cxl: Avoid double free_irq() for psl,slice interrupts
authorVaibhav Jain <vaibhav@linux.vnet.ibm.com>
Fri, 2 Jun 2017 16:56:48 +0000 (22:26 +0530)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Fri, 9 Jun 2017 09:30:53 +0000 (11:30 +0200)
commited45509b550488f6a71a4b6cc5b12265e8c1f85d
treed3f1a060851fbe17a9a93320cce51c1bc4934bc5
parent6f9193ec044a8f72d8b6ae94a5c4ab6e8b0f00ca
cxl: Avoid double free_irq() for psl,slice interrupts

During an eeh call to cxl_remove can result in double free_irq of
psl,slice interrupts. This can happen if perst_reloads_same_image == 1
and call to cxl_configure_adapter() fails during slot_reset
callback. In such a case we see a kernel oops with following back-trace:

Oops: Kernel access of bad area, sig: 11 [#1]
Call Trace:
  free_irq+0x88/0xd0 (unreliable)
  cxl_unmap_irq+0x20/0x40 [cxl]
  cxl_native_release_psl_irq+0x78/0xd8 [cxl]
  pci_deconfigure_afu+0xac/0x110 [cxl]
  cxl_remove+0x104/0x210 [cxl]
  pci_device_remove+0x6c/0x110
  device_release_driver_internal+0x204/0x2e0
  pci_stop_bus_device+0xa0/0xd0
  pci_stop_and_remove_bus_device+0x28/0x40
  pci_hp_remove_devices+0xb0/0x150
  pci_hp_remove_devices+0x68/0x150
  eeh_handle_normal_event+0x140/0x580
  eeh_handle_event+0x174/0x360
  eeh_event_handler+0x1e8/0x1f0

This patch fixes the issue of double free_irq by checking that
variables that hold the virqs (err_hwirq, serr_hwirq, psl_virq) are
not '0' before un-mapping and resetting these variables to '0' when
they are un-mapped.

Cc: stable@vger.kernel.org
Signed-off-by: Vaibhav Jain <vaibhav@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/misc/cxl/native.c