QNX6.2 devb-eide versus QNX6.1 on VIA VT8233

devb-eide in QNX6.1 is capable to run DMA mode for VIA chipset VT8233,
devb-eide in QNX6.2 does not boot with DMA.

Hardware:
MB K7T266 MSI, Athlon 1.1G, WD600

Measured transfer speed and average CPU load:
time dd if=/dev/zero of=/test.data bs=256k count=1000
hogs -s5 -p32 -%2 -n

6.1 in DMA mode: write transfer rate: 22MB/sec, 35%CPU load
6.2 in PIO mode: write transfer rate: 4MB/sec, 99%CPU load

Should 6.1 devb-eide work correct in 6.2 ???

Jiri Kristek
RETIA, a.s.

“Jioi Kristek” <jkristek@retia.cz> wrote:

devb-eide in QNX6.1 is capable to run DMA mode for VIA chipset VT8233,
devb-eide in QNX6.2 does not boot with DMA.

The VIA chipset on the ASUS CUV4X and CUV4X-D motherboards VT82C686 has
some issues with DMA that caused it to be unusable in QNX 6.2 (and Linux).
The issue would cause corruption of data.

The devb-eide driver changed significantly between 6.1 and 6.2. The 6.1
driver was not using the full capabilities of DMA, and as a result could
only operate at a maximum mode of DMA 2. This is not so for the 6.2
driver which has significantly more performance capabilites. The 6.2
driver can operate in all DMA modes, including DMA 5. If you operate
this chipset at that DMA level the corruption/issues start to occur.

The only work-around in 6.2 is to disable DMA completely in the devb-eide
driver. This may result in a slugish system.

I solved my issue with the CUV4X and it’s chipset by replacing the mobo
with a different model using a different VIA chipset. I have had no issues
with tha CUV266.

Hardware:
MB K7T266 MSI, Athlon 1.1G, WD600

I would reccomend that you look on the motherboard for the actual part
number of the VIA chips. Once you have that, then hit google and do a
search for that. There are usually postings from the Linux world that
will mention any issues with the chipset (which would potentialy affect
QNX 6.x as well).

Cheers,
Camz

Jiri Kristek <jkristek@retia.cz> wrote in article <3D3EBB21.3070808@retia.cz>…

Thanks for explanation,
Horible for me is, that a time ago with 6.1 we evaluated this boards as
one of the best concernig disk transfer rate and bought a number of
them. With 6.2 they are unusable for our purposes.

I tried to remake qnxbasedma.ifs with 6.1 version of devb-eide in 6.2.
Working, but not in DMA mode.Does anybody know why?

Did you setup manually DMA2 mode for hard drive in BIOS. It’s up for you with old devb-eide. In
some cases, if DMA mode is higher (“auto” in BIOS settings), devb-eide will not use DMA at all.
ed1k.

Thanks for explanation,
Horible for me is, that a time ago with 6.1 we evaluated this boards as
one of the best concernig disk transfer rate and bought a number of
them. With 6.2 they are unusable for our purposes.

I tried to remake qnxbasedma.ifs with 6.1 version of devb-eide in 6.2.
Working, but not in DMA mode.Does anybody know why?







camz@passageway.com wrote:

“Jioi Kristek” <> jkristek@retia.cz> > wrote:

devb-eide in QNX6.1 is capable to run DMA mode for VIA chipset VT8233,
devb-eide in QNX6.2 does not boot with DMA.


The VIA chipset on the ASUS CUV4X and CUV4X-D motherboards VT82C686 has
some issues with DMA that caused it to be unusable in QNX 6.2 (and Linux).
The issue would cause corruption of data.

The devb-eide driver changed significantly between 6.1 and 6.2. The 6.1
driver was not using the full capabilities of DMA, and as a result could
only operate at a maximum mode of DMA 2. This is not so for the 6.2
driver which has significantly more performance capabilites. The 6.2
driver can operate in all DMA modes, including DMA 5. If you operate
this chipset at that DMA level the corruption/issues start to occur.

The only work-around in 6.2 is to disable DMA completely in the devb-eide
driver. This may result in a slugish system.

