udev and pci devices
Hi all,
we stumbled across an issue of udev and busybox/modprobe. udev detects a PCI device and according to udevadm test it is selecting:
udev_rules_apply_to_event: RUN '/sbin/modprobe -bv $env{MODALIAS}' /lib/udev/rules.d/80-drivers.rules:5udev_rules_apply_to_event: RUN '/sbin/modprobe $env{MODALIAS}' /etc/udev/rules.d/local.rules:31
but as the busybox modprobe does not support the 'b' option it just fails before executing the other rules. There are some possible fixes to it.
1.)
- remove the local.rules rule for loading the module
- build busybox with CONFIG_FEATURE_MODPROBE_BLACKLIST for -b support
2.)
- remove the 80-drivers.rules and let the modprobe just be executed by the other udev rules.
comments?
holger
Quoted 11 lines · Holger Hans Peter Freyther
Hi all,
we stumbled across an issue of udev and busybox/modprobe. udev detects a PCI device and according to udevadm test it is selecting:
udev_rules_apply_to_event: RUN '/sbin/modprobe-bv $env{MODALIAS}'/lib/udev/rules.d/80-drivers.rules:5udev_rules_apply_to_event: RUN '/sbin/modprobe$env{MODALIAS}'/etc/udev/rules.d/local.rules:31
Quoted 6 lines
but as the busybox modprobe does not support the 'b' option it just fails before executing the other rules. There are some possible fixes to it.
1.)
- remove the local.rules rule for loading the module
- build busybox withCONFIG_FEATURE_MODPROBE_BLACKLISTfor -b support
This sounds like the correct thing to do I think since the 80-driver.rules file is coming from upstream, the local.rules piece is our extras piece.
Cheers,
Richard
Quoted 4 lines · Holger Hans Peter Freyther
Hi all,
we stumbled across an issue of udev and busybox/modprobe. udev detects a PCI device and according to udevadm test it is selecting:
ping? any interest in having module loading for PCI devices working?
Quoted 1 line · Holger Hans Peter Freyther
ping? any interest in having module loading for PCI devices working?
no interest?
Quoted 5 lines · Holger Hans Peter Freyther
Quoted 1 line · Holger Hans Peter Freyther
ping? any interest in having module loading for PCI devices working?
no interest?
I think up to this point most of the udev support has focused on typical hotplug items, specifically USB.
In my experience things like PCI devices are better setup statically based on the system being created. (Note, I'm not sure if we have an easy way to do that at this point or not..)
--Mark
Quoted 15 lines · Richard Purdie
Quoted 14 lines · Holger Hans Peter Freyther
Hi all,
we stumbled across an issue of udev and busybox/modprobe. udev detects a PCI device and according to udevadm test it is selecting:
udev_rules_apply_to_event: RUN '/sbin/modprobe-bv $env{MODALIAS}'/lib/udev/rules.d/80-drivers.rules:5udev_rules_apply_to_event: RUN '/sbin/modprobe$env{MODALIAS}'/etc/udev/rules.d/local.rules:31but as the busybox modprobe does not support the 'b' option it just fails before executing the other rules. There are some possible fixes to it.
Quoted 7 lines
Quoted 3 lines
1.)
- remove the local.rules rule for loading the module
- build busybox withCONFIG_FEATURE_MODPROBE_BLACKLISTfor -b supportThis sounds like the correct thing to do I think since the 80-driver.rules file is coming from upstream, the local.rules piece is our extras piece.
Seems fine to me however
How much size does CONFIG_FEATURE_MODPROBE_BLACKLIST add to busybox Is there some existing knob that could be used to enable/disable this feature
Quoted 10 lines
Cheers,
Richard
_______________________________________________
poky mailing list
[email protected]
https://lists.yoctoproject.org/listinfo/poky
Quoted 3 lines · Mark Hatle
In my experience things like PCI devices are better setup statically based on the system being created. (Note, I'm not sure if we have an easy way to do that at this point or not..)
With statically you mean putting module_autoload_NAME = "name" in the out of tree kernel module recipe? This will not 'scale' for us (e.g. one system image but n-different ethernet NICs we can put in our hardware)
cheers
holger