[ltp] "80-wire cables" issue when using SATA drives in PATA Thinkpads

[Henrique de Moraes Holschuh]

On Mon, 12 Oct 2020, Stefan Monnier wrote:
> which mentions this as a known problem.
> Is there some known workaround for it?

Many years ago I wrote a kernel patch to disable that logic for all the
internal bays of the IBM T4x thinkpads, and force-detect anything in
them as 80c.

The T43 in the latest firmware got the detection right, except either on
resume or ultrabay hotplug (can't recall which), and the patch fixed
that.

But that patch basically just did what libata.force does: you'd get
UDMA5 all the time, just like you get with libata.force=80c.

> I tried to boot with `libata.force=80c` and it does result in `hdparm`
> showing me that `udma5` is used instead of `udma2`, but that seems to be
> a lie because the actual bandwidth I see is still ~30MB/s.

This is likely the fault of the PATA-SATA bridge.

-- 
  Henrique Holschuh