timeout waiting for DMA

timeout waiting for DMA

[Jason Rappleye]

Hi,

I'm running kernel 2.4.2 on an SGI 1100 (dual PIIIs) with a Serverworks
III LE based motherboard. The disk is a Seagate ST330630A. The disk has
DMA enabled at boot time :

hda: ST330630A, ATA DISK drive
hda: 59777640 sectors (30606 MB) w/2048KiB Cache, CHS=3720/255/63, UDMA(33)

(also verified using hdparm)

but after a while (eg partway through running bonnie with a 1GB file) I
get the following errors:

Feb 24 22:51:02 nash2 kernel: hda: timeout waiting for DMA 
Feb 24 22:51:02 nash2 kernel: ide_dmaproc: chipset supported ide_dma_timeout 
func only: 14 
Feb 24 22:51:02 nash2 kernel: hda: irq timeout: status=0x58 { DriveReady
SeekComplete DataRequest }
<repeats a few times>
Feb 24 22:51:32 nash2 kernel: hda: DMA disabled

I can reenable DMA without any problems, but after some additional disk
activity (eg running bonnie again), the error occurs again. 

Additional information on my hardware is given below. Any suggestions on
how this can be resolved?

Thanks,

Jason

hdparm -i /dev/hda

/dev/hda:

 Model=ST330630A, FwRev=3.21, SerialNo=3CK0JDFE
 Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs RotSpdTol>.5% }
 RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=0
 BuffType=0(?), BuffSize=2048kB, MaxMultSect=16, MultSect=off
 DblWordIO=no, OldPIO=2, DMA=yes, OldDMA=2
 CurCHS=16383/16/63, CurSects=-66060037, LBA=yes, LBAsects=59777640
 tDMA={min:120,rec:120}, DMA modes: mword0 mword1 mword2 
 IORDY=on/off, tPIO={min:240,w/IORDY:120}, PIO modes: mode3 mode4 
 UDMA modes: mode0 mode1 *mode2 mode3 mode4 

Relevant portion of /proc/pci:

  Bus  0, device  15, function  1:
    IDE interface: PCI device 1166:0211 (ServerWorks) (rev 0).
      Master Capable.  Latency=64.  
      I/O at 0x3080 [0x308f].

Relevant portion of lspci -v:

00:0f.1 IDE interface: Relience Computer: Unknown device 0211 (prog-if 8a
[Master SecP PriP])
	Flags: bus master, medium devsel, latency 64
	I/O ports at 3080 [size=16]


--
Jason Rappleye
rappleye@buffalo.edu		
http://www.ccr.buffalo.edu/jason.htm

Re: timeout waiting for DMA

[Daniela Engert]

Hi Jason!

On Mon, 26 Feb 2001 15:41:04 -0500 (EST), Jason Rappleye wrote:

>I'm running kernel 2.4.2 on an SGI 1100 (dual PIIIs) with a Serverworks
>III LE based motherboard. The disk is a Seagate ST330630A. The disk has
>DMA enabled at boot time :

>hda: ST330630A, ATA DISK drive
>hda: 59777640 sectors (30606 MB) w/2048KiB Cache, CHS=3720/255/63, UDMA(33)

>but after a while (eg partway through running bonnie with a 1GB file) I
>get the following errors:

>Feb 24 22:51:02 nash2 kernel: hda: timeout waiting for DMA 
>Feb 24 22:51:02 nash2 kernel: ide_dmaproc: chipset supported ide_dma_timeout 
>func only: 14 
>Feb 24 22:51:02 nash2 kernel: hda: irq timeout: status=0x58 { DriveReady
>SeekComplete DataRequest }
><repeats a few times>
>Feb 24 22:51:32 nash2 kernel: hda: DMA disabled

>I can reenable DMA without any problems, but after some additional disk
>activity (eg running bonnie again), the error occurs again. 

>Additional information on my hardware is given below. Any suggestions on
>how this can be resolved?

Reduce the IDE channel speed to UltraDMA mode 1 or less.

