what is hdparm doing?

what is hdparm doing?

[Sebastian Kuzminsky]

Hi folks, hdparm's got me confused.

I've got an old PC with an ALi Corporation MS4803 (rev 01) IDE controller.
According to the manufacturer, this controller does NOT support DMA,
only PIO 0-4.  I'm booting off various Sandisk Compact Flash cards,
which I believe all support both PIO and DMA.

I'm using hdparm 6.1, and I get practically identical results with 2.4.32
and 2.6.15.4.


"hdparm /dev/hda" says:

    /dev/hda:
     multcount    =  0 (off)
     IO_support   =  0 (default 16-bit)
     unmaskirq    =  0 (off)
     using_dma    =  0 (off)
     keepsettings =  0 (off)
     readonly     =  0 (off)
     readahead    =  8 (on)
     geometry     = 1986/16/63, sectors = 2001888, start = 0


"hdparm -I /dev/hda" says:

    CompactFlash ATA device, with removable media
            Model Number:       SanDisk SDCFH-1024
            Serial Number:      003305F1104P3318
            Firmware Revision:  HDX 2.18
    Standards:
            Supported: 10
            Likely used: 10
    Configuration:
            Logical         max     current
            cylinders       1986    1986
            heads           16      16
            sectors/track   63      63
            --
            CHS current addressable sectors:    2001888
            LBA    user addressable sectors:    2001888
            device size with M = 1024*1024:         977 MBytes
            device size with M = 1000*1000:        1024 MBytes (1 GB)
    Capabilities:
            LBA, IORDY(may be)(cannot be disabled)
            Queue depth: 1
            Standby timer values: spec'd by Vendor
            R/W multiple sector transfer: Max = 1   Current = 1
            DMA: mdma0 mdma1 *mdma2
                 Cycle time: min=120ns recommended=120ns
            PIO: pio0 pio1 pio2 pio3 pio4
                 Cycle time: no flow control=120ns  IORDY flow control=120ns


The "hdparm" output shows using_dma is off.  That's refering the the
HBA, right?

As I understand the "hdparm -I" output, however, it claims the drive
(not the HBA) is using mdma2.  Is that right?  How can the drive and
the HBA use different modes?

I thought the HBA and the drive(s) would automatically use the fastest
mode that both devices support, which in this case should be PIO4.

I tried to manually switch the HBA to pio4 with "hdparm -p4 /dev/hda"
but got this error:

    /dev/hda:
     attempting to set PIO mode to 4
     HDIO_SET_PIO_MODE failed: Function not implemented

Manually switching the drive to pio4 with "hdparm -X pio4 /dev/hda"
seems to work:

    /dev/hda:
     setting xfermode to 12 (PIO flow control mode4)

But even after running this, "hdparm -I /dev/hda" shows it using mdma2
(with the * by it).


BTW, here's the relevant part of dmesg, from 2.4.32:

    Uniform Multi-Platform E-IDE driver Revision: 7.00beta4-2.4
    ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
    hda: SanDisk SDCFH-1024, CFA DISK drive
    ide2: ports already in use, skipping probe
    ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
    hda: attached ide-disk driver.
    hda: 2001888 sectors (1025 MB) w/1KiB Cache, CHS=1986/16/63
    Partition check:
     hda: hda1 hda2 hda3


Please clue in a poor monkey - what is going on here?


-- 
Sebastian Kuzminsky

Re: what is hdparm doing?

[Mark Lord]

Sebastian Kuzminsky wrote:
> Hi folks, hdparm's got me confused.
..
> "hdparm /dev/hda" says:
.
>             DMA: mdma0 mdma1 *mdma2
>                  Cycle time: min=120ns recommended=120ns
>             PIO: pio0 pio1 pio2 pio3 pio4
>                  Cycle time: no flow control=120ns  IORDY flow control=120ns
..

The *mdma2 means, that if you were to issue a READ_DMA (or WRITE_DMA)
command to the drive, the drive is expecting the host to use
the multiword-dma-2 protocol and timings for it.

Meanwhile, the host may, at its option, simply use PIO commands (READ, WRITE)
instead of DMA, any time it chooses.

Drives have both a DMA mode, and a PIO mode, each of which is selected
and used independently of the other.

Cheers

Re: what is hdparm doing?

[Sebastian Kuzminsky]

Mark Lord <liml@rtr.ca> wrote:
> Sebastian Kuzminsky wrote:
> > Hi folks, hdparm's got me confused.
> ..
> > "hdparm /dev/hda" says:
> .
> >             DMA: mdma0 mdma1 *mdma2
> >                  Cycle time: min=120ns recommended=120ns
> >             PIO: pio0 pio1 pio2 pio3 pio4
> >                  Cycle time: no flow control=120ns  IORDY flow control=120ns
> ..
> 
> The *mdma2 means, that if you were to issue a READ_DMA (or WRITE_DMA)
> command to the drive, the drive is expecting the host to use
> the multiword-dma-2 protocol and timings for it.
> 
> Meanwhile, the host may, at its option, simply use PIO commands (READ, WRITE)
> instead of DMA, any time it chooses.
> 
> Drives have both a DMA mode, and a PIO mode, each of which is selected
> and used independently of the other.

Ok thanks!

So how do I know what PIO mode the drive wants to use?

And how do I tell the HBA what PIO mode to use?

-- 
Sebastian Kuzminsky

Re: what is hdparm doing?

[Mark Lord]

Sebastian Kuzminsky wrote:
 >
> So how do I know what PIO mode the drive wants to use?
> 
> And how do I tell the HBA what PIO mode to use?
> 

You don't do either of those.
The device driver takes care of it, if needed.

In this case, they were both likely set up for PIO4.

Cheers