Re: bad dir ino 4 at offset 80: mangled entry

[nathanw%MIT.EDU@localhost (Nathan J. Williams)]

Ken Hornstein <kenh%cmf.nrl.navy.mil@localhost> writes:

>>I am using Cygwin to build the kernel + ramdisk image to be loaded onto a
>>ARM based target machine. 
>
> You're not the first person to report this.  It is clearly some issue with
> the NetBSD tools that are cross-compiled under Cygwin (my guess is
> that the problem lies in nbmakefs).  From what I have seen, the kernel
> is fine, it's just the ramdisk filesystem image is corrupt.
>
> However, since no one has stepped up and debugged the problem, it has not
> been fixed.  I don't think many of the NetBSD hard-core developers
> cross-compile under Cygwin.  If you want to solve the problem, the best
> way is to roll up your sleeves and track it down.

This is fine advice. However, one of the reasons that developers have
avoided Cygwin is that our build process runs very slowly there (high
fork overhead in Cygwin/Windows, I believe). When I've been developing
on a Windows host, I've found it faster to install NetBSD/i386 in a
virtual machine (even qemu) and do a build hosted there.

        - Nathan