pciehp: fix a race between pciehp and removing operations by sysfs

hulk inclusion
category: bugfix
bugzilla: 16100,20881,https://gitee.com/openeuler/kernel/issues/I4OG3O?from=project-issue


CVE: NA

-------------------------------------------------

When I run a stress test about pcie hotplug and removing operations by
sysfs, I got a hange task, and the following call trace is printed.

 INFO: task irq/746-pciehp:41551 blocked for more than 120 seconds.
       Tainted: P        W  OE     4.19.25-
 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
 irq/746-pciehp  D    0 41551      2 0x00000228
 Call trace:
  __switch_to+0x94/0xe8
  __schedule+0x270/0x8b0
  schedule+0x2c/0x88
  schedule_preempt_disabled+0x14/0x20
  __mutex_lock.isra.1+0x1fc/0x540
  __mutex_lock_slowpath+0x24/0x30
  mutex_lock+0x80/0xa8
  pci_lock_rescan_remove+0x20/0x28
  pciehp_configure_device+0x30/0x140
  pciehp_handle_presence_or_link_change+0x35c/0x4b0
  pciehp_ist+0x1cc/0x1d0
  irq_thread_fn+0x30/0x80
  irq_thread+0x128/0x200
  kthread+0x134/0x138
  ret_from_fork+0x10/0x18
 INFO: task bash:6424 blocked for more than 120 seconds.
       Tainted: P        W  OE     4.19.25-
 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
 bash            D    0  6424   2231 0x00000200
 Call trace:
  __switch_to+0x94/0xe8
  __schedule+0x270/0x8b0
  schedule+0x2c/0x88
  schedule_timeout+0x224/0x448
  wait_for_common+0x198/0x2a0
  wait_for_completion+0x28/0x38
  kthread_stop+0x60/0x190
  __free_irq+0x1c0/0x348
  free_irq+0x40/0x88
  pcie_shutdown_notification+0x54/0x80
  pciehp_remove+0x30/0x50
  pcie_port_remove_service+0x3c/0x58
  device_release_driver_internal+0x1b4/0x250
  device_release_driver+0x28/0x38
  bus_remove_device+0xd4/0x160
  device_del+0x128/0x348
  device_unregister+0x24/0x78
  remove_iter+0x48/0x58
  device_for_each_child+0x6c/0xb8
  pcie_port_device_remove+0x2c/0x48
  pcie_portdrv_remove+0x5c/0x68
  pci_device_remove+0x48/0xd8
  device_release_driver_internal+0x1b4/0x250
  device_release_driver+0x28/0x38
  pci_stop_bus_device+0x84/0xb8
  pci_stop_and_remove_bus_device_locked+0x24/0x40
  remove_store+0xa4/0xb8
  dev_attr_store+0x44/0x60
  sysfs_kf_write+0x58/0x80
  kernfs_fop_write+0xe8/0x1f0
  __vfs_write+0x60/0x190
  vfs_write+0xac/0x1c0
  ksys_write+0x6c/0xd8
  __arm64_sys_write+0x24/0x30
  el0_svc_common+0xa0/0x180
  el0_svc_handler+0x38/0x78
  el0_svc+0x8/0xc

When we remove a slot by sysfs.
'pci_stop_and_remove_bus_device_locked()' will be called. This function
will get the global mutex lock 'pci_rescan_remove_lock', and remove the
slot. If the irq thread 'pciehp_ist' is still running, we will wait
until it exits.

If a pciehp interrupt happens immediately after we remove the slot by
sysfs, but before we free the pciehp irq in
'pci_stop_and_remove_bus_device_locked()'. 'pciehp_ist' will hung
because the global mutex lock 'pci_rescan_remove_lock' is held by the
sysfs operation. But the sysfs operation is waiting for the pciehp irq
thread 'pciehp_ist' ends. Then a hung task occurs.

So this two kinds of operation, removing through attention buttion and
removing through /sys/devices/pci***/remove, should not be excuted at
the same time. This patch add a global variable to mark that one of these
operations is under processing. When this variable is set,  if another
operation is requested, it will be rejected.

We use a global variable 'slot_being_removed_rescaned' to mark whether a
slot is being removed or rescaned. This will cause a slot hotplug
operation is delayed if another slot is being remove or rescaned. But
if these two slots are under different root ports, they should not
influence each other. This patch make the flag
'slot_being_removed_rescanned' per root port so that one slot hotplug
operation doesn't influence slots below another root port.

We record the root port in struct pci_dev when the pci device is
initialized and added into the system instead of using
'pcie_find_root_port()' to find the root port when we need it. Because
iterating the pci tree needs the protection of
'pci_lock_rescan_remove()'. This will make the problem more complexed
because the lock is very coarse-grained. We don't need to worry about
'use-after-free' because child pci devices are always removed before the
root port device is removed.

Signed-off-by: Xiongfeng Wang <wangxiongfeng2@huawei.com>
Reviewed-by: Hanjun Guo <guohanjun@huawei.com>
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Signed-off-by: Jialin Zhang <zhangjialin11@huawei.com>
Reviewed-by: Xiongfeng Wang <wangxiongfeng2@huawei.com>
Signed-off-by: Zheng Zengkai <zhengzengkai@huawei.com>