https://bugzilla.kernel.org/show_bug.cgi?id=112911

--- Comment #2 from Lv Zheng <[email protected]> ---
The commit in question:
http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=efaed9b
may look wrong because it doesn't reset AOPOBJ_REG_CONNECTED when a region
handler is uninstalled.

But this the current ACPICA behavior.
So this is an existing issue.
Let me clarify:
ACPICA only invokes _REG(DISCONNECT) when a region object is freed (whose
reference count drops to 0).
So we can see "no handler" error messages after Linux uninstalls a region
handler.
BIOS tables are not notified by ACPICA of the _REG(DISCONNECT).
Thus region accesses can still be triggered from ACPI table and result such an
error.

There are only 2 cases a region object will be deleted:
1. When the owner table is unloaded. Unfortunately, ACPICA doesn't support
table unloading, so this is a dead case.
2. When the evaluation of a control method that creates the region object is
ended. Then it looks we needn't worry about this case, because the region
object is deleted and when the control method is evaluated again, it will be
created again with AOPOBJ_REG_CONNECTED unset.
So there won't be anything working differently from the existing code.

What I was really worrying about is:
If _REG(CONNECT) is invoked too early (for example, before the namespace is
fully loaded), and the _REG method node is not found for this region object
just because this process happens too early, then the handler driver surely
need the whole process to happen again after the namespace is loaded.
This is a known issue (but is also an existing issue) and only can be triggered
by early opregion handlers (for which, it is possible to access during the
table loading). And among the early opregion handlers, EC opregion handler is
the only one that need us to worry about.
While this case is fixed by this patch:
http://www.spinics.net/lists/linux-acpi/msg63558.html

So I don't think this commit is wrong.
What may cause reviewers to think it is wrong are all existing issues.

You really need to confirm it again by running on top of the new release
series.

Thanks and best regards
-Lv

-- 
You are receiving this mail because:
You are watching the assignee of the bug.

------------------------------------------------------------------------------
Site24x7 APM Insight: Get Deep Visibility into Application Performance
APM + Mobile APM + RUM: Monitor 3 App instances at just $35/Month
Monitor end-to-end web transactions and take corrective actions now
Troubleshoot faster and improve end-user experience. Signup Now!
http://pubads.g.doubleclick.net/gampad/clk?id=272487151&iu=/4140
_______________________________________________
acpi-bugzilla mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/acpi-bugzilla

Reply via email to