Drivers: hv: vmbus: Cleanup the vmbus unload path
v2 cover Drivers: hv: vmbus: Cleanup the vmbus unload path
This patch-set have several fixes to enable the clean unload of the vmbus. Typically, vmbus will not be unloadable when Linux is hosted on Hyper-V since the driver managing the root device needs the vmbus driver.
In this version of the patch-set, the patch Drivers: hv: vmbus: Implement the protocol for tearing down vmbus was modified based on feedback from Vitaly Kuznetsov.
Dexuan Cui (1):
hv: vmbus_free_channels(): remove the redundant free_channel()
K. Y. Srinivasan (2):
drivers: hv: vmbus: Get rid of some unused definitions
Drivers: hv: vmbus: Implement the protocol for tearing down vmbus
state
Vitaly Kuznetsov (2):
Drivers: hv: vmbus: introduce vmbus_acpi_remove
Drivers: hv: vmbus: unregister panic notifier on module unload
drivers/hv/channel_mgmt.c | 36 +++++++++++++++++++++++++++++++++--- drivers/hv/connection.c | 5 +++++ drivers/hv/hyperv_vmbus.h | 2 ++ drivers/hv/vmbus_drv.c | 16 +++++++++++++++- include/linux/hyperv.h | 20 +------------------- 5 files changed, 56 insertions(+), 23 deletions(-)
Signature · 1 line
1.7.4.1
PATCH v2 1/5 Drivers: hv: vmbus: introduce vmbus_acpi_remove
From: Vitaly Kuznetsov <[email protected]>
In case we do request_resource() in vmbus_acpi_add() we need to tear it down to be able to load the driver again. Otherwise the following crash in observed when hv_vmbus unload/load sequence is performed on a Generation2 instance:
[ 38.165701] BUG: unable to handle kernel paging request at ffffffffa00075a0 [ 38.166315] IP: [<ffffffff8107dc5f>] __request_resource+0x2f/0x50 [ 38.166315] PGD 1f34067 PUD 1f35063 PMD 3f723067 PTE 0 [ 38.166315] Oops: 0000 [#1] SMP [ 38.166315] Modules linked in: hv_vmbus(+) [last unloaded: hv_vmbus] [ 38.166315] CPU: 0 PID: 267 Comm: modprobe Not tainted 3.19.0-rc5_bug923184+ #486 [ 38.166315] Hardware name: Microsoft Corporation Virtual Machine/Virtual Machine, BIOS Hyper-V UEFI Release v1.0 11/26/2012 [ 38.166315] task: ffff88003f401cb0 ti: ffff88003f60c000 task.ti: ffff88003f60c000 [ 38.166315] RIP: 0010:[<ffffffff8107dc5f>] [<ffffffff8107dc5f>] __request_resource+0x2f/0x50 [ 38.166315] RSP: 0018:ffff88003f60fb58 EFLAGS: 00010286
--- drivers/hv/vmbus_drv.c | 10 ++++++++++ 1 files changed, 10 insertions(+), 0 deletions(-)
drivers/hv/vmbus_drv.c +10 -0
@@ -1035,6 +1035,15 @@ acpi_walk_err:return ret_val; } +static int vmbus_acpi_remove (struct acpi_device *device) +{ +int ret =0 ; + +if (hyperv_mmio.start && hyperv_mmio.end) + ret =release_resource (&hyperv_mmio); +return ret; +} +static const struct acpi_device_id vmbus_acpi_device_ids[] = { {"VMBUS" ,0 }, {"VMBus" ,0 },
@@ -1047,6 +1056,7 @@ static struct acpi_driver vmbus_acpi_driver = { .ids = vmbus_acpi_device_ids, .ops = { .add = vmbus_acpi_add, + .remove = vmbus_acpi_remove, }, };
Signature · 1 line
1.7.4.1
PATCH v2 2/5 Drivers: hv: vmbus: unregister panic notifier on module unload
From: Vitaly Kuznetsov <[email protected]>
Commit 96c1d0581d00f7abe033350edb021a9d947d8d81 ("Drivers: hv: vmbus: Add support for VMBus panic notifier handler") introduced atomic_notifier_chain_register() call on module load. We also need to call atomic_notifier_chain_unregister() on module unload as otherwise the following crash is observed when we bring hv_vmbus back:
