| Message ID | 20161121165055.7041-2-hdegoede@redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 11/21/2016 05:50 PM, Hans de Goede wrote: > We need to call drm_helper_hpd_irq_event() on resume to properly detect > monitor connection / disconnection on some laptops, use hpd_work for > this to avoid deadlocks. > Hi, this seems to introduce a hang of nouveau in 4.10-rc if the gpu is runtime resumed while no displays are connected at all. I get a permanent hang - need to power cycle to recover - if i either a) Boot a MacPro test machine which has two discrete cards, one radeon, one geforce, but the displays are only connected to the radeon, whereas the nvidia gpu has no displays connected during boot. b) On a gmux'ed MacBookPro 2010 intel + nvidia if i switch to the intel card via vgaswitcheroo (echo IGD > ...vgaswitcheroo/switch) and then after nouveau has powered down the nvidia, i use echo ON > ...vgaswitcheroo/switch) to power up the nvidia again, but now with nothing connected to its outputs. I can prevent the hang if i either boot with nouveau.runpm=0, or connect displays to the nvidia in case a), or if i remove the new schedule_work() for hpd_work in the nouveau_pmops_runtime_resume() function. Otherwise i get a hanging gpu-manager process on Ubuntu and this repeating in my kernel log: [ 246.899424] INFO: task kworker/7:1:127 blocked for more than 120 seconds. [ 246.899476] Tainted: G I 4.9.0-rc8 #60 [ 246.899511] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. [ 246.899561] kworker/7:1 D 0 127 2 0x00000000 [ 246.899573] Workqueue: pm pm_runtime_work [ 246.899576] ffff917bcbeb2680 0000000000000000 ffff917c0e8d0000 ffff917c0d890000 [ 246.899582] ffff917c165da058 ffff9c86866c7b70 ffffffff8c920fab ffff917c0de3d360 [ 246.899587] 0000000000000086 000000000de3d360 ffff917c165da058 0000000000000000 [ 246.899593] Call Trace: [ 246.899601] [<ffffffff8c920fab>] ? __schedule+0x2fb/0xb30 [ 246.899605] [<ffffffff8c921820>] schedule+0x40/0x90 [ 246.899608] [<ffffffff8c5ff8da>] rpm_resume+0x14a/0x740 [ 246.899614] [<ffffffff8c0e1c10>] ? wake_atomic_t_function+0x60/0x60 [ 246.899617] [<ffffffff8c6012e3>] pm_runtime_forbid+0x43/0x50 [ 246.899678] [<ffffffffc08f67c5>] nouveau_pmops_runtime_suspend+0xc5/0xd0 [nouveau] [ 246.899684] [<ffffffff8c4ced8d>] pci_pm_runtime_suspend+0x5d/0x190 [ 246.899687] [<ffffffff8c4ced30>] ? pci_pm_runtime_resume+0xa0/0xa0 [ 246.899690] [<ffffffff8c5ff252>] __rpm_callback+0x32/0x70 [ 246.899693] [<ffffffff8c5ff2b4>] rpm_callback+0x24/0x80 [ 246.899695] [<ffffffff8c4ced30>] ? pci_pm_runtime_resume+0xa0/0xa0 [ 246.899698] [<ffffffff8c5fffee>] rpm_suspend+0x11e/0x6f0 [ 246.899701] [<ffffffff8c60149b>] pm_runtime_work+0x7b/0xc0 [ 246.899707] [<ffffffff8c0afe58>] process_one_work+0x1f8/0x750 [ 246.899710] [<ffffffff8c0afdd9>] ? process_one_work+0x179/0x750 [ 246.899713] [<ffffffff8c0b03fb>] worker_thread+0x4b/0x4f0 [ 246.899717] [<ffffffff8c0bf8fc>] ? preempt_count_sub+0x4c/0x80 [ 246.899720] [<ffffffff8c0b03b0>] ? process_one_work+0x750/0x750 [ 246.899723] [<ffffffff8c0b7212>] kthread+0x102/0x120 [ 246.899728] [<ffffffff8c0ef546>] ? trace_hardirqs_on_caller+0x16/0x1c0 [ 246.899732] [<ffffffff8c0b7110>] ? kthread_park+0x60/0x60 [ 246.899735] [<ffffffff8c929b2a>] ret_from_fork+0x2a/0x40 [ 246.899738] INFO: lockdep is turned off. [ 246.899751] INFO: task gpu-manager:1173 blocked for more than 120 seconds. [ 246.899796] Tainted: G I 4.9.0-rc8 #60 [ 246.899832] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. [ 246.899880] gpu-manager D 0 1173 1 0x00000000 [ 246.899884] ffff917bcbd38580 0000000000000000 ffff917c0e8d5180 ffff917bcbced180 [ 246.899889] ffff917c169da058 ffff9c8688573ce0 ffffffff8c920fab ffff917c0de3d360 [ 246.899894] 0000000000000086 000000000de3d360 ffff917c169da058 0000000000000000 [ 246.899899] Call Trace: [ 246.899903] [<ffffffff8c920fab>] ? __schedule+0x2fb/0xb30 [ 246.899906] [<ffffffff8c921820>] schedule+0x40/0x90 [ 246.899909] [<ffffffff8c5ff595>] __pm_runtime_barrier+0x95/0x140 [ 246.899913] [<ffffffff8c0e1c10>] ? wake_atomic_t_function+0x60/0x60 [ 246.899915] [<ffffffff8c60109b>] pm_runtime_barrier+0x5b/0xc0 [ 246.899919] [<ffffffff8c4cce6b>] pci_config_pm_runtime_get+0x3b/0x60 [ 246.899922] [<ffffffff8c4d1bac>] pci_read_config+0x7c/0x260 [ 246.899927] [<ffffffff8c30fe60>] sysfs_kf_bin_read+0x50/0x80 [ 246.899929] [<ffffffff8c30f91a>] kernfs_fop_read+0xba/0x1c0 [ 246.899934] [<ffffffff8c277d98>] __vfs_read+0x28/0x130 [ 246.899939] [<ffffffff8c3cb7be>] ? security_file_permission+0x9e/0xc0 [ 246.899942] [<ffffffff8c278dee>] ? rw_verify_area+0x4e/0xc0 [ 246.899945] [<ffffffff8c278ef6>] vfs_read+0x96/0x140 [ 246.899948] [<ffffffff8c27a4ba>] SyS_pread64+0x7a/0x90 [ 246.899952] [<ffffffff8c929885>] entry_SYSCALL_64_fastpath+0x23/0xc6 [ 246.899955] [<ffffffff8c0eb61f>] ? trace_hardirqs_off_caller+0x1f/0xc0 [ 246.899958] INFO: lockdep is turned off. [ 369.775936] INFO: task kworker/7:1:127 blocked for more than 120 seconds. [ 369.775989] Tainted: G I 4.9.0-rc8 #60 [ 369.776024] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. [ 369.776074] kworker/7:1 D 0 127 2 0x00000000 [ 369.776086] Workqueue: pm pm_runtime_work [ 369.776090] ffff917bcbeb2680 0000000000000000 ffff917c0e8d0000 ffff917c0d890000 [ 369.776096] ffff917c165da058 ffff9c86866c7b70 ffffffff8c920fab ffff917c0de3d360 [ 369.776101] 0000000000000086 000000000de3d360 ffff917c165da058 0000000000000000 [ 369.776106] Call Trace: [ 369.776114] [<ffffffff8c920fab>] ? __schedule+0x2fb/0xb30 [ 369.776118] [<ffffffff8c921820>] schedule+0x40/0x90 [ 369.776121] [<ffffffff8c5ff8da>] rpm_resume+0x14a/0x740 [ 369.776127] [<ffffffff8c0e1c10>] ? wake_atomic_t_function+0x60/0x60 [ 369.776130] [<ffffffff8c6012e3>] pm_runtime_forbid+0x43/0x50 [ 369.776192] [<ffffffffc08f67c5>] nouveau_pmops_runtime_suspend+0xc5/0xd0 [nouveau] [ 369.776199] [<ffffffff8c4ced8d>] pci_pm_runtime_suspend+0x5d/0x190 [ 369.776202] [<ffffffff8c4ced30>] ? pci_pm_runtime_resume+0xa0/0xa0 [ 369.776205] [<ffffffff8c5ff252>] __rpm_callback+0x32/0x70 [ 369.776207] [<ffffffff8c5ff2b4>] rpm_callback+0x24/0x80 [ 369.776210] [<ffffffff8c4ced30>] ? pci_pm_runtime_resume+0xa0/0xa0 [ 369.776212] [<ffffffff8c5fffee>] rpm_suspend+0x11e/0x6f0 [ 369.776216] [<ffffffff8c60149b>] pm_runtime_work+0x7b/0xc0 [ 369.776221] [<ffffffff8c0afe58>] process_one_work+0x1f8/0x750 [ 369.776224] [<ffffffff8c0afdd9>] ? process_one_work+0x179/0x750 [ 369.776228] [<ffffffff8c0b03fb>] worker_thread+0x4b/0x4f0 [ 369.776231] [<ffffffff8c0bf8fc>] ? preempt_count_sub+0x4c/0x80 [ 369.776234] [<ffffffff8c0b03b0>] ? process_one_work+0x750/0x750 [ 369.776237] [<ffffffff8c0b7212>] kthread+0x102/0x120 [ 369.776242] [<ffffffff8c0ef546>] ? trace_hardirqs_on_caller+0x16/0x1c0 [ 369.776246] [<ffffffff8c0b7110>] ? kthread_park+0x60/0x60 [ 369.776249] [<ffffffff8c929b2a>] ret_from_fork+0x2a/0x40 [ 369.776252] INFO: lockdep is turned off. thanks, -mario > Signed-off-by: Hans de Goede <hdegoede@redhat.com> > --- > drivers/gpu/drm/nouveau/nouveau_drm.c | 11 ++++++++++- > 1 file changed, 10 insertions(+), 1 deletion(-) > > diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c b/drivers/gpu/drm/nouveau/nouveau_drm.c > index 3100fd88..b564ab8 100644 > --- a/drivers/gpu/drm/nouveau/nouveau_drm.c > +++ b/drivers/gpu/drm/nouveau/nouveau_drm.c > @@ -692,7 +692,12 @@ nouveau_pmops_resume(struct device *dev) > return ret; > pci_set_master(pdev); > > - return nouveau_do_resume(drm_dev, false); > + ret = nouveau_do_resume(drm_dev, false); > + > + /* Monitors may have been connected / disconnected during suspend */ > + schedule_work(&nouveau_drm(drm_dev)->hpd_work); > + > + return ret; > } > > static int > @@ -766,6 +771,10 @@ nouveau_pmops_runtime_resume(struct device *dev) > nvif_mask(&device->object, 0x088488, (1 << 25), (1 << 25)); > vga_switcheroo_set_dynamic_switch(pdev, VGA_SWITCHEROO_ON); > drm_dev->switch_power_state = DRM_SWITCH_POWER_ON; > + > + /* Monitors may have been connected / disconnected during suspend */ > + schedule_work(&nouveau_drm(drm_dev)->hpd_work); > + > return ret; > } > >
Hi, On 01/24/2017 02:00 AM, Mario Kleiner wrote: > On 11/21/2016 05:50 PM, Hans de Goede wrote: >> We need to call drm_helper_hpd_irq_event() on resume to properly detect >> monitor connection / disconnection on some laptops, use hpd_work for >> this to avoid deadlocks. >> > > Hi, > > this seems to introduce a hang of nouveau in 4.10-rc if the gpu is runtime resumed while no displays are connected at all. > > I get a permanent hang - need to power cycle to recover - if i either > > a) Boot a MacPro test machine which has two discrete cards, one radeon, one geforce, but the displays are only connected to the radeon, whereas the nvidia gpu has no displays connected during boot. > > b) On a gmux'ed MacBookPro 2010 intel + nvidia if i switch to the intel card