Clarify why Buffer is used instead of Package · Bumblebee-Project/bbswitch@ee0591b

Minor nit: "practice" (practise is a verb). Thanks for adding this!

Thanks for mentioning, I'll remember it for the future. I think I also made the same mistake for a nouveau patch, ah well.

acpibob · 2013-08-09T17:45:27Z

Index on a package is OK. Can you send or post a DSDT with an example of _DSM requiring a buffer?

Lekensteyn · 2013-08-10T09:18:34Z

Index on a Package is indeed OK, I thought it does not apply to Integer types, but it looks like there is also an implicit conversion rule for Integer -> Buffer? Example of _DSM that requires a Buffer:
https://github.com/Lekensteyn/acpi-stuff/blob/master/dsl/Clevo_B7130/SSDT1.dsl#L601

acpibob · 2013-08-13T17:23:58Z

Shouldn't this bug be fixed in the BIOS?

Lekensteyn · 2013-08-13T20:16:15Z

There are too many affected BIOSes which makes blacklisting a pain. I think that NVIDIA made a mistake in its reference implementation.

Some affected machines that use CreateField on Arg3:

Lenovo ThinkPad T420 (4236NVG, 4177CTO)
Clevo B7130
Clevo W150HNM-W170HN.0.1
Lenovo "4242RF1" (?)
Lenovo IdeaPad Y470
Acer Aspire 3830TG.V1.05
Dell Alienwaer M11xR3
Lenovo B560
Dell L502x
Packard Bell EasyNote TS11HR.V1.07
Acer Aspire 5755G.V1.15
(many others, simply grep for If (LEqual (_T_0, 0x1A)) with some context)

acpibob · 2013-08-13T20:21:23Z

Is the same device driver used on all of these machines (Same driver to execute the particular bad _DSM method?)

Lekensteyn · 2013-08-13T20:26:59Z

nouveau, nvidia, bbswitch are the known modules using this method. The device is always from vendor nvidia (10de) and is either of class "VGA Compatible Controller" or "3D controller". Are you going to add a blacklist in the acpi core for ignoring debug messages from this source?

acpibob · 2013-08-13T20:40:13Z

ACPICA core has no mechanism to "blacklist" individual machines. However, we may relax the (internal) definition of _DSM to allow either a buffer or a package. Not sure about this yet, however. I don't think the ACPI promoters are willing to change the spec for this at this time.

acpibob · 2013-08-13T20:56:30Z

The next question has to be: What do these machines do on windows? Does it appear that _DSM is invoked correctly?

Lekensteyn · 2013-08-13T21:09:11Z

The purpose of the function is to prepare for disabling the graphics card, that works.

ArchangeGabriel · 2013-08-14T10:33:41Z

I suppose that those lines in dmesg mean I'm affected too (Asus U43Jc) with a bug of this kind:

[   10.842446] ACPI Warning: \_SB_.PCI0.VGA_._DSM: Argument #4 type mismatch - Found [Integer], ACPI requires [Package] (20130517/nsarguments-95)
[   10.842488] ACPI Warning: \_SB_.PCI0.VGA_._DSM: Argument #4 type mismatch - Found [Integer], ACPI requires [Package] (20130517/nsarguments-95)
[   10.842509] ACPI Warning: \_SB_.PCI0.RP00.VGA_._DSM: Argument #4 type mismatch - Found [Integer], ACPI requires [Package] (20130517/nsarguments-95)
[   10.842523] ACPI Warning: \_SB_.PCI0.RP00.VGA_._DSM: Argument #4 type mismatch - Found [Integer], ACPI requires [Package] (20130517/nsarguments-95)

Lekensteyn · 2013-08-14T12:20:25Z

@ArchangeGabriel That seems to be https://bugzilla.kernel.org/show_bug.cgi?id=32602

acpibob · 2013-08-14T14:29:25Z

Argument #4 type mismatch - Found [Integer], ACPI requires [Package]

This looks like a case where the warning was useful -- the driver was passing an incorrect argument.

acpibob · 2013-08-14T19:19:11Z

Is Nvidia aware of this bug in their code?

