Building hpcboot with eVC++4

Folks:
	It looks like eMbedded Visual C++ 3 is no longer
downloadable from 
Microsoft, and it's been long enough since I peeked at
NetBSD/hpc that I 
couldn't find the copy I'd once downloaded (it's probably on
the 
now-deceased laptop, now that I think about it ;)).

The two biggest reasons for needing to generate new project
files for 
eVC++ 4 are (a) the target names changed for many of the
supported CPU 
families, and (b) evC++ 4 generates binaries incompatible
with WinCE 3.0 
by default.

The following set of patches let me build a copy of hpcboot
with evC++ 4 
targetted for ARM (it even runs!); it probably will also
work for 
SH3/SH4.  I'm not sure if hpcmips uses hpcboot anymore
(there's a 
pointer to pbsdboot in the README), but the MIPS
configuration is 
missing since I wasn't sure what CPU level we should be
targetting.

Feedback welcome.. I'd like to eventually check this in so
the next 
lurker into that part of the tree doesn't have to go
searching for tools 
that aren't published anymore.

--rafal

On Sunday 10 February 2008 21:58, Rafal Boni wrote:
> Folks:
> 	It looks like eMbedded Visual C++ 3 is no longer
downloadable from
> Microsoft, and it's been long enough since I peeked at
NetBSD/hpc that I
> couldn't find the copy I'd once downloaded (it's
probably on the
> now-deceased laptop, now that I think about it ;)).
>
> The two biggest reasons for needing to generate new
project files for
> eVC++ 4 are (a) the target names changed for many of
the supported CPU
> families, and (b) evC++ 4 generates binaries
incompatible with WinCE 3.0
> by default.
>
> The following set of patches let me build a copy of
hpcboot with evC++ 4
> targetted for ARM (it even runs!); it probably will
also work for
> SH3/SH4.  I'm not sure if hpcmips uses hpcboot anymore
(there's a
> pointer to pbsdboot in the README), but the MIPS
configuration is
> missing since I wasn't sure what CPU level we should be
targetting.
>
> Feedback welcome.. I'd like to eventually check this in
so the next
> lurker into that part of the tree doesn't have to go
searching for tools
> that aren't published anymore.
>
> --rafal
Hi 
I just had almost the same problem: I needed to recompile
hpcboot for mips and 
I found the Visual C++ 3 here: 
h
ttp://cobweb.ecn.purdue.edu/~vip/install/evt3.html

regards

Risto Sainio

On Sun, Feb 10, 2008 at 15:58:18 -0500, Rafal Boni wrote:

> 	It looks like eMbedded Visual C++ 3 is no longer
downloadable from 
> Microsoft, and it's been long enough since I peeked at
NetBSD/hpc that I 
> couldn't find the copy I'd once downloaded (it's
probably on the 
> now-deceased laptop, now that I think about it ;)).

Link to eMbedded Visual Tools 3.0 download has been broken
repeatedly
on microsoft.com.  Last time I was able to use search to
find it, but
I've just checked now, and it seems that all links their own
search
finds are broken.

I do have both the CD image of eMbedded Visual Tools 3.0
(2001
version, iirc) that also contains SDKs and the minimal (no
SDKs) 2002
edition archive along with seprately downloaded SDKs.  All
were freely
downloadable from ms web site along with the free serial
number to
activate them.

HPC:Factor - which is a good web site about H/PC machines
has a copy
of original eMbedded Visual Tools 3.0 at

http://www.hpcfactor.
com/ftp-root/software/microsoft/embdvtools30/EN_WINCE_EMBDVT
OOLS30.exe


That said, I guess that supporting eVC4 is a smart move
anyway, so
thanks for working on that!


> The following set of patches let me build a copy of
hpcboot with evC++ 4 
> targetted for ARM (it even runs!); it probably will
also work for 
> SH3/SH4.  I'm not sure if hpcmips uses hpcboot anymore
(there's a 
> pointer to pbsdboot in the README), but the MIPS
configuration is 
> missing since I wasn't sure what CPU level we should be
targetting.
> 
> Feedback welcome.. I'd like to eventually check this in
so the next 
> lurker into that part of the tree doesn't have to go
searching for tools 
> that aren't published anymore.

ARMV4 name is confusing (b/c of ARMv4 instruction set). 
Note that all
other toolchain templates use the same names already, so I
think it's
better to use plain "ARM" (etc) for eVC4 too.

SY, Uwe
-- 
uwestderr.spb.ru                       |       Zu Grunde
kommen
http://snark.ptc.spbu.
ru/~uwe/          |       Ist zu Grunde gehen

Valeriy E. Ushakov wrote:

[...re: links to archived copies of eVC++3...]
> That said, I guess that supporting eVC4 is a smart move
anyway, so
> thanks for working on that!

No problem; it was quicker getting a build with eVC++ 4 than
finding the 
links to eVC++ 3 (esp. after you count the time I spent
poking around my 
own machines looking for that archived copy ;)).

[..]
>> Feedback welcome.. I'd like to eventually check
this in so the next 
>> lurker into that part of the tree doesn't have to
go searching for tools 
>> that aren't published anymore.
> 
> ARMV4 name is confusing (b/c of ARMv4 instruction set).
 Note that all
> other toolchain templates use the same names already,
so I think it's
> better to use plain "ARM" (etc) for eVC4
too.

I agree that it's confusing, but unfortunately the names of
the targets 
have changed in eVC++4 and "ARM" is not a target
name it recognizes 
anymore (were it not for this, I probably would have been
able to get 
away with just a change to the existing templates), so to
get a 
buildable project you have to use ARMV4 (or ARMV4I or
ARMV4T).

It looks like the NWO there *is* to align the eVC++ target
name with the 
instruction set being emitted, since besides the ARM split,
the generic 
"MIPS" target has been split into
"MIPS16", "MIPSII" and
"MIPSIV" with 
the latter two also having _FP variants.

--rafal

On Tue, Feb 12, 2008 at 03:36:07AM +0300, Valeriy E. Ushakov
wrote:
> ARMV4 name is confusing (b/c of ARMv4 instruction set).
 Note that all
> other toolchain templates use the same names already,
so I think it's
> better to use plain "ARM" (etc) for eVC4
too.

I had another thought re: this last night... although eVC++
4 needs the
new target names, it might make sense to at least preserve
the names of
the directories the resulting binaries go in.. That I can
easily do if
people feel there's value in that.

Also, from a look at the supported MIPS systems on the
hpcmips page, it
looks like the correct MIPS target for eVC++ 4 should be
"MIPS II", so
I'll add the templates for that as well when I get a few
idle moments.

--rafal

-- 
  Time is an illusion; lunchtime, doubly so.     |//|     
     Rafal Boni
                   -- Ford Prefect               |//|     
rafalpobox.com