via vgaswitcheroo (echo IGD > ...vgaswitcheroo/switch) and then after nouveau has powered down the nvidia, i use echo ON > ...vgaswitcheroo/switch) to power up the nvidia again, but now with nothing connected to its outputs. > > I can prevent the hang if i either boot with nouveau.runpm=0, or connect displays to the nvidia in case a), or if i remove the new schedule_work() for hpd_work in the nouveau_pmops_runtime_resume() function. > > Otherwise i get a hanging gpu-manager process on Ubuntu and this repeating in my kernel log: Weird, Ben do you have any ideas what might be causing this ? Regards, Hans > > [ 246.899424] INFO: task kworker/7:1:127 blocked for more than 120 seconds. > [ 246.899476] Tainted: G I 4.9.0-rc8 #60 > [ 246.899511] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. > [ 246.899561] kworker/7:1 D 0 127 2 0x00000000 > [ 246.899573] Workqueue: pm pm_runtime_work > [ 246.899576] ffff917bcbeb2680 0000000000000000 ffff917c0e8d0000 ffff917c0d890000 > [ 246.899582] ffff917c165da058 ffff9c86866c7b70 ffffffff8c920fab ffff917c0de3d360 > [ 246.899587] 0000000000000086 000000000de3d360 ffff917c165da058 0000000000000000 > [ 246.899593] Call Trace: > [ 246.899601] [<ffffffff8c920fab>] ? __schedule+0x2fb/0xb30 > [ 246.899605] [<ffffffff8c921820>] schedule+0x40/0x90 > [ 246.899608] [<ffffffff8c5ff8da>] rpm_resume+0x14a/0x740 > [ 246.899614] [<ffffffff8c0e1c10>] ? wake_atomic_t_function+0x60/0x60 > [ 246.899617] [<ffffffff8c6012e3>] pm_runtime_forbid+0x43/0x50 > [ 246.899678] [<ffffffffc08f67c5>] nouveau_pmops_runtime_suspend+0xc5/0xd0 [nouveau] > [ 246.899684] [<ffffffff8c4ced8d>] pci_pm_runtime_suspend+0x5d/0x190 > [ 246.899687] [<ffffffff8c4ced30>] ? pci_pm_runtime_resume+0xa0/0xa0 > [ 246.899690] [<ffffffff8c5ff252>] __rpm_callback+0x32/0x70 > [ 246.899693] [<ffffffff8c5ff2b4>] rpm_callback+0x24/0x80 > [ 246.899695] [<ffffffff8c4ced30>] ? pci_pm_runtime_resume+0xa0/0xa0 > [ 246.899698] [<ffffffff8c5fffee>] rpm_suspend+0x11e/0x6f0 > [ 246.899701] [<ffffffff8c60149b>] pm_runtime_work+0x7b/0xc0 > [ 246.899707] [<ffffffff8c0afe58>] process_one_work+0x1f8/0x750 > [ 246.899710] [<ffffffff8c0afdd9>] ? process_one_work+0x179/0x750 > [ 246.899713] [<ffffffff8c0b03fb>] worker_thread+0x4b/0x4f0 > [ 246.899717] [<ffffffff8c0bf8fc>] ? preempt_count_sub+0x4c/0x80 > [ 246.899720] [<ffffffff8c0b03b0>] ? process_one_work+0x750/0x750 > [ 246.899723] [<ffffffff8c0b7212>] kthread+0x102/0x120 > [ 246.899728] [<ffffffff8c0ef546>] ? trace_hardirqs_on_caller+0x16/0x1c0 > [ 246.899732] [<ffffffff8c0b7110>] ? kthread_park+0x60/0x60 > [ 246.899735] [<ffffffff8c929b2a>] ret_from_fork+0x2a/0x40 > [ 246.899738] INFO: lockdep is turned off. > [ 246.899751] INFO: task gpu-manager:1173 blocked for more than 120 seconds. > [ 246.899796] Tainted: G I 4.9.0-rc8 #60 > [ 246.899832] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. > [ 246.899880] gpu-manager D 0 1173 1 0x00000000 > [ 246.899884] ffff917bcbd38580 