--==_Exmh_918408438P
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit

In message <7kjsrp$on$1@colwyn.owl.de>you write:
> In article <199906181751.TAA04194@nixalsverdrus.msro.detemobil.de>,
> 	Jan-Hinrich Fessel <fessel@DeTeMobil.de> writes:
> > Is anyone using an AST isa card or an cyclades 32Y pci successfully with 
> > -current?
[...] 
> All ports are in use and work fine.
> 
> > both are on (different) i386, 1.4-current an 1.3I.
> > ast0 at isa0 port 0x1a0-0x1bf irq 5
> > com2 at ast0 slave 0: ns8250 or ns16450, no fifo
> > com3 at ast0 slave 1: st16650, broken fifo
> > com4 at ast0 slave 2: st16650, broken fifo
> > com5 at ast0 slave 3: st16650, broken fifo
> 
> The board is really using IRQ 5 and not set to compatiblity mode? Does
> something else use IRQ 5?

Uh, yes, but only as of yesterday. Iripped out the ast board and found out it 
was jumpered to irq 15.  So I changed that, and yup, the 16450 port works. but 
the 16650 ports still do not work.

From looking at the code I assume this supposedly has to do with the detection 
of the broken fifo, because it seems that the chips are set to 
FIFO_ENABLE|FIFO_TRIGGER_14.
The 16550 seems to be detected as broken when the FIFO_TRIGGER_14 does not 
succeed, but the 16650 happily accepts this setting.
So probably there is some magic needed to reset the 16650 to "don't use FIFO"?

I'll see if I can get hold of a working 16650 and check if that works, if 
noone reports success with 16650's.

Besides, I still cannot get the cyclades stuff to produce any reasonable 
result...

Grüße
        Oskar



--==_Exmh_918408438P
Content-Type: application/pgp-signature

-----BEGIN PGP MESSAGE-----
Version: 2.6.3i

iQCVAwUBN3HxUbA63tqjeAhZAQEZrwP+OyjJ1jNiDJrtMypcc90UbPzGGiuouHDv
sT+MfDc2KDT2rLY9JVBPO4e0VCqF78KVMSiZD+lP8XBM8WYK5/QI1TayiBu0+QJ9
RsPVDQUfmA1I04W27AihR4NMWGL/oTkowV95VPLc10E5Ch16SUec4SCYG4UE79H4
eJPsOjA0p0U=
=UZIr
-----END PGP MESSAGE-----

--==_Exmh_918408438P--