Not to divert this into a distro discussion, but you should look into
Gentoo (http://www.gentoo.org) - it offers the kind of customization
and flexibility (as well as pick and choose update capabilities) that
it sounds like you are interested in.
-James
> Robert
> --
> Robert M. Stockmann - RHCE
> Network Engineer - UNIX/Linux Specialist
> crashrecovery.org [email protected]
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>
On Thu, 28 Aug 2008, Robert Hancock wrote:
> Date: Thu, 28 Aug 2008 08:41:08 -0600
> From: Robert Hancock <[email protected]>
> To: Robert M. Stockmann <[email protected]>
> Cc: [email protected]
> Subject: Re: libata, Sound on same IRQ : flaky sound
>
> Robert M. Stockmann wrote:
> > Hi,
> >
> > I have a couple of AMD64 machines with onboard sound devices running
> > and they all have one thing in common : flaky sound when one is
> > doing some heavy disk I/O on the SATA disks. Interesting enough
> > /proc/interrupts shows that libata is using the same IRQ as the sound
> > devices :
> >
> > model name : AMD Athlon(tm) 64 X2 Dual Core Processor 6000+
> > stepping : 3
> > cpu MHz : 3014.679
> > cache size : 1024 KB
> >
> > # cat /proc/interrupts
> > CPU0 CPU1
> > 0: 55869 154467838 IO-APIC-edge timer
> > 8: 0 0 IO-APIC-edge rtc
> > 9: 0 0 IO-APIC-level acpi
> > 16: 1453 499686 IO-APIC-level libata, NVidia CK8S
> > 17: 0 0 IO-APIC-level ehci_hcd:usb1
> > 18: 27633 69978987 IO-APIC-level eth0
> > NMI: 10718 12112
> > LOC: 154533874 154533852
> > ERR: 0
> > MIS: 0
> >
> > model name : AMD Opteron(tm) Processor 246
> > stepping : 10
> > cpu MHz : 1991.355
> > cache size : 1024 KB
> >
> > # cat /proc/interrupts
> > CPU0 CPU1
> > 0: 18108 19856365 IO-APIC-edge timer
> > 1: 16 7915 IO-APIC-edge i8042
> > 3: 0 56 IO-APIC-edge serial
> > 4: 0 56 IO-APIC-edge serial
> > 8: 0 0 IO-APIC-edge rtc
> > 9: 0 0 IO-APIC-level acpi
> > 12: 346 62269 IO-APIC-edge i8042
> > 14: 0 190 IO-APIC-edge ide0
> > 15: 1 189 IO-APIC-edge ide1
> > 16: 827 358917 IO-APIC-level libata, AMD AMD8111
> > 17: 0 18 IO-APIC-level ohci_hcd:usb1, ohci_hcd:usb2
> > 18: 555 231470 IO-APIC-level HiSax, nvidia
> > 19: 15179 3489796 IO-APIC-level eth0
> > NMI: 211 511
> > LOC: 19871720 19871628
> > ERR: 0
> > MIS: 0
> >
> > Kernel is 2.6.15,
> >
> > Preemption Model is (Preemptible Kernel (Low-Latency Desktop)), Memory
> > model is (Flat Memory), [*] Preempt The Big Kernel Lock is switched on.
> > Timer frequency is (1000 HZ) or (100 HZ), but changing this value is
> > of no influence. One of the machines is a genuine dual Opteron machine
> > but i'm rather disappointed with the NUMA capabilities of
> > the 2.6.15 kernel. It does do NUMA, but thats all it does, it doesn't
> > add anything compared to a SMP kernel with NUMA switched off.
> >
> > Does one really need the PREEMPT_RT approach to get rock solid
> > sound as described here ? :
> >
> > "A realtime preemption overview"
> > by Paul McKenney, August 10, 2005
> > http://lwn.net/Articles/146861/
> >
> > How about giving your sound device a proper seperate IRQ number?
> > At least libata should like eth0 have its own kernel resources.
>
> That's an issue with the way the motherboard IRQ lines are wired.
> There's nothing the kernel can do about it.
That sounds rather strange to me, as IRQ line 16 is a virtual
IRQ as part of IO-APIC.
> Normally I wouldn't expect that to make a big difference though..
fact is that when copying a iso from one SATA disk to the
other results in flaky sound when playing online internet radio.
Even starting firefox for the 1st time after booting results
in flaky sound for a short time.
> You'd really have to try a newer kernel first in order to get much help,
> though. That's a pretty ancient kernel. Quite likely the situation is
> improved in newer versions.
I sure would like todo that. But why doesn't the linux-kernel community
know howto pull virtual IRQ lines apart, as part of configuring
a linux kernel? In the days of 386/486 cpu's, IRQ's of several
add-on cards could be adjusted manually by jumpers. Why is there not
such a thing for virtual IRQ's inside the IO-APIC system?
The libata driver should like eth0 on IRQ 19 have its own IRQ.
Robert
--
Robert M. Stockmann - RHCE
Network Engineer - UNIX/Linux Specialist
crashrecovery.org [email protected]
Robert M. Stockmann wrote:
> On Thu, 28 Aug 2008, Robert Hancock wrote:
> > Robert M. Stockmann wrote:
> > > How about giving your sound device a proper seperate IRQ number?
> > > At least libata should like eth0 have its own kernel resources.
> >
> > That's an issue with the way the motherboard IRQ lines are wired.
> > There's nothing the kernel can do about it.
>
> That sounds rather strange to me, as IRQ line 16 is a virtual
> IRQ as part of IO-APIC.
The IRQ isn't virtual; in IO-APIC mode, all PCI interrupts (PCI cards
and internal PCI devices) are hard-wired to certain interrupt vectors.
Regards,
Clemens