On 24/11/16 13:49, Robin Murphy wrote: > On 24/11/16 13:29, Russell King - ARM Linux wrote: >> On Thu, Nov 24, 2016 at 01:18:39PM +0000, Robin Murphy wrote: >>> Hi Liviu, Russell, >>> >>> I'd been meaning to try digging into this if it hadn't gone away since I >>> first noticed it, but I don't really have the time and it still happens >>> with 4.9-rc and today's -next. Representative splat below, but in >>> summary what happens is that if the HDLCD fails to probe, the TDA998x >>> connector seems to get cleaned up twice, resulting in a NULL dereference >>> the second time. I got as far as sketching out the following flow from a >>> debug session (on the same 4.8-rc2 kernel), but I don't know nearly >>> enough to tell which driver is at fault: >>> >>> hdlcd_drm_bind >>> -> drm_fbdev_cma_init (fails) >>> ... >>> -> drm_mode_config_cleanup >>> ... >>> -> drm_connector_cleanup >>> -> component_unbind_all >>> ... >>> -> tda998x_unbind >>> -> drm_connector_cleanup (NULL connector) >>> >>> It's easily reproduced on Juno by booting arm64 defconfig with >>> CONFIG_CMA_SIZE_MBYTES=1 and a sufficiently large monitor connected to >>> warrant a >1MB framebuffer. >> >> It looks to me like a hdlcd bug. >> >> The probe path operates in this order: >> >> - allocates hdlcd - 1 >> - allocates drm device - 2 >> - drm_mode_config_init - 3 >> - hdlcd_load - 4 >> - binds all components - 5 >> - enables runtime PM - 6 >> - drm_vblank_init - 7 >> - drm_mode_config_reset - 8 >> - drm_kms_helper_poll_init - 9 >> - drm_fbdev_cma_init - 10 >> - drm_dev_register - 11 >> >> However, the cleanup operates in this order: >> - drm_fbdev_cma_fini - undoes 10 >> - drm_kms_helper_poll_fini - undoes 9 >> - drm_mode_config_cleanup - undoes 3 >> - drm_vblank_cleanup - undoes 7 >> - pm_runtime_disable - undoes 6 >> - component_unbind_all - undoes 5 >> - drm_irq_uninstall - undoes 4 >> - of_reserved_mem_device_release - undoes other half of 4 >> - drm_dev_unref - undoes 2 >> >> Spot the step which is out of the correct order - drm_mode_config_cleanup() >> is misplaced - it's reversing the actions of drm_mode_config_init(), not >> drm_mode_config_reset(). > > Thanks for the explanation - that saves at least a day's worth of me > trying to understand DRM code :) > >> So, drm_mode_config_cleanup() should be much later, after step 4 has >> been undone, otherwise there are paths that leave various DRM objects >> (created by drm_mode_create_standard_properties()) referenced, and >> will cause problems exactly like you're seeing here. > > Liviu, can I leave this with you then? That said, I just tried the quick and obvious thing over lunch and it does *seem* to be OK: ----->8----- From: Robin Murphy <robin.murphy@xxxxxxx> Subject: [PATCH] drm: hdlcd: Fix cleanup order If hdlcd_drm_bind() fails at drm_fbdev_cma_init(), its cleanup will call drm_mode_config_cleanup() as if to balance drm_mode_config_reset(). The net result is that drm_connector_cleanup() will clean up the active connectors long before component_unbind_all() gets called, so when the connector later tries to clean up itself after being unbound, Bad Things can happen: [ 4.121888] Unable to handle kernel NULL pointer dereference at virtual address 00000000 [ 4.129951] pgd = ffffff80091e0000 [ 4.133345] [00000000] *pgd=00000009ffffe003, *pud=00000009ffffe003, *pmd=0000000000000000 [ 4.141613] Internal error: Oops: 96000005 [#1] PREEMPT SMP [ 4.147144] Modules linked in: [ 4.150188] CPU: 0 PID: 122 Comm: kworker/u12:2 Not tainted 4.8.0-rc2+ #989 [ 4.157097] Hardware name: ARM Juno development board (r1) (DT) [ 