0000000000000000 ffff917c0e8d5180 ffff917bcbced180 > [ 246.899889] ffff917c169da058 ffff9c8688573ce0 ffffffff8c920fab ffff917c0de3d360 > [ 246.899894] 0000000000000086 000000000de3d360 ffff917c169da058 0000000000000000 > [ 246.899899] Call Trace: > [ 246.899903] [<ffffffff8c920fab>] ? __schedule+0x2fb/0xb30 > [ 246.899906] [<ffffffff8c921820>] schedule+0x40/0x90 > [ 246.899909] [<ffffffff8c5ff595>] __pm_runtime_barrier+0x95/0x140 > [ 246.899913] [<ffffffff8c0e1c10>] ? wake_atomic_t_function+0x60/0x60 > [ 246.899915] [<ffffffff8c60109b>] pm_runtime_barrier+0x5b/0xc0 > [ 246.899919] [<ffffffff8c4cce6b>] pci_config_pm_runtime_get+0x3b/0x60 > [ 246.899922] [<ffffffff8c4d1bac>] pci_read_config+0x7c/0x260 > [ 246.899927] [<ffffffff8c30fe60>] sysfs_kf_bin_read+0x50/0x80 > [ 246.899929] [<ffffffff8c30f91a>] kernfs_fop_read+0xba/0x1c0 > [ 246.899934] [<ffffffff8c277d98>] __vfs_read+0x28/0x130 > [ 246.899939] [<ffffffff8c3cb7be>] ? security_file_permission+0x9e/0xc0 > [ 246.899942] [<ffffffff8c278dee>] ? rw_verify_area+0x4e/0xc0 > [ 246.899945] [<ffffffff8c278ef6>] vfs_read+0x96/0x140 > [ 246.899948] [<ffffffff8c27a4ba>] SyS_pread64+0x7a/0x90 > [ 246.899952] [<ffffffff8c929885>] entry_SYSCALL_64_fastpath+0x23/0xc6 > [ 246.899955] [<ffffffff8c0eb61f>] ? trace_hardirqs_off_caller+0x1f/0xc0 > [ 246.899958] INFO: lockdep is turned off. > [ 369.775936] INFO: task kworker/7:1:127 blocked for more than 120 seconds. > [ 369.775989] Tainted: G I 4.9.0-rc8 #60 > [ 369.776024] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. > [ 369.776074] kworker/7:1 D 0 127 2 0x00000000 > [ 369.776086] Workqueue: pm pm_runtime_work > [ 369.776090] ffff917bcbeb2680 0000000000000000 ffff917c0e8d0000 ffff917c0d890000 > [ 369.776096] ffff917c165da058 ffff9c86866c7b70 ffffffff8c920fab ffff917c0de3d360 > [ 369.776101] 0000000000000086 000000000de3d360 ffff917c165da058 0000000000000000 > [ 369.776106] Call Trace: > [ 369.776114] [<ffffffff8c920fab>] ? __schedule+0x2fb/0xb30 > [ 369.776118] [<ffffffff8c921820>] schedule+0x40/0x90 > [ 369.776121] [<ffffffff8c5ff8da>] rpm_resume+0x14a/0x740 > [ 369.776127] [<ffffffff8c0e1c10>] ? wake_atomic_t_function+0x60/0x60 > [ 369.776130] [<ffffffff8c6012e3>] pm_runtime_forbid+0x43/0x50 > [ 369.776192] [<ffffffffc08f67c5>] nouveau_pmops_runtime_suspend+0xc5/0xd0 [nouveau] > [ 369.776199] [<ffffffff8c4ced8d>] pci_pm_runtime_suspend+0x5d/0x190 > [ 369.776202] [<ffffffff8c4ced30>] ? pci_pm_runtime_resume+0xa0/0xa0 > [ 369.776205] [<ffffffff8c5ff252>] __rpm_callback+0x32/0x70 > [ 369.776207] [<ffffffff8c5ff2b4>] rpm_callback+0x24/0x80 > [ 369.776210] [<ffffffff8c4ced30>] ? pci_pm_runtime_resume+0xa0/0xa0 > [ 369.776212] [<ffffffff8c5fffee>] rpm_suspend+0x11e/0x6f0 > [ 369.776216] [<ffffffff8c60149b>] pm_runtime_work+0x7b/0xc0 > [ 369.776221] [<ffffffff8c0afe58>] process_one_work+0x1f8/0x750 > [ 369.776224] [<ffffffff8c0afdd9>] ? process_one_work+0x179/0x750 > [ 