From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-pb0-f41.google.com ([209.85.160.41]:48382 "EHLO mail-pb0-f41.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752504Ab3FKMYw (ORCPT ); Tue, 11 Jun 2013 08:24:52 -0400 Message-ID: <51B7170E.2060608@gmail.com> Date: Tue, 11 Jun 2013 20:24:46 +0800 From: Jiang Liu MIME-Version: 1.0 To: "Alexander E. Patrakov" CC: Jiang Liu , Bjorn Helgaas , Yinghai Lu , "Rafael J. Wysocki" , linux-pci@vger.kernel.org, "linux-kernel@vger.kernel.org" Subject: Re: [PATCH] ACPIPHP: fix device destroying order issue in handling dock notification References: <1370951543-15841-1-git-send-email-jiang.liu@huawei.com> In-Reply-To: Content-Type: text/plain; charset=UTF-8 Sender: linux-pci-owner@vger.kernel.org List-ID: On Tue 11 Jun 2013 08:15:11 PM CST, Alexander E. Patrakov wrote: > 2013/6/11 Jiang Liu : >> Current ACPI glue logic expects that physical devices are destroyed >> before destroying companion ACPI devices, otherwise it will break the >> ACPI unbind logic and cause following warning messages: >> [ 185.026073] usb usb5: Oops, 'acpi_handle' corrupt >> [ 185.035150] pci 0000:1b:00.0: Oops, 'acpi_handle' corrupt >> [ 185.035515] pci 0000:18:02.0: Oops, 'acpi_handle' corrupt >> [ 180.013656] port1: Oops, 'acpi_handle' corrupt >> Please refer to https://bugzilla.kernel.org/attachment.cgi?id=104321 >> for full log message. > > This causes lockdep spew, see > https://bugzilla.kernel.org/attachment.cgi?id=104411 > > So, probably a NAK. > >> Above warning messages are caused by following scenario: >> 1) acpi_dock_notifier_call() queues a task (T1) onto kacpi_hotplug_wq >> 2) kacpi_hotplug_wq handles T1, which invokes acpi_dock_deferred_cb() >> ->dock_notify()-> handle_eject_request()->hotplug_dock_devices() >> 3) hotplug_dock_devices() first invokes registered hotplug callbacks to >> destroy physical devices, then destroys all affected ACPI devices. >> Everything seems perfect until now. But the acpiphp dock notification >> handler will queue another task (T2) onto kacpi_hotplug_wq to really >> destroy affected physical devices. >> 4) kacpi_hotplug_wq finishes T1, and all affected ACPI devices have >> been destroyed. >> 5) kacpi_hotplug_wq handles T2, which destroys all affected physical >> devices. >> >> So it breaks the ACPI glue expection because ACPI devices are destroyed >> in step 3 and physical devices are destroyed in step 5. >> >> Signed-off-by: Jiang Liu >> Cc: Bjorn Helgaas >> Cc: Yinghai Lu >> Cc: "Rafael J. Wysocki" >> Cc: linux-pci@vger.kernel.org >> Cc: linux-kernel@vger.kernel.org >> --- >> Hi all, >> We are trying to solve bug https://bugzilla.kernel.org/show_bug.cgi?id=59501 >> And seems there are multiple bugs behind bug 59501. This draft patch tries to >> fix one of those issues. I will send out form patchset once all issue have been >> resolved. >> >> Regards! >> Gerry >> --- >> drivers/pci/hotplug/acpiphp_glue.c | 31 +++++++++++++++++++------------ >> 1 file changed, 19 insertions(+), 12 deletions(-) >> >> diff --git a/drivers/pci/hotplug/acpiphp_glue.c b/drivers/pci/hotplug/acpiphp_glue.c >> index 716aa93..b132aca 100644 >> --- a/drivers/pci/hotplug/acpiphp_glue.c >> +++ b/drivers/pci/hotplug/acpiphp_glue.c >> @@ -61,7 +61,10 @@ static DEFINE_MUTEX(bridge_mutex); >> static void handle_hotplug_event_bridge (acpi_handle, u32, void *); >> static void acpiphp_sanitize_bus(struct pci_bus *bus); >> static void acpiphp_set_hpp_values(struct pci_bus *bus); >> -static