Closed arter97 closed 5 months ago
MSI helper functions within the Linux kernel may try to catch mutex [1][2] while we hold a spinlock with preemption disabled.
Fixes: #30 Link: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/kernel/irq/msi.c?h=v6.6#n448 [1] Link: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/kernel/irq/msi.c?h=v6.6#n334 [2]
Additionally, this lock specifically would go through the fastpath of mutex, so the performance should be identical to spin lock. My brief benchmark also doesn't show any performance difference.
MSI helper functions within the Linux kernel may try to catch mutex [1][2] while we hold a spinlock with preemption disabled.
Fixes: #30 Link: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/kernel/irq/msi.c?h=v6.6#n448 [1] Link: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/kernel/irq/msi.c?h=v6.6#n334 [2]