0001401: install option in modprobe.conf doesn't work

Notes
(0002498) vda 06-22-07 04:48	> For example, if the module "fred" worked better with the module "barney" already installed... ...you have to mail lkml and help kernel people to improve "fred" to not need to jump thru the hoops like this. It's not a bug. It's a "missing feature which exists in standard modprobe". I am not convinced yet we want to turn busybox modprobe into generic application launcher by having this peculiar option. Please explain why do you need that.

(0002499) quitte 06-22-07 12:59	Unfortunately the linux kernel mainatiners can do nothing about it as the problems I ran into are with proprietary modules. And getting the vendor to fix things seems to be completely impossible. The device is a dsl, wlan and phone router based on the ti ar7. I'll try to explain what I wanted to use this feature for by examples: The usb, wlan and dsl drivers all need a proprietary module loaded before they can be loaded. This module can not be distributed because of its license and the usb hci driver gives no clue why it doesn't work if that module isn't loaded in the first place. There are multiple variations of this device that all work with the same modules but need different firmware files and module options. As an alternative to letting modprobe run a script that figures out which firmware to load and which modules to use it should be doable by changing modprobe.conf from an init script. But doing it at the time that loading the module is requested seems to make a lot more sense to me. The dsl driver provides an atm interface that has to be made into an ethernet interface before being usable with pppd. While it might be possible to run the necessary scripts for setting that up through hotplug events, doing so from modprobe.conf makes more sense to me - as the device is never hotplugged.

(0002500) bernhardf 06-22-07 14:19	If you desparately need such advanced features, then i suggest you use the big modutils. Apart from that, can't you simply have an /etc/modules which content gets autoloaded: (sample from my debian host). See how all your (module related) problems are solved :) # Loop over every line in /etc/modules. echo -n 'Loading modules: ' (cat /etc/modules; echo) \| # make sure there is a LF at the end while read module args do case "$module" in \#*\|"") continue ;; esac echo -n "$module " modprobe $module $args done echo

(0002503) integrator 06-25-07 10:26	First, introducing "install" (and "remove" at the same time) would bloat modprobe uselessly. That would make modprobe depend on a shell, which has been purposedly avoided and removed in the past. Such special loading could well be handled by startup scripts. Second, you're wrong: the true way _is_ going with udev events to handle hardware (or pseudo hw) comings and goings. Have your atm driver generates appropriate events, handle them with udev by running br2684ctl (or whatever) and then spawning a ppp, and so on... Third, regarding firmwares: there is firmware loading facilities in the kernel. Use that. Have your driver request firmwares the same way others do. Besides, you're talking about proprietary modules. Using this statements clearly states that something is broken on _your_ side. We can't, and won't, fix your modules. If you definetly want this, go for the big module-init-tools. As far as I'm concerned, this is a non-issue. Denis and Bernhard, comments? Regards, Yann E. MORIN.

(0002505) bernhardf 06-25-07 10:46	I consider it closed since "06-22-07 14:19" ;)

(0002525) integrator 06-27-07 01:17	Non-issue for BusyBox.

Issue History
Date Modified	Username	Field	Change
06-21-07 18:54	quitte	New Issue
06-21-07 18:54	quitte	Status	new => assigned
06-21-07 18:54	quitte	Assigned To	=> BusyBox
06-21-07 19:02	quitte	Issue Monitored: quitte
06-22-07 04:48	vda	Note Added: 0002498
06-22-07 12:59	quitte	Note Added: 0002499
06-22-07 14:19	bernhardf	Note Added: 0002500
06-25-07 10:15	integrator	Assigned To	BusyBox => integrator
06-25-07 10:26	integrator	Note Added: 0002503
06-25-07 10:46	bernhardf	Note Added: 0002505
06-27-07 01:17	integrator	Status	assigned => closed
06-27-07 01:17	integrator	Note Added: 0002525
06-27-07 01:17	integrator	Resolution	open => won't fix

Relationships