Ciao,
  Dani

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Daniela Engert, systems engineer at MEDAV GmbH
Gräfenberger Str. 34, 91080 Uttenreuth, Germany
Phone ++49-9131-583-348, Fax ++49-9131-583-11

Re: timeout waiting for DMA

[Jasmeet Sidhu]

Iv'e seen this on a system with 36" IDE ata/100 cables.  I have the same 
exact system using 24" IDE ultra ata/100 cables and I get no such 
errors.  These two systems are exactly the same and are using the same 
drivers, same hardware.
The only difference is the type of cable used.  Could this be due to bad 
cables as well?  (I know tha CRC errors are caused by bad cables, but this 
is the only difference in the two boxes that I have).

Keep the cables as short as possible, 18" is supposed to be the limit.

And please post again if you successfully troubleshoot this problem.

Jasmeet.

At 07:35 AM 2/27/2001 +0100, Daniela Engert wrote:
>Hi Jason!
>
>On Mon, 26 Feb 2001 15:41:04 -0500 (EST), Jason Rappleye wrote:
>
> >I'm running kernel 2.4.2 on an SGI 1100 (dual PIIIs) with a Serverworks
> >III LE based motherboard. The disk is a Seagate ST330630A. The disk has
> >DMA enabled at boot time :
>
> >hda: ST330630A, ATA DISK drive
> >hda: 59777640 sectors (30606 MB) w/2048KiB Cache, CHS=3720/255/63, UDMA(33)
>
> >but after a while (eg partway through running bonnie with a 1GB file) I
> >get the following errors:
>
> >Feb 24 22:51:02 nash2 kernel: hda: timeout waiting for DMA
> >Feb 24 22:51:02 nash2 kernel: ide_dmaproc: chipset supported 
> ide_dma_timeout
> >func only: 14
> >Feb 24 22:51:02 nash2 kernel: hda: irq timeout: status=0x58 { DriveReady
> >SeekComplete DataRequest }
> ><repeats a few times>
> >Feb 24 22:51:32 nash2 kernel: hda: DMA disabled
>
> >I can reenable DMA without any problems, but after some additional disk
> >activity (eg running bonnie again), the error occurs again.
>
> >Additional information on my hardware is given below. Any suggestions on
> >how this can be resolved?
>
>Reduce the IDE channel speed to UltraDMA mode 1 or less.
>
>Ciao,
>   Dani
>
>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>Daniela Engert, systems engineer at MEDAV GmbH
>Gräfenberger Str. 34, 91080 Uttenreuth, Germany
>Phone ++49-9131-583-348, Fax ++49-9131-583-11
>
>
>-
>To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
>the body of a message to majordomo@vger.kernel.org
>More majordomo info at  http://vger.kernel.org/majordomo-info.html
>Please read the FAQ at  http://www.tux.org/lkml/


- - -
Jasmeet Sidhu
Unix Systems Administrator
ArrayComm, Inc.
jsidhu@arraycomm.com
www.arraycomm.com

[RECAP] timeout waiting for DMA

[Jason Rappleye]

Hi,

I'd like to thank those who responded to my message. Here's a recap of the
suggestions and results. My original email appears at the end. For those
of you that are getting errors that look like this:

kernel: hda: timeout waiting for DMA 
kernel: ide_dmaproc: chipset supported ide_dma_timeout func only: 14 
kernel: hda: irq timeout: status=0x58 { DriveReady SeekComplete DataRequest }
kernel: hda: DMA disabled

PLEAE read this and try the suggestions listed before posting here
again. I'm sure it'll save everyone some time and give the developers more
valuable feedback than "it doesn't work."

-Mark Hahn <hahn@coffee.psychology.mcmaster.ca> :

> Andre says that the problem is that (apparently all) ide busmaster
> engines share a common flaw: when they starve for PCI access,
> they glitch.  resetting them is the only solution, and in his latest
> patches, he has code that does this (without losing the use-dma bit).

So I applied ide.2.4.1-p8-01172001 to 2.4.2, with a reject in the hpt
source, which _I_ didn't care about (we have Serverworks boards) and
enabled the "Attempt to HACK around chipsets that TIMEOUT" option (under
ATA works in progress). This didn't work for me. YMMV.

