[LV] Is this thing on?

[Maciej W. Rozycki]

On Wed, 9 Nov 2011, Kaj-Michael Lang wrote:

> > I have uploaded some newer RPM packages, including a full version of the
> > i386 cross-compiler (FP support in the Java port may not work), more
> 
> Nice, great work. What about the kernel, where is it hosted and what 
> works? I did run 2.4 on my VaxStation 3100 many years ago. Does todays 
> kernel still run on it ?

 Sorry about the sporadic responses.

 I run a 2.6.18 derivative, which is certainly better than 2.4, but 
obviously hardly new anymore.  I won't go into the details as to why I 
haven't upgraded yet beyond that particular version, but I certainly plan 
to do that one day.

 Anyway, the derivative has been made by taking a LMO kernel (i.e. one 
from linux-mips.org) and applying VAX-specific bits from Jan-Benedict's 
repository that used to live at:

git://git.kernel.org/pub/scm/linux/kernel/git/jbglaw/vax-linux.git

and was last updated up to 2.6.23-rc4 according to my most recent daily 
snapshot.  I have no idea if the repository has been restored after the 
kernel.org outage earlier this year.  And I don't really know what the 
status of the kernel port is as Jan-Benedict has been busy and distracted 
with some more important matters recently.  I have cc-ed him in case 
(hope?) he has anything to add.

 Jan-Benedict's kernel should support a number of systems out of the box, 
including the VAXstation 4000 Model 60.  I'm not sure about the others 
like your VAXstation 3100.  My box, which is a VAXstation 4000 Model 90, 
is not fully supported, but I can run it with a local hack.  I plan to 
clean the hack up and submit it gradually (there are several functionally 
independent pieces) as I push through.  With the userland more or less 
working now I'm more confident these Model 90 changes actually make sense.

 I believe I have sorted out the biggest obstacle, that is I have figured 
out how interrupts are handled on the Model 90 (by the famous CEAC or 
CDAL-to-EDAL Adapter Chip; a specialised circuit implemented in this 
computer -- a bit unusually for a production system, especially for that 
time -- in an FPGA chip) and how EDAL devices are decoded in the system's 
address space (most importantly the DC7085/DZ11 UART ASIC needed for the 
console device and some less important bits like the RTC).

 The biggest pain about the Model 90 remaining is the SGEC Ethernet driver 
that doesn't work correctly or something (I haven't really investigated 
the cause), making initial NFS-rooted setup a bit, ahem... problematic.  
For the time being I have used my experience with MIPS DECstation systems 
and hardware I have available and worked it around by using a PMAD-AA 
TURBOchannel Ethernet board with a TURBOchannel adapter that Model 60 and 
90/90A/96 systems support (DMA isn't supported yet by the VAX port, which 
excludes some TURBOchannel FDDI network boards that could potentially work 
too, but the PMAD-AA fortunately is PIO only).

 Meanwhile I have been working on sorting out the ptrace(2) interface that 
has been suffering from incompleteness and some design, hmm... let's just 
say quirks.  I believe I have now fixed the first half, that is the said 
quirks that apply to exception handlers.  I need to deal with the syscall 
itself now, which I will hopefully be able to do early next year.  Once 
this has been been completed, I'll post patches here.  This will enable me 
adding VAX/Linux native and remote (via gdbserver) support to GDB needed 
to debug the remaining known userland problems.

 All of this should really be summarised on a web page somewhere.  One we 
currently have on sf.net is a good candidate, at least for a start; I'll 
see what I can do about it as I should have write access to that page.  
Ultimately I think we should have a wiki somewhere instead.

 What we really need are additional hands (and heads! :) ) to help with 
the project.  Any kinds of skills are good -- these don't necessarily have 
to be kernel/libc/toolchain software development skills, as we need people 
to do testing and write documentation (e.g. maintain status on web/wiki 
pages) too.  Otherwise the port will just progress as it has been so far, 
i.e. at the snail's pace.

 Meanwhile -- happy Christmas, everyone!

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