This is a follow-up on my previous set of patches for the Qemu PALcode,
which were merged except for the CTB patch.  The patch has incorporated
review feedback, but there is still some research going on about how
real DEC SRM initializes a particular field in various circumstances;
this is my best guess based on available documentation and observed
behavior of real machines, and is sufficient for the BSD operating systems.

Jason Thorpe (1):
  Provide a minimal Console Terminal Block in the HWRPB.

 hwrpb.h  | 54 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 init.c   | 36 +++++++++++++++++++++++++++++++++---
 protos.h |  2 ++
 vgaio.c  |  2 ++
 4 files changed, 91 insertions(+), 3 deletions(-)

-- 
2.30.2

> On Jun 13, 2021, at 2:09 PM, Jason Thorpe <thorpej@me.com> wrote:
> 
> but there is still some research going on about how
> real DEC SRM initializes a particular field in various circumstances;
> this is my best guess based on available documentation and observed
> behavior of real machines, and is sufficient for the BSD operating systems.

FWIW, I have since confirmed my hypothesis about how genuine DEC SRM fills in the "type" and "term_type" fields on 2 different generations of PCI-based Alpha systems (i.e. it matches the patch submitted here).

-- thorpej

On 6/14/21 5:29 PM, Jason Thorpe wrote:
> 
> 
>> On Jun 13, 2021, at 2:09 PM, Jason Thorpe <thorpej@me.com> wrote:
>>
>> but there is still some research going on about how
>> real DEC SRM initializes a particular field in various circumstances;
>> this is my best guess based on available documentation and observed
>> behavior of real machines, and is sufficient for the BSD operating systems.
> 
> FWIW, I have since confirmed my hypothesis about how genuine DEC SRM fills in the "type" and "term_type" fields on 2 different generations of PCI-based Alpha systems (i.e. it matches the patch submitted here).

Excellent, thanks.  Queued.


r~