[ 39.788877] BUG: unable to handle kernel paging request at ffffffffa00078a8 [ 39.788877] IP: [<ffffffff8109d63f>] notifier_call_chain+0x3f/0x80 ... [ 39.788877] Call Trace: [ 39.788877] [<ffffffff8109de7d>] __atomic_notifier_call_chain+0x5d/0x90 ... [ 39.788877] [<ffffffff8109d788>] ? atomic_notifier_chain_register+0x38/0x70 [ 39.788877] [<ffffffff8109d767>] ? atomic_notifier_chain_register+0x17/0x70 [ 39.788877] [<ffffffffa002814f>] hv_acpi_init+0x14f/0x1000 [hv_vmbus] [ 39.788877] [<ffffffff81002144>] do_one_initcall+0xd4/0x210
--- drivers/hv/vmbus_drv.c | 4 ++++ 1 files changed, 4 insertions(+), 0 deletions(-)
drivers/hv/vmbus_drv.c +4 -0
@@ -1108,6 +1108,10 @@ static void __exit vmbus_exit(void)hv_synic_clockevents_cleanup ();hv_remove_vmbus_irq ();vmbus_free_channels (); +if (ms_hyperv.features &HV_FEATURE_GUEST_CRASH_MSR_AVAILABLE ) { +atomic_notifier_chain_unregister (&panic_notifier_list, + &hyperv_panic_block); + }bus_unregister (&hv_bus);hv_cleanup ();for_each_online_cpu (cpu)
Signature · 1 line
1.7.4.1
PATCH v2 3/5 hv: vmbus_free_channels(): remove the redundant free_channel()
In reply to K. Y. SrinivasanFrom: Dexuan Cui <[email protected]>
free_channel() has been invoked invmbus_remove() -> hv_process_channel_removal(), or vmbus_remove() -> ... -> vmbus_close_internal() -> hv_process_channel_removal().
We also change to use list_for_each_entry_safe(), because the entry is removed in hv_process_channel_removal().
This patch fixes a bug in the vmbus unload path.
Thank Dan Carpenter for finding the issue!
--- drivers/hv/channel_mgmt.c | 11 ++++++++--- 1 files changed, 8 insertions(+), 3 deletions(-)
drivers/hv/channel_mgmt.c +8 -3
@@ -212,11 +212,16 @@ void hv_process_channel_removal(struct vmbus_channel *channel, u32 relid)void vmbus_free_channels (void ) { -struct vmbus_channel *channel; +struct vmbus_channel *channel, *tmp; + +list_for_each_entry_safe (channel, tmp, &vmbus_connection.chn_list, + listentry) { +/* if we don't set rescind to true, vmbus_close_internal() +* won't invoke hv_process_channel_removal(). +*/ + channel->rescind = true; -list_for_each_entry (channel, &vmbus_connection.chn_list, listentry) {vmbus_device_unregister (channel->device_obj); -free_channel (channel); } }
Signature · 1 line
1.7.4.1
PATCH v2 4/5 drivers: hv: vmbus: Get rid of some unused definitions
In reply to K. Y. SrinivasanGet rid of some unused definitions.
--- include/linux/hyperv.h | 19 ------------------- 1 files changed, 0 insertions(+), 19 deletions(-)
include/linux/hyperv.h +0 -19
@@ -389,10 +389,6 @@ enum vmbus_channel_message_type {CHANNELMSG_INITIATE_CONTACT =14 ,CHANNELMSG_VERSION_RESPONSE =15 ,CHANNELMSG_UNLOAD =16 , -#ifdef VMBUS_FEATURE_PARENT_OR_PEER_MEMORY_MAPPED_INTO_A_CHILD -CHANNELMSG_VIEWRANGE_ADD =17 , -CHANNELMSG_VIEWRANGE_REMOVE =18 , -#endif CHANNELMSG_COUNT };
@@ -549,21 +545,6 @@ struct vmbus_channel_gpadl_torndown {u32 gpadl; } __packed; -#ifdef VMBUS_FEATURE_PARENT_OR_PEER_MEMORY_MAPPED_INTO_A_CHILD -struct vmbus_channel_view_range_add { -struct vmbus_channel_message_header header; -PHYSICAL_ADDRESS viewrange_base; -u64 viewrange_length; -u32 child_relid; -} __packed; - -struct vmbus_channel_view_range_remove { -struct vmbus_channel_message_header header; -PHYSICAL_ADDRESS viewrange_base; -u32 child_relid; -} __packed; -#endif -struct vmbus_channel_relid_released {struct vmbus_channel_message_header header;u32 child_relid;
Signature · 1 line
1.7.4.1
PATCH v2 5/5 Drivers: hv: vmbus: Implement the protocol for tearing down vmbus state
In reply to K. Y. SrinivasanImplement the protocol for tearing down the monitor state established with the host.
