Bug 454407
| Summary: | running udevtrigger in X causes strange X behavior | ||
|---|---|---|---|
| Product: | [Fedora] Fedora | Reporter: | Marina Fadeeva <astarta> |
| Component: | dkms | Assignee: | Mario Limonciello <mario_limonciello> |
| Status: | CLOSED NEXTRELEASE | QA Contact: | Fedora Extras Quality Assurance <extras-qa> |
| Severity: | low | Docs Contact: | |
| Priority: | low | ||
| Version: | 9 | CC: | adam, ahz001, harald, jcm, kay.sievers, linux-bugs, matt_domsch, tcallawa |
| Target Milestone: | --- | ||
| Target Release: | --- | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Fixed In Version: | 2.0.21.1-2.fc11 | Doc Type: | Bug Fix |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | 2009-05-20 23:56:48 UTC | Type: | --- |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
|
Description
Marina Fadeeva
2008-07-08 10:59:37 UTC
dkms MUST NOT call udevtrigger.. udev resets all permissions, if you do that. https://bugzilla.redhat.com/show_bug.cgi?id=238383 shows an ack by Harald and Kay to add this behavior. Has something changed in udev since then? Adding Kay and Jon; this has the potential to affect Jon's KMODs too. ping udev guys. If tools shouldn't call udevtrigger, then it should not be an easily-accessible app to run. Perhaps a udevtrigger --nochangeperms or somesuch option would be better? Recent udev does not change permissions, it restores the permissions specified in the rules if they do not match the current perms of the node, they are not touched otherwise. In theory, all apps should handle "duplicated" events, which udevtrigger causes, but in real world some apps get confused by this. So, yes, it should be avoided to run udevtrigger. Why does dkms want to run udevtrigger? The loading of the module should generate all needed events and do all what is needed, right? How is that different from a device hotplug case? DKMS does not modprobe/insmod the module. It only copies it into /lib/modules/$kernelversion and runs depmod. It started calling udevtrigger a year ago https://bugzilla.redhat.com/show_bug.cgi?id=238383 so that udev causes the module to get loaded if needed/applicable. At that time, it was suggested that dkms call udevtrigger. If that's no longer the case, and there's a "better" way to handle such, please advise. But having DKMS call modprobe itself isn't really the right answer - it can't do all the "do I need this module" type of checking that udev does. This should do the trick: # find /sys -name modalias -print0 | xargs -0 cat | xargs /sbin/modprobe -a -v or without "-v" of course modaliases appear in /sys before modules that implement them are loaded? yes, generic <bus>:<vendor_id><device_id> and modules have these aliases udev does not much more for module loading Mario, see comments 7 and 8, and apply accordingly then. This bug seems to manifest itself as CTRL+F2 acting like CTRL+ALT+F2 just after installing VirtualBox until rebooting Fedora 10. http://www.virtualbox.org/ticket/3117 dkms 2.0.21.1 has the call to udevadm trigger removed. I suppose I need to get a new version built into fedora then... dkms-2.0.21.1-2.fc10 has been submitted as an update for Fedora 10. http://admin.fedoraproject.org/updates/dkms-2.0.21.1-2.fc10 ================================================================================
dkms-2.0.21.1-2.fc10
================================================================================
Release: Fedora 10
Status: pending
Type: bugfix
Karma: 0
Request: testing
Bugs: 454407 - running udevtrigger in X causes strange X behavior
Notes: Removes call to udevadm trigger, which fixes udev resetting
: permissions on device nodes. Uses
: /etc/modprobe.d/dkms.conf per module-init-tools
: expectations for naming of conf files there. Add
: Requires: lsb to avoid potential problems with RPM
: recursion.
Submitter: mdomsch
Submitted: 2009-04-15 12:47:41
Comments: mdomsch - 2009-04-15 12:47:49 (karma 0)
This update has been submitted for testing
https://admin.fedoraproject.org/updates/dkms-2.0.21.1-2.fc10
dkms-2.0.21.1-2.fc9 has been submitted as an update for Fedora 9. http://admin.fedoraproject.org/updates/dkms-2.0.21.1-2.fc9 dkms-2.0.21.1-2.fc9 has been pushed to the Fedora 9 testing repository. If problems still persist, please make note of it in this bug report. If you want to test the update, you can install it with su -c 'yum --enablerepo=updates-testing-newkey update dkms'. You can provide feedback for this update here: http://admin.fedoraproject.org/updates/F9/FEDORA-2009-3701 dkms-2.0.21.1-2.fc10 has been pushed to the Fedora 10 testing repository. If problems still persist, please make note of it in this bug report. If you want to test the update, you can install it with su -c 'yum --enablerepo=updates-testing update dkms'. You can provide feedback for this update here: http://admin.fedoraproject.org/updates/F10/FEDORA-2009-3703 Did this miss getting into Fedora 11? Yes, it missed Fedora 11. It's sitting in updates-testing for Fedora 9 and 10. I'll build for updates-testing for Fedora 11 now. dkms-2.0.21.1-2.fc11 has been submitted as an update for Fedora 11. http://admin.fedoraproject.org/updates/dkms-2.0.21.1-2.fc11 dkms-2.0.21.1-2.fc11 has been pushed to the Fedora 11 testing repository. If problems still persist, please make note of it in this bug report. If you want to test the update, you can install it with su -c 'yum --enablerepo=updates-testing update dkms'. You can provide feedback for this update here: http://admin.fedoraproject.org/updates/F11/FEDORA-2009-4960 dkms-2.0.21.1-2.fc10 has been pushed to the Fedora 10 stable repository. If problems still persist, please make note of it in this bug report. dkms-2.0.21.1-2.fc9 has been pushed to the Fedora 9 stable repository. If problems still persist, please make note of it in this bug report. dkms-2.0.21.1-2.fc11 has been pushed to the Fedora 11 stable repository. If problems still persist, please make note of it in this bug report. |