ATIIXP support (SATA, ATA133)

ATIIXP support (SATA, ATA133)

[Enrico Scholz]

Hello,

are there plans to support newer ATI IXP chipsets? I mean:

* PCI 1002:4376 which is an ATA133 controller; this should probably work
  with the already existing drivers/ide/pci/atiixp.c driver

* PCI 1002:4379 which is a SATA controller; I verified that it works
  with the drivers/ide/pci/atiixp.c driver of 2.6.9 also.

* PCI 1002:436E and 1002:437A which are both SATA controller. I saw them
  in the windoze driver file only and can not tell whether they work
  with atiixp.c


Although at least one SATA chipset works with the IDE driver, an own
driver using libata would be the better solution. Does there exist
already something which can be tested?



Enrico

Re: ATIIXP support (SATA, ATA133)

[Bartlomiej Zolnierkiewicz]

On Fri, 17 Dec 2004 11:01:05 +0100, Enrico Scholz
<enrico.scholz@informatik.tu-chemnitz.de> wrote:
> Hello,

Hi,

> are there plans to support newer ATI IXP chipsets? I mean:
> 
> * PCI 1002:4376 which is an ATA133 controller; this should probably work
>   with the already existing drivers/ide/pci/atiixp.c driver

Have you tested it?

> * PCI 1002:4379 which is a SATA controller; I verified that it works
>   with the drivers/ide/pci/atiixp.c driver of 2.6.9 also.

Care to make a patches for 1002:4376 and 1002:4379?

> * PCI 1002:436E and 1002:437A which are both SATA controller. I saw them
>   in the windoze driver file only and can not tell whether they work
>   with atiixp.c
> 
> Although at least one SATA chipset works with the IDE driver, an own
> driver using libata would be the better solution. Does there exist
> already something which can be tested?

At least some of ATi SATA chipsets are based on Silicon Image SATA
so it should be possible to use sata_sil.c for them.

Bartlomiej

Re: ATIIXP support (SATA, ATA133)

[Enrico Scholz]

[-- Attachment #1: Type: text/plain, Size: 2238 bytes --]

Bartlomiej Zolnierkiewicz <bzolnier@gmail.com> writes:

>> are there plans to support newer ATI IXP chipsets? I mean:
>> 
>> * PCI 1002:4376 which is an ATA133 controller; this should probably work
>>   with the already existing drivers/ide/pci/atiixp.c driver
>
> Have you tested it?

Yes, but with mixed results:

* a harddisk works fine and without problems

* reading from a DVD/CD-ROM drives with DMA mode gives read-errors
  (position not reproducible, no single-bit errors). I will have to
  make further tests to see if it is caused by the drive, the system
  (see below) or the driver. There are no kernel-messages regarding
  read-errors. In PIO mode, data are read correctly.


The system is an AMD64 system where:

* the i386 mode does not work (lots of lost interrupts)

* in x86_64 mode, the timer is running twice as fast as it should. Dunno
  if this affects the CD-ROM operation

Happens both with the Fedora Core 3 kernel (kernel-2.6.9-1.681_FC3) and
2.6.10-rc3-bk13.


>> * PCI 1002:4379 which is a SATA controller; I verified that it works
>>   with the drivers/ide/pci/atiixp.c driver of 2.6.9 also.
>
> Care to make a patches for 1002:4376 and 1002:4379?

ok, attached. The :4379 uses the sata_sil driver. 'hdparm -t' gives
30MB/s but because of the broken timer this should be around 60MB/s in
real.

After this 'hdparm' command I see a

| HDIO_DRIVE_CMD(null) (wait for flush complete) failed: Inappropriate ioctl for device

caused by

| ioctl(3, BLKFLSBUF, 0)                  = 0
| ioctl(3, 0x31f, 0)                      = -1 ENOTTY (Inappropriate ioctl for device)


I never saw hdparm on other SATA drives so I can not tell if this is
expected.


>> * PCI 1002:436E and 1002:437A which are both SATA controller. I saw them
>>   in the windoze driver file only and can not tell whether they work
>>   with atiixp.c
>> 
>> Although at least one SATA chipset works with the IDE driver, an own
>> driver using libata would be the better solution. Does there exist
>> already something which can be tested?
>
> At least some of ATi SATA chipsets are based on Silicon Image SATA
> so it should be possible to use sata_sil.c for them.

Thanks. As you can see in the patch, I use it for the :4379 device.



