Skip to contentSkip to conversation

rmmod/delete_module hangs forever

DISCUSSION Robin H. Johnson 1 participant 2 messages Mar 15, 2009 – Mar 19, 2009

lore
Share & tools
lore
#0 lore raw

(Please CC me on responses).

I've been meaning to report this for quite some time, but haven't got to it.

I think it may be as a result of some driver that's loaded, rather than the core, as running the git tip (0556e82 and on another machine at home and I can't reproduce it there.

I'm going to try a bisect shortly, and I'll post a followup thereafter.

Strace gives me this fun in the end:
delete_module("button", O_RDONLY|O_EXCL|O_NONBLOCK and it just sits there.

Linux bohr-int 2.6.29-rc7-00188-gebdcc81 #35 SMP Thu Mar 12 03:41:55 PDT 2009 x86_64 Intel(R) Core(TM)2 Quad CPU Q9550 @ 2.83GHz GenuineIntel GNU/Linux

Gnu C                  4.3.3
Gnu make               3.81
binutils               2.19.1
util-linux             2.14.2
mount                  support
module-init-tools      3.6
e2fsprogs              1.41.4
jfsutils               1.1.13
reiserfsprogs          3.6.21
xfsprogs               3.0.0
quota-tools            3.17.
PPP                    2.4.4
Linux C Library        2.9
Dynamic linker (ldd)   2.9
Procps                 3.2.7
Net-tools              1.60
Kbd                    1.15
Sh-utils               7.1
wireless-tools         29
Modules Loaded         loop btusb wacom nfs coretemp uinput af_packet bonding ipv6 cpufreq_ondemand bridge stp llc bnep rfcomm l2cap bluetooth snd_seq_dummy snd_seq_oss snd_seq_midi_event snd_seq snd_seq_device snd_pcm_oss snd_mixer_oss reiserfs xfs dm_mod nvram hangcheck_timer pciehp pci_hotplug nfsd lockd nfs_acl auth_rpcgss sunrpc exportfs cpufreq_stats acpi_cpufreq freq_table w83627ehf hwmon_vid snd_hda_codec_atihdmi snd_hda_codec_analog firewire_ohci firewire_core crc_itu_t snd_hda_intel snd_hda_codec ata_generic snd_hwdep mptspi snd_pcm ohci1394 mptscsih pata_acpi i2c_i801 ahci iTCO_wdt snd_timer thermal pcspkr ieee1394 i2c_core mptbase iTCO_vendor_support libata snd evdev sg processor sky2 rtc_cmos soundcore scsi_transport_spi rtc_core snd_page_alloc tpm_infineon thermal_sys intel_agp tpm rtc_lib tpm_bios button hwmon ext2 scsi_wait_scan ohci_hcd usb_storage sd_mod crc_t10dif
Signature · 4 lines
Robin Hugh Johnson
Gentoo Linux Developer & Infra Guy
E-Mail     : [email protected]
GnuPG FP   : 11AC BA4F 4778 E3F6 E4ED  F38E B27B 944E 3488 4E85
#1 lore raw
Quoted 3 lines · Robin H. Johnson

Strace gives me this fun in the end:
delete_module("button", O_RDONLY|O_EXCL|O_NONBLOCK and it just sits there.

Ok, after much head-scratching and bisect-powered reboots I have disproved for myself that this is a kernel issue.

Instead it's some weird interaction between certain BIOS settings on my motherboard (ASUS P5Q Premium) and my SCSI controller (HP SC11Xe, FusionMPT based) I use for tape specifically when there are no SCSI devices on the bus. Skipping the mptspi module, or connecting a SCSI device causes the problem to go away, ditto changing some settings in the BIOS that cause IRQs to be allocated differently.

Signature · 4 lines
Robin Hugh Johnson
Gentoo Linux Developer & Infra Guy
E-Mail     : [email protected]
GnuPG FP   : 11AC BA4F 4778 E3F6 E4ED  F38E B27B 944E 3488 4E85