Conflict between ide and usb?

Conflict between ide and usb?

[John Coppens]

Hi all.

I installed ueagle-atm, which, by itself, is working fine with an 
MT-810 modem.

The problem I have, is that when I copy a file from a DVD to harddisk,
the internet connection almost dies (it slows down terribly, so much so
that established connections actually disconnect, ping looses packets,
DNS lookup fails, etc). After copy ends, all returns to normal.

The machine is an AMD64, still in x86 mode though. CPU activity is very
low during the copy, so it's not a CPU problem. Kernel is 2.6.20.2. APIC
is enabled. No error message found in either /var/log/messages or syslog.
DVD and HD are on different IDE interfaces.

Can anyone suggest where to start looking for a solution?

Cheers,
John

$ cat interrupts
           CPU0       CPU1       
  0:   58428350    6031800    XT-PIC-XT        timer
  1:     204337          9   IO-APIC-edge      i8042
  6:          0          3   IO-APIC-edge      floppy
  7:    6031308   58426101   IO-APIC-edge      parport0
  9:          0          0   IO-APIC-fasteoi   acpi
 12:     211041    5153326   IO-APIC-edge      i8042
 14:      93535     316012   IO-APIC-edge      ide0
 15:     246534         40   IO-APIC-edge      ide1
 16:       2591    1737056   IO-APIC-fasteoi   libata, HDA Intel
 17:         41          1   IO-APIC-fasteoi   libata, ehci_hcd:usb2
 18:    7717494   19142555   IO-APIC-fasteoi   eth0
 19:          0          0   IO-APIC-fasteoi   eth1
 20:   90833792    3057782   IO-APIC-fasteoi   ohci_hcd:usb1
 21:   22135804          1   IO-APIC-fasteoi   nvidia
 22:          0          0   IO-APIC-fasteoi   saa7130[0]
NMI:          0          0 
LOC:   64460593   64460595 
ERR:          1
MIS:          0

$ lsmod
Module                  Size  Used by
pppoe                   9792  2 
pppox                   3016  1 pppoe
ppp_generic            21396  6 pppoe,pppox
slhc                    5440  1 ppp_generic
br2684                  6596  1 
i2c_viapro              7380  0 
w83627ehf              16140  0 
i2c_isa                 4096  1 w83627ehf
snd_seq_oss            26368  0 
snd_seq_midi_event      5952  1 snd_seq_oss
snd_seq                41232  4 snd_seq_oss,snd_seq_midi_event
snd_seq_device          6284  2 snd_seq_oss,snd_seq
snd_pcm_oss            34912  0 
snd_mixer_oss          13376  4 snd_pcm_oss
ipt_REJECT              3648  1 
ipt_MASQUERADE          2944  1 
ipt_LOG                 5760  1 
xt_limit                2560  1 
xt_state                2368  3 
iptable_filter          2688  1 
iptable_nat             6020  1 
nf_nat                 13740  2 ipt_MASQUERADE,iptable_nat
nf_conntrack_ipv4      11980  5 iptable_nat
nf_conntrack           44168  5
ipt_MASQUERADE,xt_state,iptable_nat,nf_nat,nf_conntrack_ipv4
ip_tables               9700  2 iptable_filter,iptable_nat
ueagle_atm             21032  0 firmware_class          7360  1 ueagle_atm
usbatm                 14464  2 ueagle_atm
atm                    33716  4 br2684,usbatm
usblp                  11136  0 
ide_scsi               13128  0 
tuner                  55272  0 
saa7134               105952  0 
snd_hda_intel          15192  3 
snd_hda_codec         149120  1 snd_hda_intel
video_buf              18500  1 saa7134
compat_ioctl32          1536  1 saa7134
ir_kbd_i2c              6928  1 saa7134
ir_common              26372  2 saa7134,ir_kbd_i2c
ehci_hcd               24332  0 
ohci_hcd               16708  0 
videodev               23488  1 saa7134
snd_pcm                61124  3 snd_pcm_oss,snd_hda_intel,snd_hda_codec
snd_timer              16836  2 snd_seq,snd_pcm
snd                    39268  9
snd_seq_oss,snd_seq,snd_seq_device,snd_pcm_oss,snd_mixer_oss,snd_hda_intel,snd_hda_codec,snd_pcm,snd_timer
snd_page_alloc          7368  2 snd_hda_intel,snd_pcm
sg                     19032  0 i2c_nforce2             4800  0 
nvidia               6817908  22 
v4l2_common            21312  3 tuner,saa7134,videodev
v4l1_compat            12036  2 saa7134,videodev
k8temp                  4544  0 
usbcore               101768  6 ueagle_atm,usbatm,usblp,ehci_hcd,ohci_hcd
ne2k_pci                8736  0 
8390                    7808  1 ne2k_pci
forcedeth              37192  0 
parport_pc             22628  0 
parport                30728  1 parport_pc

Re: Conflict between ide and usb?

[Lee Revell]

On 3/16/07, John Coppens <john@jcoppens.com> wrote:
> The problem I have, is that when I copy a file from a DVD to harddisk,
> the internet connection almost dies (it slows down terribly, so much so
> that established connections actually disconnect, ping looses packets,
> DNS lookup fails, etc). After copy ends, all returns to normal.

Sounds like DMA starvation, it's been seen in the past with some SATA
controllers.  Presumably vendors do it to improve benchmark scores.

You can confirm it by forcing the hard drive and/or DVD drive to a
lower speed if your BIOS allows it.

Lee

Re: Conflict between ide and usb?

