Description of problem: BUG: lock held when returning to user space! Version-Release number of selected component (if applicable): kernel-2.6.25-0.121.rc5.git4.fc9.x86_64 kernel-2.6.25-0.141.rc6.git5.fc9.x86_64 How reproducible: load the module Steps to Reproduce: 1. 2. 3. Actual results: excerpt Mar 25 17:57:51 f9 console-kit-daemon[2135]: WARNING: Couldn't read /proc/2875/environ: Error reading file '/proc/2875/environ': No such process Mar 26 04:15:30 f9 kernel: warning: process `system-config-f' used the deprecated sysctl system call with 1.23. Mar 26 04:15:30 f9 kernel: Mar 26 04:15:30 f9 kernel: ================================================ Mar 26 04:15:30 f9 kernel: [ BUG: lock held when returning to user space! ] Mar 26 04:15:30 f9 kernel: ------------------------------------------------ Mar 26 04:15:30 f9 kernel: system-config-f/4088 is leaving the kernel with locks still held! Mar 26 04:15:30 f9 kernel: 1 lock held by system-config-f/4088: Mar 26 04:15:30 f9 kernel: #0: (&priv->stats_lock){--..}, at: [_end+109199257/2109760552] prism54_get_wireless_stats+0x21/0x9c [prism54] Mar 26 04:15:54 f9 kerneloops: Submitted 1 kernel oopses to www.kerneloops.org Expected results: Additional info: this happens with and without the firmware[1]. in the later case the box behaves crazy. on "input field" eg. console or firefox adresse line all letters gets repeated indefinitely. "a" becomes "aaaaaaaaaaaaaaaaaaaaaaaaa....." same seen on F8 kernel since 2.6.24.3-34.fc8 [1] I got two of this prism54 cards. up to F9 beta on this box, where this bz is about, always isl3886 firmware was loaded eg. p54pci module was active. on the other box was always isl3890 firmware loaded eg. prism54. both cards were operating.
Created attachment 299083 [details] /var/log/messages exerpt Here is a similar problem, maybe. Unfortunately, I could not repeat the problem, and I was fooling around with NetworkManager at the time - disabling and enabling wireless and changing between various APs. The good news is that iwl3945 and NetworkManager managed to work in other conditions, and didn't fail at all when I rebooted to the current version of F9beta.
Peter, I believe the issue you are seeing is actually the same as in bug 438584 -- please use that to track the issue.
Created attachment 299218 [details] 0001-prism54-correct-thinko-in-prism54-Convert-stats_s.patch