[LV] The kernel build process & tools any changes .

[Maciej W. Rozycki]

On Mon, 4 Jun 2007, Mr. James W. Laferriere wrote:

>  	I see we are still under the weather or the gun to get things 
> done by those we made promises to & those we care about .

 Well, I have a schedule which I keep more or less, but it does not 
specify any dates, only dependencies.  Right now I am in the kernel mode 
for the other architecture, which is MIPS.

 I have just finished porting zs.c to the serial subsystem.  This is for 
the DECstation.  There may still be issues and I still have to look at how 
to make it work in a clean way with an LK201 keyboard and a VSXXX mouse 
automatically for the virtual console, but priority has dropped 
significantly on this one.  I shall use the experience gained here to make 
some updates to dz.c and make it a platform device -- it should work for 
our port automatically then (perhaps with some small changes to the 
interrupt handlers).

 I have started a driver for the TURBOchannel Dual SCSI cards (DECstation 
again), which may be eventually of use for our port as well, but I have 
stubled across some problems in the recently modernised esp_scsi core 
(which we will probably want to use eventually as well).  Thus this is 
stalled until I get documentation for the chips involved and get through 
the relevant part of the SCSI standard, so that I am able to work on the 
core as well.  The card itself should work for our port eventually too, 
except there are some mechanical constraints which result from the way 
TURBOchannel options are mounted in VAXstations.  They may make attaching 
SCSI cables an interesting experience.

 The next on the plate is porting sb1250_duart.c to the serial subsystem.  
This is for the Broadcom SOC used on a couple of their boards.  It should 
not take more than a day or two; certainly not after the bloody zs.c 
experience (FYI, DEC have wired some of the modem lines incorrectly, 
making the handling of them rather painful).  I should be doing it 
essentially right now.  No use for our port whatsoever though. ;-)

 Then I plan to switch into the user mode.  And there, first on the plate 
is GCC 4.1 and glibc 2.4.  This is where I left off a while ago.  These 
versions, while a little bit outdated by now, are not that unreasonable as 
they still sort of support running with linuxthreads rather than NPTL 
only.  This has the advantage of making a port of glibc to VAX/Linux 
possible without the need of defining a TLS ABI right from the beginning.  
Which is what I would rather avoid.  Any interest in a port of glibc?  
Testers?

>  	Tho sure would be nice to see a bit here in a month other than from 
> myself .

 Was it nice to see, then? ;-)

  Maciej
_______________________________________________
Linux-Vax mailing list
Linux-Vax at pergamentum.com
http://www.pergamentum.com/mailman/listinfo/linux-vax