-Jasmeet Sidhu <jsidhu@arraycomm.com>
> Iv'e seen this on a system with 36" IDE ata/100 cables.  I have the same 
> exact system using 24" IDE ultra ata/100 cables and I get no such 
> errors.  These two systems are exactly the same and are using the same 
> drivers, same hardware.
> The only difference is the type of cable used.  Could this be due to bad 
> cables as well?  (I know tha CRC errors are caused by bad cables, but
> this is the only difference in the two boxes that I have).

> Keep the cables as short as possible, 18" is supposed to be the limit.

Our cables are ~ 12" long, and are sliced up every 5 lines or so to fit
"neatly" in our 1U cases. So I don't think this is contributing to the
problem, at least not in our case. Still, a good thing to check.

-Daniela Engert <dani@ngrt.de>

> Reduce the IDE channel speed to UltraDMA mode 1 or less.

Both with and without Andre's patch, udma mode 1 and 0 both exhibited the
same problem. However, switching to mulitword DMA 2 (hdparm -X34 /dev/hdX) 
worked just fine WITH Andre's patch. Without the patch we still experienced 
the problem. Hmm...I just got an email saying that the new BIOS for our
machines disables udma in lieu of PIO mode 4 because it "doesn't work" on
our boards (Serverworks LE III chipset). Great. So you might have better
luck with udma 1 and 0 than we are. 

So, at the moment, _I_ personally suggest a) applying Andre's latest ide
patch, b) trying a lower udma setting or multiword 2 DMA, and c) checking
your cables. You're not really "losing" performance by dropping down to a
different udma mode or mword2, since it's certainly better than pio 4 :-)

Andre, I can replicate this behavior 100% of the time, if you need someone
to do testing let me know, I'm more than happy to help out in any way I
can. Though I might not be a good candidate due to the damn Serverworks
chipset :-(

Hope this helps,

j

On Mon, 26 Feb 2001, Jason Rappleye wrote:

> 
> Hi,
> 
> I'm running kernel 2.4.2 on an SGI 1100 (dual PIIIs) with a Serverworks
> III LE based motherboard. The disk is a Seagate ST330630A. The disk has
> DMA enabled at boot time :
> 
> hda: ST330630A, ATA DISK drive
> hda: 59777640 sectors (30606 MB) w/2048KiB Cache, CHS=3720/255/63, UDMA(33)
> 
> (also verified using hdparm)
> 
> but after a while (eg partway through running bonnie with a 1GB file) I
> get the following errors:
> 
> Feb 24 22:51:02 nash2 kernel: hda: timeout waiting for DMA 
> Feb 24 22:51:02 nash2 kernel: ide_dmaproc: chipset supported ide_dma_timeout 
> func only: 14 
> Feb 24 22:51:02 nash2 kernel: hda: irq timeout: status=0x58 { DriveReady
> SeekComplete DataRequest }
> <repeats a few times>
> Feb 24 22:51:32 nash2 kernel: hda: DMA disabled
> 
> I can reenable DMA without any problems, but after some additional disk
> activity (eg running bonnie again), the error occurs again. 
> 
> Additional information on my hardware is given below. Any suggestions on
> how this can be resolved?
> 
> Thanks,
> 
> Jason
> 
> hdparm -i /dev/hda
> 
> /dev/hda:
> 
>  Model=ST330630A, FwRev=3.21, SerialNo=3CK0JDFE
>  Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs RotSpdTol>.5% }
>  RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=0
>  BuffType=0(?), BuffSize=2048kB, MaxMultSect=16, MultSect=off
>  DblWordIO=no, OldPIO=2, DMA=yes, OldDMA=2
>  CurCHS=16383/16/63, CurSects=-66060037, LBA=yes, LBAsects=59777640
>  tDMA={min:120,rec:120}, DMA modes: mword0 mword1 mword2 
>  IORDY=on/off, tPIO={min:240,w/IORDY:120}, PIO modes: mode3 mode4 
>  UDMA modes: mode0 mode1 *mode2 mode3 mode4 
> 
> Relevant portion of /proc/pci:
> 
>   Bus  0, device  15, function  1:
>     IDE interface: PCI device 1166:0211 (ServerWorks) (rev 0).
>       Master Capable.  Latency=64.  
>       I/O at 0x3080 [0x308f].
> 
> Relevant portion of lspci -v:
> 
> 00:0f.1 IDE interface: Relience Computer: Unknown device 0211 (prog-if 8a
> [Master SecP PriP])
> 	Flags: bus master, medium devsel, latency 64
> 	I/O ports at 3080 [size=16]
> 
> 
> --
> Jason Rappleye
> rappleye@buffalo.edu		
> http://www.ccr.buffalo.edu/jason.htm
> 
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 

--
Jason Rappleye
rappleye@buffalo.edu		
http://www.ccr.buffalo.edu/jason.htm

timeout waiting for DMA...

