A couple of small, paid userland hacker projects.

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view

A couple of small, paid userland hacker projects.

Juli Mallett
Hey there,

I'm looking for someone to implement each of the following:

(1) Make pkg_add(1) not use system(3) to execute external programs to
do things that it can implement internally (i.e. calling out to tar(1)
is fine, calling out to mkdir(1) is not.)  Alternately, rewrite
pkg_add as a sh(1) shell script, with perhaps a minimal utility
written in C (pkg_admin?) to muck with /var/db/pkg.

(2) Add an option to specify an alternate root for pkg_add(1) to
install into.  This means that if pkg_add(1) were passed an alternate
root of '/foo' instead of updating /var/db/pkg, it would update
/foo/var/db/pkg; instead of untarring to /usr/local/bin it would untar
to /foo/usr/local/bin.  This is instead of chroot(2) functionality; I
need to be able to install packages onto a filesystem which contains
binaries for another (incompatible) architecture, so the chroot(2)
approach doesn't work.

(3) Make kldload(8) warn if something that looks like a filename
without a path (i.e. there is a . in it or it ends in .ko or whatever,
but there is no slash in it — kldload(2) uses a slash to determine
whether something is a path) is specified on the command line and:
(a) that file exists in the current directory
(b) the current directory is not in the module search path *OR* is not
the first directory in the module search path in which a file with
that name exists.  Consider a module search path of
/boot/modules;/boot/kernel — if I am in /boot/kernel and I type
'kldload if_foo.ko' and there is a file named 'if_foo.ko' in
/boot/modules, the module that will be loaded is not the one that I
would expect to be loaded unless I instead spell it 'kldload
./if_foo.ko'.  Likewise if I am in /boot/monkeys and I type the same
command, there is no chance that the if_foo.ko in /boot/monkeys will
be loaded.
NB: I have kernel patches to make kldload(2) smarter, but I would like
to get kldload(8) to warn about it in 8.x and then fix kldload(2) in

Each of these is fairly small, hopefully on the order of a few hours.
Please contact me as soon as possible with a flat rate quote and
qualifications (unless you're a FreeBSD committer, obviously) if
you're interested, I'd like to see these finished within the next two

[hidden email] mailing list
To unsubscribe, send any mail to "[hidden email]"