Lekensteyn · 2013-08-14T21:39:14Z

I don't know and I think they would not care/fix this because of backwards compatibility. I just peeked in the GPL wrapper of the nvidia proprietary driver and found this in nv-acpi.c:1118 (version 313.26):

     // 3rd dsm argument always a dword. inParamSize checked on entry to this routine.
     dsm_params[3].buffer.type    = ACPI_TYPE_BUFFER;
     dsm_params[3].buffer.length  = inParamSize;
     memcpy(argument3, pInParams, dsm_params[3].buffer.length);
     dsm_params[3].buffer.pointer = argument3;

acpibob · 2013-08-15T17:15:02Z

I think Nvidia should be informed of this problem. Their code (both driver and ASL) is in clear violation of the ACPI specification.

ArchangeGabriel · 2013-11-24T13:54:10Z

Just for your information, the ACPI warning I had before disappeared, however I have those ones now, that are, I supposed, related to this commit this time:

[   32.246370] bbswitch: module verification failed: signature and/or required key missing - tainting kernel
[   32.246675] bbswitch: version 0.7
[   32.246682] bbswitch: Found integrated VGA device 0000:00:02.0: \_SB_.PCI0.VGA_
[   32.246689] bbswitch: Found discrete VGA device 0000:01:00.0: \_SB_.PCI0.RP00.VGA_
[   32.246704] ACPI Warning: \_SB_.PCI0.RP00.VGA_._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20130725/nsarguments-95)
[   32.246783] bbswitch: detected an Optimus _DSM function
[   32.246796] pci 0000:01:00.0: enabling device (0000 -> 0003)
[   32.246850] bbswitch: Succesfully loaded. Discrete card 0000:01:00.0 is on
[   32.249734] bbswitch: disabling discrete graphics
[   32.249747] ACPI Warning: \_SB_.PCI0.RP00.VGA_._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20130725/nsarguments-95)

Lekensteyn · 2013-11-24T14:09:53Z

@ArchangeGabriel The ACPI Warning here is harmless (though it can be annoying because it gets repeatedly printed during toggles). The module verification warning is unrelated, it means that the kernel you use (Ubuntu?) has module verification enabled and that the key was not available when building bbswitch (obviously, the keys must be kept secret or it would be rather useless to have module signing).

petamem · 2016-11-05T07:28:01Z

Harmless it may be, but +1 for the annoyance. 3y later - make at least some option to turn it off.

[  185.121418] NVRM: loading NVIDIA UNIX x86_64 Kernel Module  375.10  Fri Oct 14 10:30:06 PDT 2016 (using threaded interrupts)
[  185.124792] ACPI Warning: \_SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20160422/nsarguments-95)
[  185.124886] ACPI Warning: \_SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20160422/nsarguments-95)
[  185.124939] ACPI Warning: \_SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20160422/nsarguments-95)
[  185.124988] ACPI Warning: \_SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20160422/nsarguments-95)
[  185.125024] ACPI Warning: \_SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20160422/nsarguments-95)
[  185.125073] ACPI Warning: \_SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20160422/nsarguments-95)
[  185.125107] ACPI Warning: \_SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20160422/nsarguments-95)
[  185.224526] ACPI Warning: \_SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20160422/nsarguments-95)
[  185.521644] ACPI Warning: \_SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20160422/nsarguments-95)
[  204.924800] bbswitch: device 0000:01:00.0 is in use by driver 'nvidia', refusing OFF
[  207.374261] ACPI Warning: \_SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20160422/nsarguments-95)
[  207.374326] ACPI Warning: \_SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20160422/nsarguments-95)
[  207.374373] ACPI Warning: \_SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20160422/nsarguments-95)
[  207.374421] ACPI Warning: \_SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20160422/nsarguments-95)
[  207.374469] ACPI Warning: \_SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20160422/nsarguments-95)
[  207.374533] ACPI Warning: \_SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20160422/nsarguments-95)
[  207.374579] ACPI Warning: \_SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20160422/nsarguments-95)
[  207.378308] ACPI Warning: \_SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20160422/nsarguments-95)
[  207.517085] ACPI Warning: \_SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20160422/nsarguments-95)
[  214.540205] bbswitch: device 0000:01:00.0 is in use by driver 'nvidia', refusing OFF