Enrico


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: support for 1002:437[69] ATI IXP devices --]
[-- Type: text/x-patch, Size: 1285 bytes --]

2004-12-20  Enrico Scholz  <enrico.scholz@informatik.tu-chemnitz.de>
	* drivers/ide/pci/atiixp.c: handle 1002:4376 IDE controller also
	* drivers/scsi/sata_sil.c: handle 1002:4379 SATA controller also

--- Linuxa-2.6.10-rc3/drivers/scsi/sata_sil.c.atiixp	2004-12-03 22:52:13.000000000 +0100
+++ linux-2.6.10-rc3/drivers/scsi/sata_sil.c	2004-12-20 20:23:14.000000000 +0100
@@ -71,6 +71,7 @@ static struct pci_device_id sil_pci_tbl[
 	{ 0x1095, 0x0240, PCI_ANY_ID, PCI_ANY_ID, 0, 0, sil_3112 },
 	{ 0x1095, 0x3512, PCI_ANY_ID, PCI_ANY_ID, 0, 0, sil_3112 },
 	{ 0x1095, 0x3114, PCI_ANY_ID, PCI_ANY_ID, 0, 0, sil_3114 },
+	{ 0x1002, 0x4379, PCI_ANY_ID, PCI_ANY_ID, 0, 0, sil_3112 },
 	{ }	/* terminate list */
 };
 
--- linux-2.6.10-rc3/drivers/ide/pci/atiixp.c.atiixp	2004-12-20 19:34:12.000000000 +0100
+++ linux-2.6.10-rc3/drivers/ide/pci/atiixp.c	2004-12-20 20:21:34.000000000 +0100
@@ -348,6 +348,7 @@ static int __devinit atiixp_init_one(str
 static struct pci_device_id atiixp_pci_tbl[] = {
 	{ PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_IXP_IDE, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0},
 	{ PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_IXP2_IDE, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0},
+	{ PCI_VENDOR_ID_ATI, 0x4376, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0},
 	{ 0, },
 };
 MODULE_DEVICE_TABLE(pci, atiixp_pci_tbl);

Re: ATIIXP support (SATA, ATA133)

[Enrico Scholz]

Enrico Scholz <enrico.scholz@informatik.tu-chemnitz.de> writes:

>>> * PCI 1002:4376 which is an ATA133 controller; this should probably
>>>   work with the already existing drivers/ide/pci/atiixp.c driver
>>
>> Have you tested it?
>
> Yes, but with mixed results:
>
> * a harddisk works fine and without problems
>
> * reading from a DVD/CD-ROM drives with DMA mode gives read-errors
>   (position not reproducible, no single-bit errors). I will have to
>   make further tests to see if it is caused by the drive, the system
>   (see below) or the driver. There are no kernel-messages regarding
>   read-errors. In PIO mode, data are read correctly.

Ok, I did some testings; here the results:

* things are fine with another CD drive (PLEXTOR CD-R PX-W4824A)

* the previously mentioned one (LG HL-DT-STDVDRRW GWA-4083B) works after
  executing

  | hdparm -a 64 /dev/hda

  Any number above 64 makes it fail.




Enrico

Re: ATIIXP support (SATA, ATA133)

[Bartlomiej Zolnierkiewicz]

On Mon, 03 Jan 2005 18:53:20 +0100, Enrico Scholz
<enrico.scholz@informatik.tu-chemnitz.de> wrote:
> Enrico Scholz <enrico.scholz@informatik.tu-chemnitz.de> writes:
> 
> >>> * PCI 1002:4376 which is an ATA133 controller; this should probably
> >>>   work with the already existing drivers/ide/pci/atiixp.c driver
> >>
> >> Have you tested it?
> >
> > Yes, but with mixed results:
> >
> > * a harddisk works fine and without problems
> >
> > * reading from a DVD/CD-ROM drives with DMA mode gives read-errors
> >   (position not reproducible, no single-bit errors). I will have to
> >   make further tests to see if it is caused by the drive, the system
> >   (see below) or the driver. There are no kernel-messages regarding
> >   read-errors. In PIO mode, data are read correctly.

there is confirmation from ATI now, patch applied

> Ok, I did some testings; here the results:
> 
> * things are fine with another CD drive (PLEXTOR CD-R PX-W4824A)
> 
> * the previously mentioned one (LG HL-DT-STDVDRRW GWA-4083B) works after
>   executing
> 
>   | hdparm -a 64 /dev/hda
> 
>   Any number above 64 makes it fail.
 
weird