HDIO_SET_DMA failed: Operation not permitted

dlittlebear · n00b Joined: 16 Apr 2004 Posts: 4

I know this is not the first nor the last on this subject.

I have looked over google.com/linux and searched these forums.
I thought at first it was due to the lack of an 80 conductor cable, replaced the one that was in there with 2 different ones and still same problem.

The PC.
AMD 1.4ghz
/dev/hdd is the one I am working with
abit KT7A (KT133 chipset)

Gentoo 2004.0
kernel 2.6.5
I started with stage1 if it matters.

# hdparm -d1 /dev/hdd

/dev/hdd:
setting using_dma to 1 (on)
HDIO_SET_DMA failed: Operation not permitted
using_dma = 0 (off)

# hdparm -tT /dev/hdd

/dev/hdd:
Timing buffer-cache reads: 396 MB in 2.00 seconds = 197.54 MB/sec
Timing buffered disk reads: 6 MB in 3.06 seconds = 1.96 MB/sec

# lspci | grep IDE
00:07.1 IDE interface: VIA Technologies, Inc. VT82C586A/B/VT82C686/A/B/VT823x/A/C PIPC Bus Master IDE (rev 06)

# cat /usr/src/linux/.config | grep DMA | grep -v "#"
CONFIG_GENERIC_ISA_DMA=y
CONFIG_BLK_DEV_IDEDMA_PCI=y
CONFIG_IDEDMA_PCI_AUTO=y
CONFIG_BLK_DEV_ADMA=y
CONFIG_BLK_DEV_IDEDMA=y
CONFIG_IDEDMA_AUTO=y

# cat /usr/src/linux/.config | grep VIA82CXX | grep -v "#"
CONFIG_BLK_DEV_VIA82CXXX=y

# cat /usr/src/linux/.config | grep IDE | grep -v "#"
CONFIG_IDE=y
CONFIG_BLK_DEV_IDE=y
CONFIG_BLK_DEV_IDEDISK=y
CONFIG_IDEDISK_MULTI_MODE=y
CONFIG_BLK_DEV_IDECD=y
CONFIG_IDE_TASKFILE_IO=y
CONFIG_IDE_GENERIC=y
CONFIG_BLK_DEV_IDEPCI=y
CONFIG_IDEPCI_SHARE_IRQ=y
CONFIG_BLK_DEV_IDEDMA_PCI=y
CONFIG_IDEDMA_PCI_AUTO=y
CONFIG_BLK_DEV_IDEDMA=y
CONFIG_IDEDMA_AUTO=y

# dmesg | grep ide
BIOS-provided physical RAM map:
CPU: After generic identify, caps: 0183fbff c1c7fbff 00000000 00000000
CPU: After vendor identify, caps: 0183fbff c1c7fbff 00000000 00000000
ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
ide1 at 0x170-0x177,0x376 on irq 15
/dev/ide/host0/bus0/target0/lun0: p1
/dev/ide/host0/bus0/target1/lun0: p1
/dev/ide/host1/bus0/target1/lun0: p1 p2 p3
VP_IDE: port 0x01f0 already claimed by ide0
VP_IDE: port 0x0170 already claimed by ide1

EDIT

# dmesg | grep IDE
Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
VP_IDE: IDE controller at PCI slot 0000:00:07.1
VP_IDE: chipset revision 6
VP_IDE: not 100% native mode: will probe irqs later
VP_IDE: VIA vt82c686b (rev 40) IDE UDMA100 controller on pci0000:00:07.1
VP_IDE: port 0x01f0 already claimed by ide0
VP_IDE: port 0x0170 already claimed by ide1
VP_IDE: neither IDE port enabled (BIOS)

with CAPS IDE, noticed it is not seeing either IDE port.

