Most recent kernel where this bug did not occur: either 2.6.12 or 2.6.15 - not sure, not tested long enough to tell those won't oops, just the pccard worked better and did never get such an oops. Distribution: Debian / selfcompiled K + sw.suspend2-2.2.5 patches Hardware Environment: notebook Cel.P4 + pccard modem Software Environment: Debian/GNU/Linux Problem Description: kernel oops I had problems with this pccard hw/sw, ttyS* would disappear after eg. 1-2 calls - eg. call, then wait for callback, but by then ttyS* may have gone. Temp. solution was to cardctl eject; cardctl insert; then go on. On some of these sequences got: kernel BUG at drivers/serial/8250.c:1401! invalid opcode: 0000 [#1] PREEMPT Modules linked in: ppp_async ppp_generic slhc nfs lockd nfs_acl sunrpc rfcomm l2 cap bluetooth ipt_ULOG ipv6 nbd irda crc_ccitt autofs4 iptable_mangle iptable_fi lter ip_tables x_tables usbhid ehci_hcd uhci_hcd dm_mod snd_intel8x0 snd_ac97_co dec snd_pcm_oss snd_mixer_oss snd_pcm snd_timer snd soundcore snd_page_alloc snd _ac97_bus evdev as_iosched deadline_iosched i915 drm capability commoncap smbfs usb_storage scsi_mod i2c_dev i2c_core cpufreq_userspace cpufreq_stats cpufreq_po wersave cpufreq_ondemand cpufreq_conservative hangcheck_timer non_fatal p4_clock mod speedstep_lib freq_table tun 8250_pnp 8250_pci 8250_acpi dummy thermal eth13 94 processor fan button battery ac softdog cpuid agpgart serial_cs 8250 serial_c ore pcmcia firmware_class yenta_socket rsrc_nonstatic pcmcia_core ohci1394 ieee1 394 nls_iso8859_15 8139too mii lp parport_pc parport CPU: 0 EIP: 0060:[<cf967952>] Not tainted VLI EFLAGS: 00010246 (2.6.16.13-ss2.2.5-p4.1 #4) EIP is at serial_unlink_irq_chain+0x53/0x5d [8250] eax: 00000000 ebx: cf96ba34 ecx: 00000000 edx: 00000005 esi: cf96be8c edi: 00000202 ebp: cf96be8c esp: c4785e00 ds: 007b es: 007b ss: 0068 Process mgetty (pid: 22875, threadinfo=c4784000 task=c2d45a90) Stack: <0>00000000 00000000 c6650f00 cf96be8c cf95f3a7 cf96be8c 00000000 0000028 6 ce7962bc cdfed800 cf9609ca ce7962bc 00000002 c0598a40 ce7962d4 cdfed800 00000000 00000001 c0598a40 c024c54c cdfed800 c0598a40 00000000 00000000 Call Trace: [<cf95f3a7>] uart_shutdown+0xdc/0xe8 [serial_core] [<cf9609ca>] uart_close+0xe1/0x217 [serial_core] [<c024c54c>] release_dev+0x746/0x74b [<c014eb7d>] free_hot_cold_page+0x105/0x15a [<c015632e>] zap_pte_range+0xa5/0x232 [<c014eb7d>] free_hot_cold_page+0x105/0x15a [<c01920d2>] inotify_dentry_parent_queue_event+0xd5/0xff [<c01e6883>] _atomic_dec_and_lock+0x2b/0x44 [<c024ca11>] tty_release+0x14/0x1f [<c016a0a5>] __fput+0x1bf/0x1d1 [<c0168582>] filp_close+0x4d/0x79 [<c011d8ab>] close_files+0x7e/0x88 [<c011d91d>] put_files_struct+0x2a/0x54 [<c011e3f6>] do_exit+0x13e/0x4a7 [<c011e7ca>] do_group_exit+0x3c/0xbc [<c0102ba7>] sysenter_past_esp+0x54/0x75 Code: 2a 39 00 74 18 89 74 24 04 89 1c 24 e8 33 fe ff ff 8b 5c 24 08 8b 74 24 0c 83 c4 10 c3 89 5c 24 04 89 14 24 e8 8f 18 7e f0 eb da <0f> 0b 79 05 af 90 96 cf eb cc 83 ec 14 89 5c 24 10 c7 44 24 04 <1>Fixing recursive fault but reboot is needed! Steps to reproduce: sequences of carctl eject|insert seems to trigger the bug.
Where are these "sw.suspend2-2.2.5" patches?
www.suspend2.net
Towards the top of drivers/serial/serial_core.c, there is a #undef DEBUG Could you define that, rebuild, reproduce the oops, and send the entire kernel log please? Thanks.
defined,compiled, installed. 2.6.16.20 though. Will do stress-test when I'll have the chance to be phisically near that pc (can't risk lockup from remote).
Any progress?
no, I had a few occasions to stress that hw part but have seen no oops so far. BTW I'm going to upgrade to 2.6.16.27 shortly. I'll try on next weekend to hammer at it again.
Created attachment 8594 [details] oops trying to reboot after playing a bit with the pccard modem and installing the new K hm, might be unrelated, but got an oops trying to reboot after playing a bit with the pccard modem and installing the new K - pls see dmesg output attached
This bug needs a new owner; please arrange for others to be spammed with bugzilla mail. I gave up serial maintainership. Unfortunately the bugzilla disease is that I still get the bug mail.
Paolo, is this still a problem with latest kernels? Thanks.
the pc where I got the problem is out of reach now, and is currently loaded with Debian 4.0 and kernel .18.8 - the most recent version that got me little or no compat-headaches, while the modem (which was the reson for using a serial device) function is done by slmodem+alsa+snd_intel8x0m. Since I've recently compiled a bunch of .21.5, I may be able to try lastest stable on that pc and the pccard modem that oopsed as in original report, within next week (hopely).
Paolo, I suspect the bug has been worked out in recent kernel, closing the bug. Sorry Russell, I am relieving you from unneeded emails, in case the bug gets reopened.
Yep, didn't see it again from .18+. Likely FIXED thanks
sorry, the 'DEFERRED WILL_FIX_LATER' was a due to mis-aiming the pointer. I think it can be CLOSED.