On Monday 30 July 2001 07:41 pm, Ron Johnson wrote:
> Do I remember properly that because of the bug in this chip,
> that the LM80 2.4 kernel SRPMs have DMA disabled on IDE CDRs and
> CDRWs? And that that means that creating CDR* disks from IDE
> CDR* devices is impossible on motherboards that have the 686b?
I believe this business of the VIA-IDE bug is greatly overblown. It
only rarely supposedly affects large transfers across IDE channels
(>100mb). I have a VIA chipset board (Soyo k7vta pro) and have never
experienced this bug. I just burned backup copies of Mandrake Freq2
last night. I have 2 IBM HDDs on ide0, and a CDrom and CD-RW on ide1.
Both LM 8.0 (various 2.4.x kernels) and Freq2 (2.4.5-5) have set up all
the drives using 32bit and DMA during install. No problems, either
burning CD's or while testing moving a 700mb file across ide.
In researching this bug before I bought the VIA board I came across
three elements that are said to provoke the bug. Older VIA chipsets
(there's been bascially 3 steppings since the chip was introduced). SB
sound cards, particularly SB Live! PCI, and then mostly with Abit
motherboards. Newer Western Digital HDD's. I have none of the above,
and since my research was done on the Net, take it with a grain of salt.
I have had some correspondence with a friend who has all the above, 2
new WD HDDs, a pci SB Live!, on a Abit KT7a, yet he's never experienced
the bug or had problems burning CD's either.
--
Tom Brinkman Galveston Bay