I can post the full dmesg if needed.

any suggestions appreciated. I hope this is enough information. I read through posts on this same problem for other distros too.

firephoto · Posted: Fri Apr 16, 2004 6:36 am Post subject:

I have the same chipset on a KR7A-RAID board. Kernel 2.6.5 here.

alari · Posted: Fri Apr 16, 2004 10:26 am Post subject:

I have a Epox 8RDA3i motherboard witch Quantum Fireball lct15 15gb HDD Ultra ATA 66.When i booted Gentoo 1.4 cd i was able to set the dma but now i use 2.6.4-ck and ia cannot set dma, it says operation no permitted?
Do I have to enable something in kernel, special IDE dma controllers ?

dlittlebear · n00b Joined: 16 Apr 2004 Posts: 4

thoughts · Posted: Fri Apr 16, 2004 10:55 am Post subject:

I'm pretty sure this is a software/kernel issue. I had Slackware on my box for almost a year, both hard drives using DMA. I just installed Gentoo 2004.0 the other day, same box but on the second hard drive, and I got "operation not permitted" when I tried to set the DMA on either drive. That was kernel 2.4.22... yesterday I installed kernel 2.6.3, and DMA works fine on both drives. Reboot back into 2.4.22, DMA is gone. No changes to the BIOS or hardware.

So I don't know which kernel option it is, but I'm pretty sure it's that, and not a BIOS setting.

-Anthony
http://nodivisions.com/

Titan. · Posted: Fri Apr 16, 2004 1:02 pm Post subject:

Are you sure you have VIA82CXXX chipset support enabled under "Device Drivers --> ATA/ATAPI/MFM/RLL support"? I had the same problem on my 8K3A+...

rebecca · Posted: Fri Apr 16, 2004 3:09 pm Post subject:

It might not be a case of just having the VIA driver in there but also removing the generic support. From the look of dlittlebear's dmesg output, the generic driver is kicking in befor the VIA one and taking control. Recompiling the kernel and deselcting 'generic/default IDE chipset support' and 'Generic PCI IDE Chipset Support' should allow the VIA driver to take over. Worked for me, anyway. Hope it helps.

Titan. · Posted: Fri Apr 16, 2004 3:25 pm Post subject:

I have both "Generic PCI bus-master DMA", "Generic PCI IDE Chipset" and "generic/default IDE chipset" support compiled in.

firephoto · Posted: Fri Apr 16, 2004 3:32 pm Post subject:

You shouldn't need this option in the kernel.

rebecca · Posted: Fri Apr 16, 2004 3:35 pm Post subject:

dlittlebear · n00b Joined: 16 Apr 2004 Posts: 4

did the suggestion mentioned about killing the other ide option and recompiling the kernel. Nothing, new results.

# hdparm -d1 /dev/hdd

/dev/hdd:
setting using_dma to 1 (on)
HDIO_SET_DMA failed: Operation not permitted
using_dma = 0 (off)

# hdparm -tT /dev/hdd

/dev/hdd:
Timing buffer-cache reads: 476 MB in 2.01 seconds = 236.85 MB/sec
Timing buffered disk reads: 8 MB in 3.97 seconds = 2.02 MB/sec

# cat /usr/src/linux/.config | grep DMA | grep -v "#"
CONFIG_GENERIC_ISA_DMA=y
CONFIG_BLK_DEV_IDEDMA_PCI=y
CONFIG_IDEDMA_PCI_AUTO=y
CONFIG_IDEDMA_ONLYDISK=y
CONFIG_BLK_DEV_ADMA=y
CONFIG_BLK_DEV_IDEDMA=y
CONFIG_IDEDMA_AUTO=y

# cat /usr/src/linux/.config | grep VIA82CXX | grep -v "#"
CONFIG_BLK_DEV_VIA82CXXX=y
# cat /usr/src/linux/.config | grep IDE | grep -v "#"
CONFIG_IDE=y
CONFIG_BLK_DEV_IDE=y
CONFIG_BLK_DEV_IDECD=y
CONFIG_BLK_DEV_IDEPCI=y
CONFIG_IDEPCI_SHARE_IRQ=y
CONFIG_BLK_DEV_IDEDMA_PCI=y
CONFIG_IDEDMA_PCI_AUTO=y
CONFIG_IDEDMA_ONLYDISK=y
CONFIG_BLK_DEV_IDEDMA=y
CONFIG_IDEDMA_AUTO=y

# dmesg | grep ide
BIOS-provided physical RAM map:
CPU: After generic identify, caps: 0183fbff c1c7fbff 00000000 00000000
CPU: After vendor identify, caps: 0183fbff c1c7fbff 00000000 00000000
ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
ide1 at 0x170-0x177,0x376 on irq 15
/dev/ide/host0/bus0/target0/lun0: p1
/dev/ide/host1/bus0/target0/lun0: unknown partition table
/dev/ide/host1/bus0/target1/lun0: p1 p2 p3
VP_IDE: port 0x01f0 already claimed by ide0
VP_IDE: port 0x0170 already claimed by ide1
ide2: BM-DMA at 0xe400-0xe407, BIOS settings: hde:pio, hdf:DMA
ide3: BM-DMA at 0xe408-0xe40f, BIOS settings: hdg:DMA, hdh:DMA

the last 2 lines here are my PCI IDE controller.