4.162981] Workqueue: deferwq deferred_probe_work_func [ 4.168173] task: ffffffc975d93200 task.stack: ffffffc975dac000 [ 4.174055] PC is at drm_connector_cleanup+0x58/0x1c0 [ 4.179074] LR is at tda998x_unbind+0x24/0x40 [ 4.183401] pc : [<ffffff80084c46f0>] lr : [<ffffff800850414c>] pstate: 00000045 [ 4.190750] sp : ffffffc975dafa10 [ 4.194041] x29: ffffffc975dafa10 x28: ffffffc9768152a8 [ 4.199325] x27: ffffffc97ff46450 x26: ffffff8008d99000 [ 4.204608] x25: dead000000000100 x24: dead000000000200 [ 4.209891] x23: ffffffc976bf91e8 x22: 0000000000000000 [ 4.215172] x21: ffffffc976bf9170 x20: ffffffc976bf9170 [ 4.220454] x19: ffffffc976bf9018 x18: 0000000000000000 [ 4.225737] x17: 0000000074ce71ee x16: 000000008ff5d35f [ 4.231019] x15: ffffffc97681e91c x14: ffffffffffffffff [ 4.236301] x13: ffffffc97681e185 x12: 0000000000000038 [ 4.241583] x11: 0101010101010101 x10: 0000000000000000 [ 4.246866] x9 : 0000000040000000 x8 : 0000000000210d00 [ 4.252148] x7 : ffffffc97fea8c00 x6 : 000000000000001b [ 4.257430] x5 : ffffff80084b7b8c x4 : 0000000000000080 [ 4.262712] x3 : ffffff8008504128 x2 : ffffffc975df3800 [ 4.267993] x1 : 0000000000000000 x0 : 0000000000000000 ... [ 4.750937] [<ffffff80084c46f0>] drm_connector_cleanup+0x58/0x1c0 [ 4.756990] [<ffffff800850414c>] tda998x_unbind+0x24/0x40 [ 4.762354] [<ffffff8008507918>] component_unbind.isra.4+0x28/0x50 [ 4.768492] [<ffffff8008507a0c>] component_unbind_all+0xcc/0xd8 [ 4.774373] [<ffffff80084d5adc>] hdlcd_drm_bind+0x234/0x418 [ 4.779909] [<ffffff8008507b58>] try_to_bring_up_master+0x140/0x1a0 [ 4.786133] [<ffffff8008507c50>] component_add+0x98/0x170 [ 4.791496] [<ffffff8008504b90>] tda998x_probe+0x18/0x20 [ 4.796774] [<ffffff80086bf914>] i2c_device_probe+0x164/0x258 [ 4.802481] [<ffffff800850d094>] driver_probe_device+0x204/0x2b0 [ 4.808447] [<ffffff800850d28c>] __device_attach_driver+0x9c/0xf8 [ 4.814498] [<ffffff800850b108>] bus_for_each_drv+0x58/0x98 [ 4.820033] [<ffffff800850cd64>] __device_attach+0xc4/0x138 [ 4.825567] [<ffffff800850d338>] device_initial_probe+0x10/0x18 [ 4.831446] [<ffffff800850c124>] bus_probe_device+0x94/0xa0 [ 4.836981] [<ffffff800850c5b0>] deferred_probe_work_func+0x78/0xb0 [ 4.843207] [<ffffff80080d2998>] process_one_work+0x118/0x378 [ 4.848914] [<ffffff80080d2c40>] worker_thread+0x48/0x498 [ 4.854276] [<ffffff80080d8918>] kthread+0xd0/0xe8 [ 4.859036] [<ffffff8008082e90>] ret_from_fork+0x10/0x40 [ 4.864314] Code: f2fbd5b9 f2fbd5b8 f8478ee0 eb17001f (f9400013) [ 4.870472] ---[ end trace a643cfe4ce1d838b ]--- Fix this by moving the drm_mode_config_cleanup() much later such that it correctly balances drm_mode_config_init(). Suggested-by: Russell King <linux@xxxxxxxxxxxxxxx> Signed-off-by: Robin Murphy <robin.murphy@xxxxxxx> --- drivers/gpu/drm/arm/hdlcd_drv.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/arm/hdlcd_drv.c b/drivers/gpu/drm/arm/hdlcd_drv.c index 59b76054edc9..1a4fff7c0a7c 100644 --- a/drivers/gpu/drm/arm/hdlcd_drv.c +++ b/drivers/gpu/drm/arm/hdlcd_drv.c @@ -420,7 +420,6 @@ static int hdlcd_drm_bind(struct device *dev) err_fbdev: drm_kms_helper_poll_fini(drm); - drm_mode_config_cleanup(drm); drm_vblank_cleanup(drm); err_vblank: pm_runtime_disable(drm->dev); @@ -432,6 +431,7 @@ static int hdlcd_drm_bind(struct device *dev) drm_irq_uninstall(drm); of_reserved_mem_device_release(drm->dev); err_free: + drm_mode_config_cleanup(drm); dev_set_drvdata(dev, NULL); drm_dev_unref(drm); -- 2.10.2.dirty _______________________________________________ dri-devel mailing list dri-devel@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/dri-devel