[Jason Gillis]

Hi all,

   I'm seeing the "timeout waiting for DMA" problems that I've noticed
several others are running up against.  I have a UDMA-66 drive that I
have _full_ of data.  After some amount of disk activity (some can be
very little [a few minutes], or a long time [hours]), I get the
following on the console (copied the best I can):

ide_dmaproc: chipset supported ide_dma_timeout func only: 14
hdc: irq timeout: status=0x50 { DriveReady SeekComplete }
hda: timeout waiting for DMA
ide_dmaproc: chipset supported ide_dma_timeout func only: 14
hda: DMA disabled
hdb: DMA disabled
hdc: timeout waiting for DMA
ide_dmaproc: chipset supported ide_dma_timeout func only: 14
hdc: irq timeout: status=0x50 { DriveReady SeekComplete }
ide0: reset timed out, status=0xd0
hdb: status timeout: status=0xd0 { Busy }
hdb: drive not ready for command
hdc: timeout waiting for DMA

   The IDE controller in question is a Promise 20262 Ultra/66 controller.

   I'd appreciate any suggestions that I can get at this point.  I've
spent lots of time building the 2.4.2 kernel, and that didn't seem to
help.  I had previously been using the 2.2.17 kernel with the latest
IDE patch from people/hedrick to no avail.

   At this point, I'd be more than happy to test out new IDE drivers,
if that's what things take.

   Included below is some info from my system.  I'm not sure what's
useful, so I probably put more here than I needed to.  If there's
something missing, please let me know.  Also, I'm not subscribed to
the list, so a CC back to me would be appreciated!

   Jason
   jgillis@acm.org

===========================

[root@excedrin jgillis]# cat /proc/dma
 4: cascade
 5: GFA1/CS4231 record
 7: CS4231 playback
[root@excedrin jgillis]# cat /proc/interrupts
           CPU0       CPU1
  0:     635966     747681    IO-APIC-edge  timer
  1:          2          0    IO-APIC-edge  keyboard
  2:          0          0          XT-PIC  cascade
 10:       2936       2947   IO-APIC-level  ide0, ide1
 11:          0          0    IO-APIC-edge  InterWave
 12:       9391       9560   IO-APIC-level  eth1
 14:      16727      17082   IO-APIC-level  eth0
NMI:    1383582    1383582
LOC:    1383678    1383677
ERR:          0

===========================

[root@excedrin jgillis]# uname -a
Linux excedrin 2.4.2 #1 SMP Sun Mar 18 10:10:31 PST 2001 i686 unknown


===========================
Stuff from dmesg:


ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
PIIX3: IDE controller on PCI bus 00 dev 39
PCI: Enabling device 00:07.1 (0000 -> 0001)
PIIX3: chipset revision 0
PIIX3: not 100% native mode: will probe irqs later
PIIX3: neither IDE port enabled (BIOS)
PDC20262: IDE controller on PCI bus 00 dev 70
PDC20262: chipset revision 1
PDC20262: not 100% native mode: will probe irqs later
PDC20262: (U)DMA Burst Bit ENABLED Primary PCI Mode Secondary PCI Mode.
    ide0: BM-DMA at 0xac00-0xac07, BIOS settings: hda:DMA, hdb:DMA
    ide1: BM-DMA at 0xac08-0xac0f, BIOS settings: hdc:DMA, hdd:pio
hda: FUJITSU M1623TAU, ATA DISK drive
hdb: Maxtor 86480D6, ATA DISK drive
hdc: WDC WD450AA, ATA DISK drive
ide0 at 0x9c00-0x9c07,0xa002 on irq 10
ide1 at 0xa400-0xa407,0xa802 on irq 10
hda: 3324996 sectors (1702 MB) w/128KiB Cache, CHS=3298/16/63, DMA
hdb: 12594960 sectors (6449 MB) w/256KiB Cache, CHS=13328/15/63, UDMA(33)
hdc: 87930864 sectors (45021 MB) w/2048KiB Cache, CHS=87233/16/63, UDMA(66)
Partition check:
 hda: [PTBL] [824/64/63] hda1 hda2
 hdb: [PTBL] [784/255/63] hdb1
 hdc: [PTBL] [5473/255/63] hdc1


