On Fri, Jun 17, 2016 at 08:47:04AM -0000, Patchwork wrote: > == Series Details == > > Series: series starting with [1/7] drm: Automatically register/unregister all connectors > URL : https://patchwork.freedesktop.org/series/8814/ > State : warning > > == Summary == > > Series 8814v1 Series without cover letter > http://patchwork.freedesktop.org/api/1.0/series/8814/revisions/1/mbox > > Test drv_module_reload_basic: > pass -> DMESG-WARN (ro-bdw-i7-5557U) > pass -> DMESG-WARN (ro-ilk1-i5-650) > pass -> DMESG-WARN (ro-hsw-i3-4010u) > pass -> DMESG-WARN (ro-bdw-i7-5600u) > pass -> DMESG-WARN (ro-snb-i7-2620M) > pass -> DMESG-WARN (ro-bdw-i5-5250u) > pass -> DMESG-WARN (ro-hsw-i7-4770r) > pass -> DMESG-WARN (ro-ivb-i7-3770) > pass -> DMESG-WARN (ro-ivb2-i7-3770) > pass -> DMESG-WARN (ro-byt-n2820) > pass -> DMESG-WARN (fi-skl-i7-6700k) So this is because i915 is one of the bad drivers still using driver->unload and not updating its init ordering: [ 160.838919] ------------[ cut here ]------------ [ 160.838933] WARNING: CPU: 2 PID: 5784 at fs/sysfs/group.c:237 sysfs_remove_group+0x86/0x90 [ 160.838938] sysfs group ffffffff81cc8380 not found for kobject 'intel_backlight' [ 160.838941] Modules linked in: i915(-) i2c_algo_bit drm_kms_helper snd_hda_codec_hdmi snd_hda_codec_generic syscopyarea sysfillrect sysimgblt fb_sys_fops x86_pkg_temp_thermal intel_powerclamp drm snd_hda_codec snd_hwdep snd_hda_core coretemp snd_pcm crct10dif_pclmul crc32_pclmul mei_me ghash_clmulni_intel lpc_ich mei sdhci_pci sdhci mmc_core e1000e ptp pps_core [last unloaded: snd_hda_intel] [ 160.838998] CPU: 2 PID: 5784 Comm: rmmod Tainted: G U 4.7.0-rc2-gfxbench-RO_Patchwork_1203+ #1 [ 160.839003] Hardware name: LENOVO 4287CTO/4287CTO, BIOS 8DET47WW (1.17 ) 05/31/2011 [ 160.839006] 0000000000000000 ffff880212bd3bc0 ffffffff8140e085 ffff880212bd3c10 [ 160.839016] 0000000000000000 ffff880212bd3c00 ffffffff8107a996 000000ed81c797e0 [ 160.839024] 0000000000000000 ffffffff81cc8380 ffff88020bc6c290 ffff88020bed0000 [ 160.839032] Call Trace: [ 160.839040] [<ffffffff8140e085>] dump_stack+0x67/0x92 [ 160.839047] [<ffffffff8107a996>] __warn+0xc6/0xe0 [ 160.839052] [<ffffffff8107a9fa>] warn_slowpath_fmt+0x4a/0x50 [ 160.839060] [<ffffffff817a92e9>] ? mutex_unlock+0x9/0x10 [ 160.839065] [<ffffffff8125b375>] ? kernfs_find_and_get_ns+0x45/0x50 [ 160.839070] [<ffffffff8125ec66>] sysfs_remove_group+0x86/0x90 [ 160.839076] [<ffffffff81523e22>] dpm_sysfs_remove+0x52/0x60 [ 160.839083] [<ffffffff81517834>] device_del+0x44/0x240 [ 160.839088] [<ffffffff810cbd7a>] ? up_write+0x1a/0x40 [ 160.839093] [<ffffffff81517a49>] device_unregister+0x19/0x60 [ 160.839099] [<ffffffff81475fb9>] backlight_device_unregister+0x89/0x90 [ 160.839155] [<ffffffffa039a25f>] intel_backlight_unregister+0x2f/0x50 [i915] [ 160.839192] [<ffffffffa0367ad9>] intel_modeset_cleanup+0x29/0xb0 [i915] [ 160.839225] [<ffffffffa03a257c>] i915_driver_unload+0x6c/0x1b0 [i915] [ 160.839241] [<ffffffffa018799c>] drm_dev_unregister+0x2c/0xb0 [drm] [ 160.839253] [<ffffffffa0187f2e>] drm_put_dev+0x1e/0x50 [drm] [ 160.839274] [<ffffffffa02d9300>] i915_pci_remove+0x10/0x20 [i915] [ 160.839278] [<ffffffff81457cd4>] pci_device_remove+0x34/0xb0 [ 160.839281] [<ffffffff8151ba9c>] __device_release_driver+0x9c/0x150 [ 160.839285] [<ffffffff8151c666>] driver_detach+0xb6/0xc0 [ 160.839288] [<ffffffff8151b4b3>] bus_remove_driver+0x53/0xd0 [ 160.839291] [<ffffffff8151d127>] driver_unregister+0x27/0x50 [ 160.839294] [<ffffffff81456d55>] pci_unregister_driver+0x25/0x70 [ 160.839307] [<ffffffffa0189292>] drm_pci_exit+0x72/0x90 [drm] [ 160.839339] [<ffffffffa03a2d06>] i915_exit+0x20/0x209 [i915] [ 160.839343] [<ffffffff8110ba39>] SyS_delete_module+0x199/0x1f0 [ 160.839347] [<ffffffff817acae9>] entry_SYSCALL_64_fastpath+0x1c/0xac [ 160.839351] ---[ end trace d3039df93d0212ed ]--- So perhaps we have to wait until more drivers are de-midlayered? Or do we apply and fix up only reports (not many are going to be quite as complex as i915, as evidenced by the lack of connector unregister callbacks elsewhere)? At any rate, we will have an ordering conflict between drm-misc and next-queue (i.e. bisecting through the drm-misc branch will generate warns for i915). -Chris -- Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