Lekensteyn · 2016-11-05T12:53:26Z

@acpibob Btw, NVIDIA's Windows driver also passes Buffer instead of a Package, see this ACPI trace:
#115 (comment)

@petamem You could try to fill a bug with the ACPI subsystem in kernel bugzilla (https://bugzilla.kernel.org/), but it seems unlikely to get fixed since it is a violation of the ACPI spec.

RyanNerd · 2017-02-17T09:38:30Z

Ok so this is a feature and not a bug. Why do I have deja vu of Microsoft errors (error 5 occurred reporting error 5)?

acpibob · 2017-02-17T15:18:36Z

How about a mechanism to report the warning once only, then go silent?

acpibob · 2017-02-17T15:43:00Z

Actually, the mechanism already exists, but this particular warning is currently always enabled. Probably because it is rather serious (and should probably be an error, not just a warning).

mirh · 2017-03-26T01:28:52Z

~~wonders why robert and zeta seems to work on separate tracks~~ Ok, being divided by the ocean seems a valid reason.

Nevertheless, guess this should be fixed when acpica/acpica#189 lands?
It will be a good race to see if that pull maturation will come earlier or later than fglrx (which I'm pretty sure, together with this, are really suffering the otherwise wise choice of sharing Windows code) total degradation [:

zetalog · 2017-03-31T09:22:59Z

Nevertheless, guess this should be fixed when acpica/acpica#189 lands?

Not related, this bug is talking about implicit source conversion.
While #189 is talking about NameString in packages.

masli5 · 2017-11-23T10:22:19Z

The bug is really affecting my boot time, what exactly should I do to fix it?

Lekensteyn · 2017-11-23T17:31:56Z

@masli5 The message is not a bug. It is triggered when nouveau tries to do runtime PM on "older" laptops (changing power state introduces some delay). It is also triggered by the nouveau and proprietary nvidia driver which needs to lookup some information (like ROM).

masli5 · 2017-11-24T11:16:16Z

@Lekensteyn Ok, it's a messagge. However, during boot it is displayed about 15 times, in 6-7 times it follows a 5 second delay (so I'm assuming that nouveau is waiting for 5 seconds for whatever is sending the messagge to do its thing, but that doesn't happen). Since my laptop is a year old, with Core i7, nvidia gtx950m and 8GB RAM, I find 2 minutes boots a bit too much (am I wrong?) and the messagge seems to follow the main problem. Is there a fix?

mirh · 2017-11-24T11:18:25Z

ACPICA guys are like constantly working on it.

acpibob · 2017-11-30T16:53:06Z

I don't see why a few messages would increase boot time significantly.

Lekensteyn · 2017-11-30T19:30:58Z

The messages are a red herring, they occur also in normal situations. Some hybrid graphics laptops (Intel iGPU + Nvidia GTX >= 9xx) however have an issue where the ACPI code from the vendor is stuck in an infinite loop, polling for device power on. If the device never regains power, you can observe this delay. See also https://bugzilla.kernel.org/show_bug.cgi?id=156341

Most likely system sleep (suspend) is also broken for you and shutdown takes a while. @masli5

acpibob · 2017-11-30T21:50:21Z

The infinite loop timeout in ACPICA is configurable, you might try changing that. Other than that, there is not much that ACPICA can do about such things.