===========================

[root@excedrin jgillis]# cat /proc/pci
...
  Bus  0, device  14, function  0:
    Unknown mass storage controller: Promise Technology, Inc. 20262 (rev 1).
      IRQ 10.
      Master Capable.  Latency=64.
      I/O at 0x9c00 [0x9c07].
      I/O at 0xa000 [0xa003].
      I/O at 0xa400 [0xa407].
      I/O at 0xa800 [0xa803].
      I/O at 0xac00 [0xac3f].
      Non-prefetchable 32 bit memory at 0xe2000000 [0xe201ffff].

===========================

[root@excedrin jgillis]# /sbin/hdparm -i /dev/hda

/dev/hda:

 Model=FUJITSU M1623TAU, FwRev=5243, SerialNo=00002003
 Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs RotSpdTol>.5% }
 RawCHS=3298/16/63, TrkSize=0, SectSize=0, ECCbytes=4
 BuffType=0(?), BuffSize=128kB, MaxMultSect=16, MultSect=off
 DblWordIO=no, maxPIO=2(fast), DMA=yes, maxDMA=2(fast)
 CurCHS=3298/16/63, CurSects=-1176502222, LBA=yes
 LBA CHS=824/64/63 Remapping, LBA=yes, LBAsects=3324996
 tDMA={min:120,rec:120}, DMA modes: sword0 sword1 sword2 mword0 mword1
*mword2
 IORDY=yes, tPIO={min:240,w/IORDY:120}, PIO modes: mode3 mode4

[root@excedrin jgillis]# /sbin/hdparm -i /dev/hdb

/dev/hdb:

 Model=Maxtor 86480D6, FwRev=NAVX171F, SerialNo=L6066EZA
 Config={ Fixed }
 RawCHS=13328/15/63, TrkSize=0, SectSize=0, ECCbytes=20
 BuffType=3(DualPortCache), BuffSize=256kB, MaxMultSect=16, MultSect=off
 DblWordIO=no, maxPIO=2(fast), DMA=yes, maxDMA=2(fast)
 CurCHS=13328/15/63, CurSects=789577920, LBA=yes
 LBA CHS=784/255/63 Remapping, LBA=yes, LBAsects=12594960
 tDMA={min:120,rec:120}, DMA modes: mword0 mword1 mword2
 IORDY=on/off, tPIO={min:120,w/IORDY:120}, PIO modes: mode3 mode4
 UDMA modes: mode0 mode1 *mode2

[root@excedrin jgillis]# /sbin/hdparm -i /dev/hdc

/dev/hdc:

 Model=WDC WD450AA, FwRev=10.09K11, SerialNo=WD-WMA2E1006236
 Config={ HardSect NotMFM HdSw>15uSec SpinMotCtl Fixed DTR>5Mbs FmtGapReq }
 RawCHS=16383/16/63, TrkSize=57600, SectSize=600, ECCbytes=40
 BuffType=3(DualPortCache), BuffSize=2048kB, MaxMultSect=16, MultSect=off
 DblWordIO=no, maxPIO=2(fast), DMA=yes, maxDMA=0(slow)
 CurCHS=16383/16/63, CurSects=-66060037, LBA=yes
 LBA CHS=1023/256/63 Remapping, LBA=yes, LBAsects=87930864
 tDMA={min:120,rec:120}, DMA modes: mword0 mword1 mword2
 IORDY=on/off, tPIO={min:120,w/IORDY:120}, PIO modes: mode3 mode4
 UDMA modes: mode0 mode1 mode2

[root@excedrin jgillis]#

===========================