---
Changes in V2: Call vmbus_disconnect earlier - Vitaly Kuznetsov <[email protected]>
drivers/hv/channel_mgmt.c | 25 +++++++++++++++++++++++++ drivers/hv/connection.c | 5 +++++ drivers/hv/hyperv_vmbus.h | 2 ++ drivers/hv/vmbus_drv.c | 2 +- include/linux/hyperv.h | 1 + 5 files changed, 34 insertions(+), 1 deletions(-)
drivers/hv/channel_mgmt.c +25 -0
@@ -422,6 +422,30 @@ static void init_vp_index(struct vmbus_channel *channel, const uuid_le *type_gui }/* +* vmbus_unload_response - Handler for the unload response. +*/ +static void vmbus_unload_response (struct vmbus_channel_message_header *hdr) +{ +/* +* This is a global event; just wakeup the waiting thread. +* Once we successfully unload, we can cleanup the monitor state. +*/ +complete (&vmbus_connection.unload_event); +} + +void vmbus_initiate_unload (void ) +{ +struct vmbus_channel_message_header hdr; + +init_completion (&vmbus_connection.unload_event); +memset (&hdr,0 ,sizeof (struct vmbus_channel_message_header )); + hdr.msgtype =CHANNELMSG_UNLOAD ; +vmbus_post_msg (&hdr,sizeof (struct vmbus_channel_message_header )); + +wait_for_completion (&vmbus_connection.unload_event); +} + +/* * vmbus_onoffer - Handler for channel offers from vmbus in parent partition. * */
@@ -717,6 +741,7 @@ struct vmbus_channel_message_table_entry {CHANNELMSG_INITIATE_CONTACT, 0, NULL}, {CHANNELMSG_VERSION_RESPONSE, 1, vmbus_onversion_response}, {CHANNELMSG_UNLOAD, 0, NULL}, + {CHANNELMSG_UNLOAD_RESPONSE, 1, vmbus_unload_response}, };/*
drivers/hv/connection.c +5 -0
@@ -227,6 +227,11 @@ cleanup:void vmbus_disconnect (void ) { +/* +* First send the unload request to the host. +*/ +vmbus_initiate_unload (); +if (vmbus_connection.work_queue) {drain_workqueue (vmbus_connection.work_queue);destroy_workqueue (vmbus_connection.work_queue);
drivers/hv/hyperv_vmbus.h +2 -0
@@ -647,6 +647,7 @@ struct vmbus_connection {atomic_t next_gpadl_handle; +struct completion unload_event;/* * Represents channel interrupts. Each bit position represents a * channel. When a channel sends an interrupt via VMBUS, it finds its
@@ -741,6 +742,7 @@ void hv_vss_onchannelcallback(void *);int hv_fcopy_init (struct hv_util_service *);void hv_fcopy_deinit (void );void hv_fcopy_onchannelcallback (void *); +void vmbus_initiate_unload (void );static inline void hv_poll_channel (struct vmbus_channel *channel,void (*cb)(void *))
drivers/hv/vmbus_drv.c +1 -1
@@ -1106,6 +1106,7 @@ static void __exit vmbus_exit(void) vmbus_connection.conn_state =DISCONNECTED ;hv_synic_clockevents_cleanup (); +vmbus_disconnect ();hv_remove_vmbus_irq ();vmbus_free_channels ();if (ms_hyperv.features &HV_FEATURE_GUEST_CRASH_MSR_AVAILABLE ) {
@@ -1118,7 +1119,6 @@ static void __exit vmbus_exit(void)smp_call_function_single (cpu, hv_synic_cleanup,NULL ,1 );acpi_bus_unregister_driver (&vmbus_acpi_driver);hv_cpu_hotplug_quirk (false); -vmbus_disconnect (); }
include/linux/hyperv.h +1 -0
@@ -389,6 +389,7 @@ enum vmbus_channel_message_type {CHANNELMSG_INITIATE_CONTACT =14 ,CHANNELMSG_VERSION_RESPONSE =15 ,CHANNELMSG_UNLOAD =16 , +CHANNELMSG_UNLOAD_RESPONSE =17 , CHANNELMSG_COUNT };
Signature · 1 line
1.7.4.1