-Original file line number
+Diff line change
@@ Expand Up @@
         params[1].integer.value = revid;
         params[2].type = ACPI_TYPE_INTEGER;
         params[2].integer.value = func;
+        /* Although the ACPI spec defines Arg3 as a Package, in practise
+         * implementations expect a Buffer (CreateWordField and Index functions are
+         * applied to it). */
         params[3].type = ACPI_TYPE_BUFFER;
         params[3].buffer.length = 4;
         if (args) {
@@ Expand Down @@

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Commit

There are no files selected for viewing

This comment has been minimized.

This comment has been minimized.

31 comments on commit `ee0591b`

acpibob commented on `ee0591b` Aug 9, 2013

Lekensteyn commented on `ee0591b` Aug 10, 2013

acpibob commented on `ee0591b` Aug 13, 2013

Lekensteyn commented on `ee0591b` Aug 13, 2013

acpibob commented on `ee0591b` Aug 13, 2013

Lekensteyn commented on `ee0591b` Aug 13, 2013

acpibob commented on `ee0591b` Aug 13, 2013

acpibob commented on `ee0591b` Aug 13, 2013

Lekensteyn commented on `ee0591b` Aug 13, 2013

ArchangeGabriel commented on `ee0591b` Aug 14, 2013

Lekensteyn commented on `ee0591b` Aug 14, 2013

acpibob commented on `ee0591b` Aug 14, 2013 •

edited by Lekensteyn

acpibob commented on `ee0591b` Aug 14, 2013

Lekensteyn commented on `ee0591b` Aug 14, 2013

acpibob commented on `ee0591b` Aug 15, 2013

ArchangeGabriel commented on `ee0591b` Nov 24, 2013

Lekensteyn commented on `ee0591b` Nov 24, 2013

petamem commented on `ee0591b` Nov 5, 2016 •

edited by Lekensteyn

Lekensteyn commented on `ee0591b` Nov 5, 2016 •

edited

RyanNerd commented on `ee0591b` Feb 17, 2017

acpibob commented on `ee0591b` Feb 17, 2017

acpibob commented on `ee0591b` Feb 17, 2017

mirh commented on `ee0591b` Mar 26, 2017 •

edited

zetalog commented on `ee0591b` Mar 31, 2017

masli5 commented on `ee0591b` Nov 23, 2017

Lekensteyn commented on `ee0591b` Nov 23, 2017

masli5 commented on `ee0591b` Nov 24, 2017

mirh commented on `ee0591b` Nov 24, 2017

acpibob commented on `ee0591b` Nov 30, 2017

Lekensteyn commented on `ee0591b` Nov 30, 2017

acpibob commented on `ee0591b` Nov 30, 2017

Commit

There are no files selected for viewing

This comment has been minimized.

amonakov Aug 6, 2013

This comment has been minimized.

