Re: Strange silsata* issue (was: Changes for mpt*)

["Jonathan A. Kollasch" <jakllsch%kollasch.net@localhost>]

On Wed, Jul 01, 2026 at 03:31:43PM +0000, John Klos wrote:
> [  32.4070337] siisata0: SOFT_RESET failed on port 3 (error 26 PSS 0x1 PIS 0x20000), resetting

The error 26 here is described by the SiI3124A Data Sheet as:

26 PLDCMDERRORMASTERABORT: "A PCI Master Abort occurred while the SiI3124
was fetching a Port Request Block (PRB) from host memory."

A PCI Master Abort here means that nothing on the PCI bus provided the
data the SiI3124 requested.  This could also mean that the IOMMU did not
allow/map the access to host memory.  That said, I thought I had tested
siisata(4) on a AlphaStation 200 successfully when I was developing it.
But that was a SiI3124 on a 32-bit-only card on a 33MHz/32-bit only bus.
So the possibility of a driver issue seems unlikely.  That leaves the
possibility of a PCI(-X) device, bus, or bridge compatibility issue, or
of course a bad electrical connection somewhere.

If there's a compatibility or electrical connection issue, that might
explain spurious (mis)behavior of other devices or interrupts.

Is your 3124 card a PCI-X card or a PCI card?  Sometimes there can be
issues using a PCI-X chip connected to a PCI or PCI-X host via a
non-PCI-X card.  See re(4)'s BUGS section about the XN-152 card design
for example.

Alas, all my Alphas (including a CS20) are no longer operational at the
moment, so I'm unable to do testing myself.

	Jonathan