Fedora Account System
Red Hat Associate
Red Hat Customer
Description of problem: 2.6.23-0.29.rc0.git6.fc8PAE seems to 'lose' USB events (cursor, keyboard). I can temporarily work around by pulling out and reinserting the device. I sometimes get the following: Jul 16 17:07:08 localhost kernel: usb 5-6.1: USB disconnect, address 5 Jul 16 17:07:08 localhost kernel: *** SLUB sysfs_dir_cache: Padding check failed in slab @0xc1678d3c [Not tainted] Jul 16 17:07:08 localhost kernel: [<c04061a8>] show_trace_log_lvl+0x1a/0x2f Jul 16 17:07:08 localhost kernel: [<c0406c18>] show_trace+0x12/0x14 Jul 16 17:07:08 localhost kernel: [<c0406c71>] dump_stack+0x16/0x18 Jul 16 17:07:08 localhost kernel: [<c0480758>] slab_err+0x51/0x57 Jul 16 17:07:08 localhost kernel: [<c04807c2>] slab_pad_check+0x64/0x90 Jul 16 17:07:08 localhost kernel: [<c0480928>] check_slab+0x13a/0x147 Jul 16 17:07:08 localhost kernel: [<c04813be>] __slab_free+0xbd/0x278 Jul 16 17:07:08 localhost kernel: [<c0481d32>] kmem_cache_free+0x77/0x97 Jul 16 17:07:08 localhost kernel: [<c04c00a1>] release_sysfs_dirent+0x6e/0x83 Jul 16 17:07:08 localhost kernel: [<c04c02c4>] sysfs_addrm_finish+0x1dd/0x1f6 Jul 16 17:07:08 localhost kernel: [<c04befde>] sysfs_hash_and_remove+0x52/0x7f Jul 16 17:07:08 localhost kernel: [<c04c1107>] sysfs_remove_link+0xb/0xd Jul 16 17:07:08 localhost kernel: [<c056c570>] bus_remove_device+0x22/0x82 Jul 16 17:07:08 localhost kernel: [<c056ad21>] device_del+0x212/0x288 Jul 16 17:07:08 localhost kernel: [<c0582b22>] usb_disable_device+0x5f/0xbc Jul 16 17:07:08 localhost kernel: [<c057f0f6>] usb_disconnect+0x9a/0x134 Jul 16 17:07:08 localhost kernel: [<c057f7fe>] hub_thread+0x373/0xa49 Jul 16 17:07:08 localhost kernel: [<c043bb0f>] kthread+0x3b/0x64 Jul 16 17:07:08 localhost kernel: [<c0405c8b>] kernel_thread_helper+0x7/0x10 Jul 16 17:07:08 localhost kernel: ======================= Jul 16 17:07:08 localhost kernel: @@@ SLUB sysfs_dir_cache: Restoring slab padding (0x5a) from 0xdfdcbfc0-0xdfdcbfff Jul 16 17:07:11 localhost kernel: usb 5-6.1: new low speed USB device using ehci_hcd and address 9 Jul 16 17:07:12 localhost kernel: usb 5-6.1: configuration #1 chosen from 1 choice Jul 16 17:07:12 localhost kernel: input: Microsoft Basic Optical Mouse as /class/input/input8 Jul 16 17:07:12 localhost kernel: input: USB HID v1.10 Mouse [Microsoft Basic Optical Mouse] on usb-0000:00:1d.7-6.1 Jul 16 17:09:13 localhost kernel: usb 5-6.1: USB disconnect, address 9 Jul 16 17:09:16 localhost kernel: usb 5-6.1: new low speed USB device using ehci_hcd and address 10 Jul 16 17:09:16 localhost kernel: usb 5-6.1: configuration #1 chosen from 1 choice Jul 16 17:09:16 localhost kernel: input: Microsoft Basic Optical Mouse as /class/input/input9 Jul 16 17:09:16 localhost kernel: input: USB HID v1.10 Mouse [Microsoft Basic Optical Mouse] on usb-0000:00:1d.7-6.1 Jul 16 17:09:50 localhost ntpd[2603]: synchronized to 62.44.101.22, stratum 2 Jul 16 17:09:50 localhost ntpd[2603]: kernel time sync status change 0001 [root@localhost ~]# Version-Release number of selected component (if applicable): 2.6.23-0.29.rc0.git6.fc8PAE How reproducible: Every time Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
Have similar (worse?) issues with 2.6.23-0.30.rc0.git6.fc8PAE: USB keyboard missed the first character typed, and USB mouse is 'lost'. After about 60 seconds of 'use', I get a hard system freeze: have to power cycle to recover. Nothing in /var/log/messages...... I'm running on Thinkpad X60.
*** This bug has been marked as a duplicate of 248783 ***
I'm not convinced it's the same issue that Nicolas has. You should've not merged them. It's trivial to dup bugs, but divorcing them is much harder. If the kernel crashes, it's time to get netconsole or serial console set up, unless you're able to save dmesg and sync in these 60 seconds.
Sorry for prematurely merging. Would you prefer that I open another with a copy of this info? Here is where I am at: Each kernel AFTER kernel-PAE-2.6.23-0.15.rc0.git1.fc8, up to and including kernel-PAE-2.6.23-0.41.rc0.git14.fc8 exhibits the same behavior: at gnome greeter screen, USB mouse is frozen and the first character typed (on USB keyboard) is 'lost' (so typing 'abc' for login yields 'bc'). After the first lost character, I can gnome login, but I have to pull out and reinsert the usb mouse to regain the cursor. ALT-F1 works, and I can enter text on 'console'. Some time later, typically after I've brought up firefox (or some such), system freezes. Can take up to 5-10 minutes. Looks like kernel-2.6.23-0.43.rc0.git16.fc8 just finished on koji. I will download/install and see if anything has changed......
No joy with kernel-PAE-2.6.23-0.43.rc0.git16.fc8: same behavior as above: kernels after 0.15.rc0.git1.fc8 drop USB keyboard characters, USB mouse gets lost, system freezes. This time I tried running 'watch -n 2 "dmesg | tail"' in a terminal window, but got no data on the crash. [I was running rhythmbox with Internet radio at the time.... got hard quarter second audio 'loop'.] Sigh.... I'll look into setting up net-/serial-console. BTW, I the last 'SLUB padding check failed' message I got was with 2.6.23-0.29.rc0.git6.fc8PAE.
I'm having no luck setting up 'netconsole'. I've setup /etc/sysconfig/netconsole and enabled the netconsole service. I've setup a FC6 machine with 'nc -l -u 7654' (7654 is the port number I config'ed in /etc/sysconfig/netconsole), but I get no packets. I can 'nc -u 192.168.1.102 7654' from the Rawhide machine and the packets get there. I completely disabled the firewall on the FC6 machine, but no luck. Are there any issues with netconsole and e1000 driver? Am I doing something dumb?
Created attachment 159803 [details] log file showing problem with USB mouse removal OK, I got a crash with log with .45 (from koji). Has several BUG's, lots of Slab INFOs, etc. I believe I can reproduce this.