IB/hfi1: Fix potential deadlock on &irq_src_lock and &dd->uctxt_lock
authorChengfeng Ye <dg573847474@gmail.com>
Tue, 26 Sep 2023 10:11:16 +0000 (10:11 +0000)
committerLeon Romanovsky <leon@kernel.org>
Wed, 25 Oct 2023 08:12:43 +0000 (11:12 +0300)
commit2f19c4b8395ccb6eb25ccafee883c8cfbe3fc193
tree8191c7e67f09bb03bb793dadd601f4fac2d35ef5
parent7a1c2abf9a2be7d969b25e8d65567933335ca88e
IB/hfi1: Fix potential deadlock on &irq_src_lock and &dd->uctxt_lock

handle_receive_interrupt_napi_sp() running inside interrupt handler
could introduce inverse lock ordering between &dd->irq_src_lock
and &dd->uctxt_lock, if read_mod_write() is preempted by the isr.

          [CPU0]                                        |          [CPU1]
hfi1_ipoib_dev_open()                                   |
--> hfi1_netdev_enable_queues()                         |
--> enable_queues(rx)                                   |
--> hfi1_rcvctrl()                                      |
--> set_intr_bits()                                     |
--> read_mod_write()                                    |
--> spin_lock(&dd->irq_src_lock)                        |
                                                        | hfi1_poll()
                                                        | --> poll_next()
                                                        | --> spin_lock_irq(&dd->uctxt_lock)
                                                        |
                                                        | --> hfi1_rcvctrl()
                                                        | --> set_intr_bits()
                                                        | --> read_mod_write()
                                                        | --> spin_lock(&dd->irq_src_lock)
<interrupt>                                             |
   --> handle_receive_interrupt_napi_sp()               |
   --> set_all_fastpath()                               |
   --> hfi1_rcd_get_by_index()                          |
   --> spin_lock_irqsave(&dd->uctxt_lock)               |

This flaw was found by an experimental static analysis tool I am
developing for irq-related deadlock.

To prevent the potential deadlock, the patch use spin_lock_irqsave()
on &dd->irq_src_lock inside read_mod_write() to prevent the possible
deadlock scenario.

Signed-off-by: Chengfeng Ye <dg573847474@gmail.com>
Link: https://lore.kernel.org/r/20230926101116.2797-1-dg573847474@gmail.com
Acked-by: Dennis Dalessandro <dennis.dalessandro@cornelisnetworks.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
drivers/infiniband/hw/hfi1/chip.c