Re: DMA not works in Linux 2.6.12, but in Windows works fine.

Re: DMA not works in Linux 2.6.12, but in Windows works fine.

[Sergey Vlasov]

[-- Attachment #1: Type: text/plain, Size: 1967 bytes --]

On Mon, 23 May 2005 14:25:32 +0400 Ivan G wrote:

> DMA not works in Linux 2.6.12, but in Windows works fine.
> 
> DMA not works with HDD and CD drives connected by 80-conductor
> cable to secondary IDE port (ide1).
> 
> Hardware description:
> 
>    1) Motherboard has chipset Intel, Giga-byte
>    2) HDD Seagate ST3160023AS (Serial ATA)
>    3) HDD Seagate ST3200822A (IDE ATA)
>    4) SONY CD-RW CRX320E, IDE ATAPI CD/DVD-ROM
> 
> 
> Hardware connections:
> 
>    ST3160023AS ---> SATA0 ---> BIOS mapping ---> ide0 Pri master 
>  (hda)
>                     SATA1 ---> BIOS mapping ---> ide0 Pri slave
>    CRX320E     --------------------------------> ide1 Sec master 
>  (hdc)
>    ST3200822A  --------------------------------> ide1 Sec slave 
>   (hdd)
[skip]
> ide0: I/O resource 0x1F0-0x1F7 not free.
> ide0: ports already in use, skipping probe
> Probing IDE interface ide1...
> hdc: SONY CD-RW CRX320E, ATAPI CD/DVD-ROM drive
> hdd: ST3200822A, ATA DISK drive
> Probing IDE interface ide2...
> Probing IDE interface ide3...
> Probing IDE interface ide4...
> Probing IDE interface ide5...
> ide1 at 0x170-0x177,0x376 on irq 15
> hdd: max request size: 1024KiB
> hdd: 390721968 sectors (200049 MB) w/8192KiB Cache, CHS=24321/255/63
> hdd: cache flushes supported
>   hdd: hdd1 hdd2
> hdc: ATAPI 52X DVD-ROM CD-R/RW drive, 2048kB Cache
> Uniform CD-ROM driver Revision: 3.20
> libata version 1.10 loaded.
> ata_piix version 1.03
> ata_piix: combined mode detected
[skip]

This is a known problem - if the Intel ICH5/6 controller is used in
combined mode (SATA mapped to legacy IDE ports), DMA for PATA devices
does not work.  If you reconfigure the controller in BIOS to not use the
combined mode (so that the SATA part becomes a separate PCI device), DMA
for PATA devices will work fine.

To IDE developers: Is something planned to work around this problem?
AFAIK, there are some machines where BIOS does not provide an option to
turn off the combined mode.

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: DMA not works in Linux 2.6.12, but in Windows works fine.

[Jeff Garzik]

Sergey Vlasov wrote:
> To IDE developers: Is something planned to work around this problem?
> AFAIK, there are some machines where BIOS does not provide an option to
> turn off the combined mode.

This limitation exists because two drivers (drivers/ide and libata) want 
to use different parts of the same hardware.

Once libata can do ATAPI 100%, when we can just let libata handle both 
PATA and SATA, which will enable DMA again.

	Jeff

Re: DMA not works in Linux 2.6.12, but in Windows works fine.

[Mark Lord]

Note to others:

If you are brave and pull down Jeff's libata-dev tree,
then it already has ATAPI DMA working (mostly) therein.

That's what I've done on my newest ICH6M machine here,
and both the PATA HD and PATA DVD-RW are working just fine
in "combined mode" using ata_piix.  After applying a bugfix
patch for ATAPI error handling (which Jeff has just posted
here today somewhere for general comments).

Cheers.