[John Coppens]

On Fri, 16 Mar 2007 15:52:12 -0400
"Lee Revell" <rlrevell@joe-job.com> wrote:

> On 3/16/07, John Coppens <john@jcoppens.com> wrote:
> > The problem I have, is that when I copy a file from a DVD to harddisk,
> > the internet connection almost dies (it slows down terribly, so much
> > so that established connections actually disconnect, ping looses
> > packets, DNS lookup fails, etc). After copy ends, all returns to
> > normal.
> 
> Sounds like DMA starvation, it's been seen in the past with some SATA
> controllers.  Presumably vendors do it to improve benchmark scores.

Hi Lee.

In this case SATA isn't used, as the copy is from IDE to IDE. But I
doubt bandwidth has anything to do with it. The DVD transfer rate rarely 
reaches 3.3 MB/s - which by itself isn't normal.

> You can confirm it by forcing the hard drive and/or DVD drive to a
> lower speed if your BIOS allows it.

No... I didn't find any way to lower the speed.

Thanks
John

Re: Conflict between ide and usb?

[Mark Lord]

John Coppens wrote:
> Hi all.
> 
> I installed ueagle-atm, which, by itself, is working fine with an 
> MT-810 modem.
> 
> The problem I have, is that when I copy a file from a DVD to harddisk,
> the internet connection almost dies (it slows down terribly, so much so
> that established connections actually disconnect, ping looses packets,
> DNS lookup fails, etc). After copy ends, all returns to normal.
> 
> The machine is an AMD64, still in x86 mode though. CPU activity is very
> low during the copy, so it's not a CPU problem. Kernel is 2.6.20.2. APIC
> is enabled. No error message found in either /var/log/messages or syslog.
> DVD and HD are on different IDE interfaces.
> 
> Can anyone suggest where to start looking for a solution?
..
>  16:       2591    1737056   IO-APIC-fasteoi   libata, HDA Intel
>  17:         41          1   IO-APIC-fasteoi   libata, ehci_hcd:usb2
..

It pretty much has to be an IRQ sharing issue,
most likely in the USB driver side of the house.

I don't know enough about the USB drivers really,
but the interrupt routine inside the ueagle-atm.c file
doesn't seem too "sharing friendly".  But perhaps that
part is handled by the layers above (?).

That's where I'd start looking if I were you.

-ml

Re: Conflict between ide and usb?

[Lennart Sorensen]

On Fri, Mar 16, 2007 at 05:12:53PM -0300, John Coppens wrote:
> No... I didn't find any way to lower the speed.

Do you even have DMA enabled for the DVD drive?  Without it it will be
very slow and painful for the CPU.  I also have noticed that many fast
(16x) DVD writers must have an 80 wire cable or they won't work
correctly and do nasty things to the system.

Check /proc/ide/hdX/settings|grep dma and see if it says 1 0 1 or 0 0 1.

--
Len Sorensen

Re: Conflict between ide and usb?

[Mark Lord]

Mark Lord wrote:
> John Coppens wrote:
>
>> I installed ueagle-atm, which, by itself, is working fine with an 
>> MT-810 modem.
>>
>> The problem I have, is that when I copy a file from a DVD to harddisk,
>> the internet connection almost dies (it slows down terribly, so much so
>> that established connections actually disconnect, ping looses packets,
>> DNS lookup fails, etc). After copy ends, all returns to normal.
>>
>> The machine is an AMD64, still in x86 mode though. CPU activity is very
>> low during the copy, so it's not a CPU problem. Kernel is 2.6.20.2. APIC
>> is enabled. No error message found in either /var/log/messages or syslog.
>> DVD and HD are on different IDE interfaces.
>>
>> Can anyone suggest where to start looking for a solution?
> ..
>>  16:       2591    1737056   IO-APIC-fasteoi   libata, HDA Intel
>>  17:         41          1   IO-APIC-fasteoi   libata, ehci_hcd:usb2
> ..
> 
> It pretty much has to be an IRQ sharing issue,
> most likely in the USB driver side of the house.

Oh, wait a minute.. that's not the correct IRQ, is it?

Your ueagle-atm must be on this one:

 20:   90833792    3057782   IO-APIC-fasteoi   ohci_hcd:usb1

In which case, it's not likely to be an interrupt sharing issue.
Can you confirm this?

Also, try moving your ueagle-atm to one of the ehci usb ports
and see how that affects things.

Re: Conflict between ide and usb?

[John Coppens]

On Fri, 16 Mar 2007 16:34:15 -0400
lsorense@csclub.uwaterloo.ca (Lennart Sorensen) wrote:

> Do you even have DMA enabled for the DVD drive?  Without it it will be
> very slow and painful for the CPU.  I also have noticed that many fast
> (16x) DVD writers must have an 80 wire cable or they won't work
> correctly and do nasty things to the system.
> 
> Check /proc/ide/hdX/settings|grep dma and see if it says 1 0 1 or 0 0 1.

I _knew_ I enabled the DMA, but it seems that is switches off when I start
the transfer:

hdc: DMA disabled
hdc: ide_intr: huh? expected NULL handler on exit
hdc: ATAPI reset complete
ISO 9660 Extensions: RRIP_1991A
VFS: busy inodes on changed media.

This is on hdc, which is the DVD. Harddisk has DMA on.

John

Re: Conflict between ide and usb?

[John Coppens]

On Fri, 16 Mar 2007 16:34:15 -0400
lsorense@csclub.uwaterloo.ca (Lennart Sorensen) wrote:

> I also have noticed that many fast
> (16x) DVD writers must have an 80 wire cable or they won't work
> correctly and do nasty things to the system.

That was the problem! This is the first machine I bought assembled, 
and I hadn't noticed that the CD/DVD cables were the old 40 wire
cables. I do hope this is the last of the problems.

Thanks guys! Changed the cable, and speed went up from 3.3 to 19MB/s
(probably limited by the older harddrive too)... Thas was spectacular!
Of course, now DMA didn't switch off either.

Cheers,
John