Lekensteyn Aug 6, 2013

31 comments on commit ee0591b

acpibob commented on ee0591b Aug 9, 2013

Choose a reason for hiding this comment

Lekensteyn commented on ee0591b Aug 10, 2013

Choose a reason for hiding this comment

acpibob commented on ee0591b Aug 13, 2013

Choose a reason for hiding this comment

Lekensteyn commented on ee0591b Aug 13, 2013

Choose a reason for hiding this comment

acpibob commented on ee0591b Aug 13, 2013

Choose a reason for hiding this comment

Lekensteyn commented on ee0591b Aug 13, 2013

Choose a reason for hiding this comment

acpibob commented on ee0591b Aug 13, 2013

Choose a reason for hiding this comment

acpibob commented on ee0591b Aug 13, 2013

Choose a reason for hiding this comment

Lekensteyn commented on ee0591b Aug 13, 2013

Choose a reason for hiding this comment

ArchangeGabriel commented on ee0591b Aug 14, 2013

Choose a reason for hiding this comment

Lekensteyn commented on ee0591b Aug 14, 2013

Choose a reason for hiding this comment

acpibob commented on ee0591b Aug 14, 2013 • edited by Lekensteyn

Choose a reason for hiding this comment

acpibob commented on ee0591b Aug 14, 2013

Choose a reason for hiding this comment

Lekensteyn commented on ee0591b Aug 14, 2013

Choose a reason for hiding this comment

acpibob commented on ee0591b Aug 15, 2013

Choose a reason for hiding this comment

ArchangeGabriel commented on ee0591b Nov 24, 2013

Choose a reason for hiding this comment

Lekensteyn commented on ee0591b Nov 24, 2013

Choose a reason for hiding this comment

petamem commented on ee0591b Nov 5, 2016 • edited by Lekensteyn

Choose a reason for hiding this comment

Lekensteyn commented on ee0591b Nov 5, 2016 • edited

Choose a reason for hiding this comment

RyanNerd commented on ee0591b Feb 17, 2017

Choose a reason for hiding this comment

acpibob commented on ee0591b Feb 17, 2017

Choose a reason for hiding this comment

acpibob commented on ee0591b Feb 17, 2017

Choose a reason for hiding this comment

mirh commented on ee0591b Mar 26, 2017 • edited

Choose a reason for hiding this comment

zetalog commented on ee0591b Mar 31, 2017

Choose a reason for hiding this comment

masli5 commented on ee0591b Nov 23, 2017

Choose a reason for hiding this comment

Lekensteyn commented on ee0591b Nov 23, 2017

Choose a reason for hiding this comment

masli5 commented on ee0591b Nov 24, 2017

Choose a reason for hiding this comment

mirh commented on ee0591b Nov 24, 2017

Choose a reason for hiding this comment

acpibob commented on ee0591b Nov 30, 2017

Choose a reason for hiding this comment

Lekensteyn commented on ee0591b Nov 30, 2017

Choose a reason for hiding this comment

acpibob commented on ee0591b Nov 30, 2017

Choose a reason for hiding this comment

31 comments on commit `ee0591b`

acpibob commented on `ee0591b` Aug 9, 2013

Lekensteyn commented on `ee0591b` Aug 10, 2013

acpibob commented on `ee0591b` Aug 13, 2013

Lekensteyn commented on `ee0591b` Aug 13, 2013

acpibob commented on `ee0591b` Aug 13, 2013

Lekensteyn commented on `ee0591b` Aug 13, 2013

acpibob commented on `ee0591b` Aug 13, 2013

acpibob commented on `ee0591b` Aug 13, 2013

Lekensteyn commented on `ee0591b` Aug 13, 2013

ArchangeGabriel commented on `ee0591b` Aug 14, 2013

Lekensteyn commented on `ee0591b` Aug 14, 2013

acpibob commented on `ee0591b` Aug 14, 2013 •

edited by Lekensteyn

acpibob commented on `ee0591b` Aug 14, 2013

Lekensteyn commented on `ee0591b` Aug 14, 2013

acpibob commented on `ee0591b` Aug 15, 2013

ArchangeGabriel commented on `ee0591b` Nov 24, 2013

Lekensteyn commented on `ee0591b` Nov 24, 2013

petamem commented on `ee0591b` Nov 5, 2016 •

edited by Lekensteyn

Lekensteyn commented on `ee0591b` Nov 5, 2016 •

edited

RyanNerd commented on `ee0591b` Feb 17, 2017

acpibob commented on `ee0591b` Feb 17, 2017

acpibob commented on `ee0591b` Feb 17, 2017

mirh commented on `ee0591b` Mar 26, 2017 •

edited

zetalog commented on `ee0591b` Mar 31, 2017

masli5 commented on `ee0591b` Nov 23, 2017

Lekensteyn commented on `ee0591b` Nov 23, 2017

masli5 commented on `ee0591b` Nov 24, 2017

mirh commented on `ee0591b` Nov 24, 2017

acpibob commented on `ee0591b` Nov 30, 2017

Lekensteyn commented on `ee0591b` Nov 30, 2017

acpibob commented on `ee0591b` Nov 30, 2017