acpi: Fix possible recursive locking in hwregs.c

Calling pm-suspend might trigger a recursive lock in it's code path. In function acpi_hw_clear_acpi_status, acpi_os_release_lock holds the lock acpi_gbl_hardware_lock before calling acpi_hw_register_write(), then without releasing acpi_gbl_hardware_lock, this function calls acpi_ev_walk_gpe_list, which also tries to hold acpi_gbl_hardware_lock and thus causes possible recursive lock.

Following patch fixes this scenario by just releasing acpi_gbl_hardware_lock before calling acpi_ev_walk_gpe_list.

[This patch was created against 3.0-rc3. Since kernel.org is down, I don't have the updated tree. So, before applying it might require some adjustment.]

Signed-off-by Rakib Mullick <[email protected]>

---

drivers/acpi/acpica/hwregs.c +3 -0

index 55accb7..e3110ac 100644
@@ -269,6 +269,9 @@ acpi_status acpi_hw_clear_acpi_status(void)
 
        status = acpi_hw_register_write(ACPI_REGISTER_PM1_STATUS,
                                        ACPI_BITMASK_ALL_FIXED_STATUS);
+
+       acpi_os_release_lock(acpi_gbl_hardware_lock, lock_flags);
+
        if (ACPI_FAILURE(status)) {
                goto unlock_and_exit;
        }

@@ -278,7 +281,6 @@ acpi_status acpi_hw_clear_acpi_status(void)
        status = acpi_ev_walk_gpe_list(acpi_hw_clear_gpe_block, NULL);
 

      unlock_and_exit:
-       acpi_os_release_lock(acpi_gbl_hardware_lock, lock_flags); return_ACPI_STATUS(status);
 }

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Quoted 4 lines · Rakib Mullick

Calling pm-suspend might trigger a recursive lock in it's code path. In function acpi_hw_clear_acpi_status, acpi_os_release_lock holds the lock acpi_gbl_hardware_lock before calling acpi_hw_register_write(), then without releasing acpi_gbl_hardware_lock, this function calls acpi_ev_walk_gpe_list, which also tries to hold acpi_gbl_hardware_lock and thus causes possible recursive lock.

No, acpi_ev_walk_gpe_list holds acpi_gbl_gpe_lock.

Quoted 8 lines · Lin Ming

Quoted 4 lines · Rakib Mullick

Calling pm-suspend might trigger a recursive lock in it's code path. In function acpi_hw_clear_acpi_status, acpi_os_release_lock holds the lock acpi_gbl_hardware_lock before calling acpi_hw_register_write(), then without releasing acpi_gbl_hardware_lock, this function calls acpi_ev_walk_gpe_list, which also tries to hold acpi_gbl_hardware_lock and thus causes possible recursive lock.

No, acpi_ev_walk_gpe_list holds acpi_gbl_gpe_lock.

Yes, right. Actually, acpi_os_release_lock() tries to hold acpi_gbl_gpe_lock without releasing acpi_gbl_hardware_lock. Right?

Thanks,
Rakib

Quoted 7 lines · Rakib Mullick

Quoted 6 lines · Lin Ming

Quoted 4 lines · Rakib Mullick

Calling pm-suspend might trigger a recursive lock in it's code path. In function acpi_hw_clear_acpi_status, acpi_os_release_lock holds the lock acpi_gbl_hardware_lock before calling acpi_hw_register_write(), then without releasing acpi_gbl_hardware_lock, this function calls acpi_ev_walk_gpe_list, which also tries to hold acpi_gbl_hardware_lock and thus causes possible recursive lock.

Quoted 4 lines

Quoted 2 lines

No, acpi_ev_walk_gpe_list holds acpi_gbl_gpe_lock.

Yes, right. Actually, acpi_os_release_lock() tries to hold acpi_gbl_gpe_lock without releasing acpi_gbl_hardware_lock. Right?

Do you mean "acpi_ev_walk_gpe_list() tries to hold acpi_gbl_gpe_lock..."?

We have below locks sequence. I don't see the problem.

acpi_hw_clear_acpi_status:

    <acquire acpi_gbl_hardware_lock>

        acpi_ev_walk_gpe_list:
            <acquire acpi_gbl_gpe_lock>
            ...
            <release acpi_gbl_gpe_lock>
        ...
    <release acpi_gbl_hardware_lock>

Quoted 3 lines

Thanks,
Rakib