369.776228] [<ffffffff8c0b03fb>] worker_thread+0x4b/0x4f0 > [ 369.776231] [<ffffffff8c0bf8fc>] ? preempt_count_sub+0x4c/0x80 > [ 369.776234] [<ffffffff8c0b03b0>] ? process_one_work+0x750/0x750 > [ 369.776237] [<ffffffff8c0b7212>] kthread+0x102/0x120 > [ 369.776242] [<ffffffff8c0ef546>] ? trace_hardirqs_on_caller+0x16/0x1c0 > [ 369.776246] [<ffffffff8c0b7110>] ? kthread_park+0x60/0x60 > [ 369.776249] [<ffffffff8c929b2a>] ret_from_fork+0x2a/0x40 > [ 369.776252] INFO: lockdep is turned off. > > thanks, > -mario > > >> Signed-off-by: Hans de Goede <hdegoede@redhat.com> >> --- >> drivers/gpu/drm/nouveau/nouveau_drm.c | 11 ++++++++++- >> 1 file changed, 10 insertions(+), 1 deletion(-) >> >> diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c b/drivers/gpu/drm/nouveau/nouveau_drm.c >> index 3100fd88..b564ab8 100644 >> --- a/drivers/gpu/drm/nouveau/nouveau_drm.c >> +++ b/drivers/gpu/drm/nouveau/nouveau_drm.c >> @@ -692,7 +692,12 @@ nouveau_pmops_resume(struct device *dev) >> return ret; >> pci_set_master(pdev); >> >> - return nouveau_do_resume(drm_dev, false); >> + ret = nouveau_do_resume(drm_dev, false); >> + >> + /* Monitors may have been connected / disconnected during suspend */ >> + schedule_work(&nouveau_drm(drm_dev)->hpd_work); >> + >> + return ret; >> } >> >> static int >> @@ -766,6 +771,10 @@ nouveau_pmops_runtime_resume(struct device *dev) >> nvif_mask(&device->object, 0x088488, (1 << 25), (1 << 25)); >> vga_switcheroo_set_dynamic_switch(pdev, VGA_SWITCHEROO_ON); >> drm_dev->switch_power_state = DRM_SWITCH_POWER_ON; >> + >> + /* Monitors may have been connected / disconnected during suspend */ >> + schedule_work(&nouveau_drm(drm_dev)->hpd_work); >> + >> return ret; >> } >> >>
diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c b/drivers/gpu/drm/nouveau/nouveau_drm.c index 3100fd88..b564ab8 100644 --- a/drivers/gpu/drm/nouveau/nouveau_drm.c +++ b/drivers/gpu/drm/nouveau/nouveau_drm.c @@ -692,7 +692,12 @@ nouveau_pmops_resume(struct device *dev) return ret; pci_set_master(pdev); - return nouveau_do_resume(drm_dev, false); + ret = nouveau_do_resume(drm_dev, false); + + /* Monitors may have been connected / disconnected during suspend */ + schedule_work(&nouveau_drm(drm_dev)->hpd_work); + + return ret; } static int @@ -766,6 +771,10 @@ nouveau_pmops_runtime_resume(struct device *dev) nvif_mask(&device->object, 0x088488, (1 << 25), (1 << 25)); vga_switcheroo_set_dynamic_switch(pdev, VGA_SWITCHEROO_ON); drm_dev->switch_power_state = DRM_SWITCH_POWER_ON; + + /* Monitors may have been connected / disconnected during suspend */ + schedule_work(&nouveau_drm(drm_dev)->hpd_work); + return ret; }
We need to call drm_helper_hpd_irq_event() on resume to properly detect monitor connection / disconnection on some laptops, use hpd_work for this to avoid deadlocks. Signed-off-by: Hans de Goede <hdegoede@redhat.com> --- drivers/gpu/drm/nouveau/nouveau_drm.c | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-)