I solved my issue with the CUV4X and it’s chipset by replacing the mobo
with a different model using a different VIA chipset. I have had no issues
with tha CUV266.


Hardware:
MB K7T266 MSI, Athlon 1.1G, WD600


I would reccomend that you look on the motherboard for the actual part
number of the VIA chips. Once you have that, then hit google and do a
search for that. There are usually postings from the Linux world that
will mention any issues with the chipset (which would potentialy affect
QNX 6.x as well).

Cheers,
Camz

"Ji Kristek <jkristek@retia.cz> wrote:

devb-eide in QNX6.1 is capable to run DMA mode for VIA chipset VT8233,
devb-eide in QNX6.2 does not boot with DMA.

The 6.2 driver doesn’t support DMA on the VT8233 south bridge. We just
received docs for the 8233 on Monday, so support should be available RSN.

Hardware:
MB K7T266 MSI, Athlon 1.1G, WD600

Measured transfer speed and average CPU load:
time dd if=/dev/zero of=/test.data bs=256k count=1000
hogs -s5 -p32 -%2 -n

6.1 in DMA mode: write transfer rate: 22MB/sec, 35%CPU load
6.2 in PIO mode: write transfer rate: 4MB/sec, 99%CPU load

Should 6.1 devb-eide work correct in 6.2 ???

The 6.1 driver should work, but you will have to specify the following
options to diskboot:

-o “devb-eide,eide ioport=0x1f0,irq=14,dma eide ioport=0x170,irq=15,dma”

Kevin Chiles <kchiles@qnx.com> wrote:

"Ji Kristek <> jkristek@retia.cz> > wrote:
devb-eide in QNX6.1 is capable to run DMA mode for VIA chipset VT8233,
devb-eide in QNX6.2 does not boot with DMA.
Should 6.1 devb-eide work correct in 6.2 ???
The 6.1 driver should work, but you will have to specify the following
options to diskboot:
-o “devb-eide,eide ioport=0x1f0,irq=14,dma eide ioport=0x170,irq=15,dma”

Can you not also force UDMA2 (if you are confident it works) on the 6.2
driver with the option “eide master=udma=2” … or is this post-6.2 syntax?

John Garvey <jgarvey@qnx.com> wrote:

Kevin Chiles <> kchiles@qnx.com> > wrote:
"Ji Kristek <> jkristek@retia.cz> > wrote:
devb-eide in QNX6.1 is capable to run DMA mode for VIA chipset VT8233,
devb-eide in QNX6.2 does not boot with DMA.
Should 6.1 devb-eide work correct in 6.2 ???
The 6.1 driver should work, but you will have to specify the following
options to diskboot:
-o “devb-eide,eide ioport=0x1f0,irq=14,dma eide ioport=0x170,irq=15,dma”

Can you not also force UDMA2 (if you are confident it works) on the 6.2
driver with the option “eide master=udma=2” … or is this post-6.2 syntax?

The 6.2 driver doesn’t try to use DMA on unknown chipsets, there were just
too many problems. The next release of the driver has a new option to
enable it.

“Kevin Chiles” <kchiles@qnx.com> pí¹e v diskusním pøíspìvku
news:ahonlc$4kq$1@nntp.qnx.com

John Garvey <> jgarvey@qnx.com> > wrote:
Kevin Chiles <> kchiles@qnx.com> > wrote:
"Ji Kristek <> jkristek@retia.cz> > wrote:
devb-eide in QNX6.1 is capable to run DMA mode for VIA chipset VT8233,
devb-eide in QNX6.2 does not boot with DMA.
Should 6.1 devb-eide work correct in 6.2 ???
The 6.1 driver should work, but you will have to specify the following
options to diskboot:
-o “devb-eide,eide ioport=0x1f0,irq=14,dma eide
ioport=0x170,irq=15,dma”

Can you not also force UDMA2 (if you are confident it works) on the 6.2
driver with the option “eide master=udma=2” … or is this post-6.2
syntax?

The 6.2 driver doesn’t try to use DMA on unknown chipsets, there were just
too many problems. The next release of the driver has a new option to
enable it.

Can we have it in a short time?