Hi Am 15.12.24 um 19:18 schrieb Dr. David Alan Gilbert:
Hey Gerd, Thomas, I've got the following oops that looks bochs-drm related on the current HEAD ( 4800575d8c0b2f354ab05ab1c4749e45e213bf73 ) and it's been there for at least a few days; this is
[...]
The oops has : [ 78.463760][ T1] bochs_pci_driver_init+0x8a/0xc0 in it, hence why I'm blaming that. (Other odd observation, the Tuxen flicker heavily during booting!) [ 72.756014][ T1] bochs-drm 0000:00:02.0: vgaarb: deactivate vga console [ 72.758258][ T1] [drm] Found bochs VGA, ID 0xb0c5. [ 72.758793][ T1] [drm] Framebuffer size 16384 kB @ 0xfd000000, mmio @ 0xfebf0000. [ 72.767777][ T1] [drm] Initialized bochs-drm 1.0.0 for 0000:00:02.0 on minor 2 [ 72.839222][ T1] fbcon: bochs-drmdrmfb (fb1) is primary device [ 72.839311][ T1] fbcon: Remapping primary device, fb1, to tty 1-63 [ 78.402163][ T1] bochs-drm 0000:00:02.0: [drm] fb1: bochs-drmdrmfb frame buffer device [ 78.459984][ T1] BUG: unable to handle page fault for address: ffff8dd345604004 [ 78.463246][ T1] #PF: supervisor write access in kernel mode [ 78.463760][ T1] #PF: error_code(0x0002) - not-present page [ 78.463760][ T1] PGD 72001067 P4D 72001067 PUD 72002067 PMD 7fbe1067 PTE 800ffffffa9fb060 [ 78.463760][ T1] Oops: Oops: 0002 [#1] PREEMPT SMP DEBUG_PAGEALLOC NOPTI [ 78.463760][ T1] CPU: 2 UID: 0 PID: 1 Comm: swapper/0 Tainted: G W N 6.13.0-rc2+ #363 6c653a430ed30aae3dac648429c492a2726da3d7 [ 78.463760][ T1] Tainted: [W]=WARN, [N]=TEST [ 78.463760][ T1] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.3-3.fc41 04/01/2014 [ 78.463760][ T1] RIP: 0010:devm_drm_dev_init_release+0x4e/0x140
[...]
[ 78.463760][ T1] ---[ end Kernel panic - not syncing: Fatal exception ]--- The config is a fairly full yes-config ish; see attached.
Thanks for reporting. I've been able to reproduce the problem by setting CONFIG_DEBUG_TEST_DRIVER_REMOVE <https://elixir.bootlin.com/linux/v6.13-rc2/K/ident/CONFIG_DEBUG_TEST_DRIVER_REMOVE>=y. The attached patch fixes the problem for me. Could you please test and report back the results.
