On Tue, May 13, 2008 at 3:12 AM, Thomas Bächler <thomas@archlinux.org> wrote:
Dan McGee schrieb:
Spoke too early. Does this mean nvidia could benefit from a recompile? I don't know if some symbols changed since the initial compile for .25. Damn xscreensaver killing my machine.
BUG: unable to handle kernel NULL pointer dereference at 00000148 IP: [<f930957c>] :nvidia:os_set_mlock_capability+0xc/0x30 *pde = 00000000 Oops: 0002 [#1] PREEMPT SMP Modules linked in: ipv6 nfs nfsd lockd nfs_acl auth_rpcgss sunrpc exportfs ipt_REJECT xt_tcpudp nf_conntrack_ipv4 xt_state nf_conntrack iptable_filter ip_tables x_tables sg sd_mod ext2 usb_storage scsi_mod parport_pc snd_seq_oss snd_seq_midi_event snd_seq snd_seq_device firewire_ohci firewire_core crc_itu_t snd_intel8x0 snd_ac97_codec serio_raw ppdev snd_pcm_oss snd_mixer_oss ac97_bus snd_pcm nvidia(P) snd_timer snd lp parport ohci1394 psmouse soundcore snd_page_alloc ppp_generic slhc pcspkr ieee1394 ehci_hcd ohci_hcd shpchp pci_hotplug sis_agp agpgart thermal evdev processor fan button battery ac it87 hwmon_vid eeprom i2c_sis96x i2c_core floppy 8139too mii rtc_cmos rtc_core rtc_lib usbcore ext3 jbd mbcache ide_cd_mod cdrom ide_disk ide_pci_generic sis5513 ide_core
Pid: 17241, comm: hypnowheel Tainted: P (2.6.25-ARCH #1) EIP: 0060:[<f930957c>] EFLAGS: 00010286 CPU: 0 EIP is at os_set_mlock_capability+0xc/0x30 [nvidia] EAX: 00000000 EBX: 00000000 ECX: 00000000 EDX: 00000286 ESI: f7afc000 EDI: f7140c00 EBP: f2c63ea8 ESP: f2c63e9c DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068 Process hypnowheel (pid: 17241, ti=f2c62000 task=f6f7f200 task.ti=f2c62000) Stack: f908aadf f7021280 00000006 f2c63ec8 f908c44c 0000005b f94d0da0 f2c13c80 f908d97b f7afc000 00000001 f2c63ef8 f908de85 f94d0da0 f2c13c80 f94d0da0 1cd47067 c03fd1e0 b7fdffff b7fdf000 0000005b f2de4400 00000000 f2c63f28 Call Trace: [<f908aadf>] _nv001924rm+0xb/0x24 [nvidia] [<f908c44c>] rm_write_watch_init+0x54/0x78 [nvidia] [<f908d97b>] _nv001997rm+0x2b/0x770 [nvidia] [<f908de85>] _nv001997rm+0x535/0x770 [nvidia] [<f908b524>] rm_ioctl+0x1c/0x24 [nvidia] [<f9306a7e>] nv_kern_ioctl+0xae/0x330 [nvidia] [<f9306d38>] nv_kern_unlocked_ioctl+0x18/0x20 [nvidia] [<f9306d20>] nv_kern_unlocked_ioctl+0x0/0x20 [nvidia] [<c019178b>] vfs_ioctl+0x2b/0x90 [<c0191857>] do_vfs_ioctl+0x67/0x2e0 [<c0191b36>] sys_ioctl+0x66/0x70 [<c01050d8>] sysenter_past_esp+0x6d/0xa5 ======================= Code: 80 0d 4d f9 c3 8d 76 00 8d bc 27 00 00 00 00 f3 c3 8d b4 26 00 00 00 00 8d bc 27 00 00 00 00 64 a1 00 b0 3f c0 8b 80 bc 04 00 00 <c7> 80 48 01 00 00 ff ff ff ff 64 a1 00 b0 3f c0 81 88 bc 01 00 EIP: [<f930957c>] os_set_mlock_capability+0xc/0x30 [nvidia] SS:ESP 0068:f2c63e9c ---[ end trace 71ca9a3959efaf65 ]---
Apparently, these changes http://repos.archlinux.org/viewvc.cgi/kernel26/repos/testing-i686/config?r1=376&r2=1507 require rebuild of all modules. Could you test if rebuilding external modules helps?
I rebuilt last night and did not come back to a locked machine this morning, so I think we are OK if we rebuilt modules. To Pierre- I use nvidia-96xx and not the main version, so that is definitely not the problem. And everything worked fine already for the original 2.6.25. -Dan