Coda File System

Re: Build concerns

From: Greg Troxel <gdt_at_ir.bbn.com>
Date: 28 Jan 2005 18:59:05 -0500
"M. Anthony Aiello" <aiello_at_cs.virginia.edu> writes:

> Building the most current userland coda tarballs on the mac, I had to
> make the following changes:
> 
> 1) in coda-src/vol/vldb.h:
> 	Add 	"#define uint32_t unsigned long" to the top of the file.

This ought to be in <stdint.h>, which ought to be included.  But
perhaps not...

> 2) in coda-src/venus/worker.cc:
> 	Add 	"#define __NetBSD__"
> 		"#include "pioctl.h"
> 	to the top of the file.
> 
> The second change, however, was a guess: I have no idea if the NetBSD
> values for those constants are meaningful/reasonable for the Mac; I
> actually have no idea what those constants are controlling (and didn't
> bother to look too closely).

Quite likely, since the mac kernel support is a port of the BSD
support.

IMHO, pioctl.h ought to be part of the OS, and provided along with the
kernel coda module, but that view is not widely shared.

Probably a __DARWIN__ ifdef branch needs to be made in various places.
In general assuming darwin is like NetBSD is not correct, although
for many specifics it is probably ok.

> Following the instructions on the OpenDarwin coda port page, I was
> able to connect to and get directory listings from the cmu public test
> server, so my change in 2) appears to have worked....
> 
> Tony
> 

-- 
        Greg Troxel <gdt_at_ir.bbn.com>
Received on 2005-01-28 19:00:20