]> www.infradead.org Git - users/jedix/linux-maple.git/commit
iommu/vt-d: Fix system hang on reboot -f
authorYunhui Cui <cuiyunhui@bytedance.com>
Mon, 10 Mar 2025 02:47:44 +0000 (10:47 +0800)
committerJoerg Roedel <jroedel@suse.de>
Mon, 10 Mar 2025 08:31:02 +0000 (09:31 +0100)
commit9ce7603ad3cbae64003087de57834e8c5c856a20
tree3f32785c3d8e398dd0d4ce3d18e786f532c516a8
parent0ad2507d5d93f39619fc42372c347d6006b64319
iommu/vt-d: Fix system hang on reboot -f

We found that executing the command ./a.out &;reboot -f (where a.out is a
program that only executes a while(1) infinite loop) can probabilistically
cause the system to hang in the intel_iommu_shutdown() function, rendering
it unresponsive. Through analysis, we identified that the factors
contributing to this issue are as follows:

1. The reboot -f command does not prompt the kernel to notify the
application layer to perform cleanup actions, allowing the application to
continue running.

2. When the kernel reaches the intel_iommu_shutdown() function, only the
BSP (Bootstrap Processor) CPU is operational in the system.

3. During the execution of intel_iommu_shutdown(), the function down_write
(&dmar_global_lock) causes the process to sleep and be scheduled out.

4. At this point, though the processor's interrupt flag is not cleared,
 allowing interrupts to be accepted. However, only legacy devices and NMI
(Non-Maskable Interrupt) interrupts could come in, as other interrupts
routing have already been disabled. If no legacy or NMI interrupts occur
at this stage, the scheduler will not be able to run.

5. If the application got scheduled at this time is executing a while(1)-
type loop, it will be unable to be preempted, leading to an infinite loop
and causing the system to become unresponsive.

To resolve this issue, the intel_iommu_shutdown() function should not
execute down_write(), which can potentially cause the process to be
scheduled out. Furthermore, since only the BSP is running during the later
stages of the reboot, there is no need for protection against parallel
access to the DMAR (DMA Remapping) unit. Therefore, the following lines
could be removed:

down_write(&dmar_global_lock);
up_write(&dmar_global_lock);

After testing, the issue has been resolved.

Fixes: 6c3a44ed3c55 ("iommu/vt-d: Turn off translations at shutdown")
Co-developed-by: Ethan Zhao <haifeng.zhao@linux.intel.com>
Signed-off-by: Ethan Zhao <haifeng.zhao@linux.intel.com>
Signed-off-by: Yunhui Cui <cuiyunhui@bytedance.com>
Link: https://lore.kernel.org/r/20250303062421.17929-1-cuiyunhui@bytedance.com
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Signed-off-by: Joerg Roedel <jroedel@suse.de>
drivers/iommu/intel/iommu.c