void handle_hotplug_event_func(acpi_handle handle, u32 type, void *context); >> +static void __handle_hotplug_event_func(acpi_handle handle, u32 type, >> + void *context); >> +static void handle_hotplug_event_func(acpi_handle handle, u32 type, >> + void *context); >> static void free_bridge(struct kref *kref); >> >> /* callback routine to check for the existence of a pci dock device */ >> @@ -147,7 +150,7 @@ static int post_dock_fixups(struct notifier_block *nb, unsigned long val, >> >> >> static const struct acpi_dock_ops acpiphp_dock_ops = { >> - .handler = handle_hotplug_event_func, >> + .handler = __handle_hotplug_event_func, >> }; >> >> /* Check whether the PCI device is managed by native PCIe hotplug driver */ >> @@ -1065,20 +1068,13 @@ static void handle_hotplug_event_bridge(acpi_handle handle, u32 type, >> alloc_acpi_hp_work(handle, type, context, _handle_hotplug_event_bridge); >> } >> >> -static void _handle_hotplug_event_func(struct work_struct *work) >> +static void __handle_hotplug_event_func(acpi_handle handle, u32 type, >> + void *context) >> { >> - struct acpiphp_func *func; >> + struct acpiphp_func *func = context; >> char objname[64]; >> struct acpi_buffer buffer = { .length = sizeof(objname), >> .pointer = objname }; >> - struct acpi_hp_work *hp_work; >> - acpi_handle handle; >> - u32 type; >> - >> - hp_work = container_of(work, struct acpi_hp_work, work); >> - handle = hp_work->handle; >> - type = hp_work->type; >> - func = (struct acpiphp_func *)hp_work->context; >> >> acpi_scan_lock_acquire(); >> >> @@ -1115,6 +1111,17 @@ static void _handle_hotplug_event_func(struct work_struct *work) >> } >> >> acpi_scan_lock_release(); >> +} >> + >> +static void _handle_hotplug_event_func(struct work_struct *work) >> +{ >> + struct acpiphp_func *func; >> + struct acpi_hp_work *hp_work; >> + >> + hp_work = container_of(work, struct acpi_hp_work, work); >> + func = (struct acpiphp_func *)hp_work->context; >> + __handle_hotplug_event_func(hp_work->handle, hp_work->type, >> + hp_work->context); >> kfree(hp_work); /* allocated in handle_hotplug_event_func */ >> put_bridge(func->slot->bridge); >> } >> -- >> 1.8.1.2 >> > > > > -- > Alexander E. Patrakov Hi Alexander, Sorry for the deadlock, I have no machine for testing:( Below patch should fix the deadlock issue. Regards! ---- diff --git a/drivers/pci/hotplug/acpiphp_glue.c b/drivers/pci/hotplug/acpiphp_glue.c index 0302645..699b8ca 100644 --- a/drivers/pci/hotplug/acpiphp_glue.c +++ b/drivers/pci/hotplug/acpiphp_glue.c @@ -1075,8 +1075,6 @@ static void _handle_hotplug_event_func(acpi_handle handle, u32 type, struct acpi_buffer buffer = { .length = sizeof(objname), .pointer = objname }; - acpi_scan_lock_acquire(); - acpi_get_name(handle, ACPI_FULL_PATHNAME, &buffer); switch (type) { @@ -1108,8 +1106,6 @@ static void _handle_hotplug_event_func(acpi_handle handle, u32 type, warn("notify_handler: unknown event type 0x%x for %s\n", type, objname); break; } - - acpi_scan_lock_release(); } static void _handle_hotplug_event_cb(struct work_struct *work) @@ -1119,8 +1115,10 @@ static void _handle_hotplug_event_cb(struct work_struct *work) hp_work = container_of(work, struct acpi_hp_work, work); func = (struct acpiphp_func *)hp_work->context; + acpi_scan_lock_acquire(); _handle_hotplug_event_func(hp_work->handle, hp_work->type, hp_work->context); + acpi_scan_lock_release(); kfree(hp_work); /* allocated in handle_hotplug_event_func */ put_bridge(func->slot->bridge); }