# dmesg | grep IDE
Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
VP_IDE: IDE controller at PCI slot 0000:00:07.1
VP_IDE: chipset revision 6
VP_IDE: not 100% native mode: will probe irqs later
VP_IDE: VIA vt82c686b (rev 40) IDE UDMA100 controller on pci0000:00:07.1
VP_IDE: port 0x01f0 already claimed by ide0
VP_IDE: port 0x0170 already claimed by ide1
VP_IDE: neither IDE port enabled (BIOS)
PDC20267: IDE controller at PCI slot 0000:00:11.0

the last line here is my PCI IDE controller.

slightly faster on the hdparm -tT but no where near where I need it.

firephoto · Posted: Sat Apr 17, 2004 2:52 am Post subject:

That has the highpoint raid chipset right? Are you using the drive on that conroller perhaps? If so you need CONFIG_BLK_DEV_HPT366 enabled in your kernel.

Also I noticed the drive is /dev/hdd which should mean it's on the 4th ide channel? Where are you cd drive(s)? Your drive isn't slaved off of a cdrom is it? Your dmesg output for the ide stuff is different than mine too. *edit Ok saw you did a grep IDE and ide.

dlittlebear · n00b Joined: 16 Apr 2004 Posts: 4

not even using the PCI IDE controller yet. Tried, but failed. edited lilo and fstab but have to do more work than what I thought. The HDD /dev/hdd was slaved off the dvdrw drive. The onboard chipset is via, the PCI one is promise controller, I have recompiled the kernel to support that. After looking at this more it looks like I need to disable IDE yet have the via portion take its place. been though the kernel a few hrs now and recompiled more than 5x trying stuff.

npaust · Posted: Wed Apr 28, 2004 10:19 pm Post subject:

I had the same problem with performance down around 1.96 MB/s transfer rates. My dmesg looked just like the poster's with the same errors about the generic IDE driver not letting the via driver work.

Recompiling the kernel and removing the generic IDE driver resulted in my writes going up to the 26 MB/s range.

And for reference, I'm using an 800 MHz VIA mini-itx motherboard.

dyqik · Posted: Fri May 07, 2004 1:06 pm Post subject:

I had a problem getting DMA working on my new Vaio V505DP.

Originaly I had compiled in all the chipsets as modules, as well as the Generic driver above.

Compiling in the Intel ICH driver and leaving out the Generic drive in my kernel (2.6.5-gentoo-r1) as describe above fixed that problem.

69_rs_ss · n00b Joined: 26 Dec 2003 Posts: 65 Location: NY, USA

I'm having the same problem with my drives but it is using the Nforce2 chipset. I have tried all the suggestions above but none have worked. Does anyone know how to get this working since I am even getting a message about it while I am booting up.

seringen · Posted: Wed May 12, 2004 9:19 pm Post subject:

thank you thank you thank you.
getting rid of generic ide support really helped

this took me from a disk read of ~4MB/s to ~18MB/s on my old laptop. Nothing to write home about, but definitely better than it was.

I've known a number of people with this problem, hopefully more people will notice this.

thanks again!

possumjc · Posted: Fri May 14, 2004 12:18 am Post subject:

dlittlebear,
The line from your dmesg output:

69_rs_ss · n00b Joined: 26 Dec 2003 Posts: 65 Location: NY, USA

Is there a way to check the kernel settings off the livecd. Using the livecd I have DMA turned on but after install it keeps erroring. I wanted to see if I was missing something.

possumjc · Posted: Thu May 20, 2004 3:53 pm Post subject:

69_rs_ss,
What settings are you wanting to check? You can run dmesg to see the output of the kernel buring boot. You can also run hdaparm off of the live CD to check drive settings.

Regards,
Joel

69_rs_ss · n00b Joined: 26 Dec 2003 Posts: 65 Location: NY, USA

Ok, I figured it out. I was trying to recompile the kernel but I forgot to mount /boot when copying the new kernel over. Once I mounted it, DMA turned on fine. Thanks for the help.

ufoq · n00b Joined: 25 Mar 2004 Posts: 33