The driver may sleep under a spinlock, and the function call paths are: gma_power_begin (acquire the spinlock) gma_resume_pci pci_set_power_state __pci_start_power_transition (drivers/pci/pci.c) msleep --> may sleep

gma_power_begin (acquire the spinlock) gma_resume_pci pci_enable_device pci_enable_device_flags (drivers/pci/pci.c) do_pci_enable_device pci_set_power_state __pci_start_power_transition msleep --> may sleep

A possible fix is to replace msleep with mdelay in __pci_start_power_transition in drivers/pci/pci.c.

These bugs are found by my static analysis tool and my code review.

Thanks, Jia-Ju Bai