Best regards Thomas
Dave
-- -- Thomas Zimmermann Graphics Driver Developer SUSE Software Solutions Germany GmbH Frankenstrasse 146, 90461 Nuernberg, Germany GF: Ivo Totev, Andrew Myers, Andrew McDonald, Boudien Moerman HRB 36809 (AG Nuernberg)
From bebba3b34d5df5aa7c882f080633b43ddb87f4ad Mon Sep 17 00:00:00 2001 From: Thomas Zimmermann <tzimmermann@xxxxxxx> Date: Mon, 16 Dec 2024 09:21:46 +0100 Subject: [PATCH] drm/bochs: Do not put DRM device in PCI remove callback Removing the bochs PCI device should mark the DRM device as unplugged without removing it. Hence clear the respective call to drm_dev_put() from bochs_pci_remove(). Fixes a double unref in devm_drm_dev_init_release(). An example error message is shown below: [ 32.958338] BUG: KASAN: use-after-free in drm_dev_put.part.0+0x1b/0x90 [ 32.958850] Write of size 4 at addr ffff888152134004 by task (udev-worker)/591 [ 32.959574] CPU: 3 UID: 0 PID: 591 Comm: (udev-worker) Tainted: G E 6.13.0-rc2-1-default+ #3417 [ 32.960316] Tainted: [E]=UNSIGNED_MODULE [ 32.960637] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.16.3-2-gc13ff2cd-prebuilt.qemu.org 04/01/2014 [ 32.961429] Call Trace: [ 32.961433] <TASK> [ 32.961439] dump_stack_lvl+0x68/0x90 [ 32.961452] print_address_description.constprop.0+0x88/0x330 [ 32.961461] ? preempt_count_sub+0x14/0xc0 [ 32.961473] print_report+0xe2/0x1d0 [ 32.961479] ? srso_alias_return_thunk+0x5/0xfbef5 [ 32.963725] ? __virt_addr_valid+0x143/0x320 [ 32.964077] ? srso_alias_return_thunk+0x5/0xfbef5 [ 32.964463] ? drm_dev_put.part.0+0x1b/0x90 [ 32.964817] kasan_report+0xce/0x1a0 [ 32.965123] ? drm_dev_put.part.0+0x1b/0x90 [ 32.965474] kasan_check_range+0xff/0x1c0 [ 32.965806] drm_dev_put.part.0+0x1b/0x90 [ 32.966138] release_nodes+0x84/0xc0 [ 32.966447] devres_release_all+0xd2/0x110 [ 32.966788] ? __pfx_devres_release_all+0x10/0x10 [ 32.967177] ? preempt_count_sub+0x14/0xc0 [ 32.967523] device_unbind_cleanup+0x16/0xc0 [ 32.967886] really_probe+0x1b7/0x570 [ 32.968207] __driver_probe_device+0xca/0x1b0 [ 32.968568] driver_probe_device+0x4a/0xf0 [ 32.968907] __driver_attach+0x10b/0x290 [ 32.969239] ? __pfx___driver_attach+0x10/0x10 [ 32.969598] bus_for_each_dev+0xc0/0x110 [ 32.969923] ? __pfx_bus_for_each_dev+0x10/0x10 [ 32.970291] ? bus_add_driver+0x17a/0x2b0 [ 32.970622] ? srso_alias_return_thunk+0x5/0xfbef5 [ 32.971011] bus_add_driver+0x19a/0x2b0 [ 32.971335] driver_register+0xd8/0x160 [ 32.971671] ? __pfx_bochs_pci_driver_init+0x10/0x10 [bochs] [ 32.972130] do_one_initcall+0xba/0x390 [...] After unplugging the DRM device, clients will close their references. Closing the final reference will also release the DRM device. Reported-by: Dr. David Alan Gilbert <dave@xxxxxxxxxxx> Closes: https://lore.kernel.org/lkml/Z18dbfDAiFadsSdg@gallifrey/ Fixes: 04826f588682 ("drm/bochs: Allocate DRM device in struct bochs_device") Cc: Thomas Zimmermann <tzimmermann@xxxxxxx> Cc: Gerd Hoffmann <kraxel@xxxxxxxxxx> Cc: virtualization@xxxxxxxxxxxxxxx Signed-off-by: Thomas Zimmermann <tzimmermann@xxxxxxx> --- drivers/gpu/drm/tiny/bochs.c | 1 - 1 file changed, 1 deletion(-) diff --git a/drivers/gpu/drm/tiny/bochs.c b/drivers/gpu/drm/tiny/bochs.c index 89a699370a59..c67e1f906785 100644 --- a/drivers/gpu/drm/tiny/bochs.c +++ b/drivers/gpu/drm/tiny/bochs.c @@ -757,7 +757,6 @@ static void bochs_pci_remove(struct pci_dev *pdev) drm_dev_unplug(dev); drm_atomic_helper_shutdown(dev); - drm_dev_put(dev); } static void bochs_pci_shutdown(struct pci_dev *pdev) -- 2.47.1
