RE: [PATCH] [2/2] Allow override acpi_gbl_pre_defined_names (aka change that _OS).

RE: [PATCH] [2/2] Allow override acpi_gbl_pre_defined_names (aka change that _OS).

[Grover, Andrew]

> From: Ducrot Bruno [mailto:ducrot-kk6yZipjEM5g9hUCZPvPmw@public.gmane.org] 
> This is the linux part.  It permit to pass a apci_os_name= boot args
> in order to override the \_OS.

I'm still not clear on why this needs to be changeable at runtime. If we do
decide that the BIOS writers are so Windows-centric that we need to
masquerade as one, why wouldn't we just change the #define and recompile?

Dave, can you help me understand how a distro would find this useful, over
just statically changing the reported _OS string?

Regards -- Andy


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Re: [PATCH] [2/2] Allow override acpi_gbl_pre_defined_names (aka change that _OS).

[Ducrot Bruno]

On Wed, Feb 05, 2003 at 10:55:45AM -0800, Grover, Andrew wrote:
> > From: Ducrot Bruno [mailto:ducrot-kk6yZipjEM5g9hUCZPvPmw@public.gmane.org] 
> > This is the linux part.  It permit to pass a apci_os_name= boot args
> > in order to override the \_OS.
> 
> I'm still not clear on why this needs to be changeable at runtime. If we do
> decide that the BIOS writers are so Windows-centric that we need to
> masquerade as one, why wouldn't we just change the #define and recompile?
> 
> Dave, can you help me understand how a distro would find this useful, over
> just statically changing the reported _OS string?

I can not speak for Dave, but suppose that you say to an end user:
Hum, yes, your mouse do not work after S1..

Get the latest kernel of the distro via apt-get, rpm, what ever, then
edit the file /path/to/kernel/include/acpi/platform/aclinux.h

find the
#define ACPI_OS_NAME                "Linux"
line then change so that
#define ACPI_OS_NAME                "Microsoft Windows NT"

recompile, etc.

Now suppose that you said to the same end user:
Add acpi_os_name="Microsoft Windows NT" at boot time, and now
your mouse will work without any tricks after S1 and with gpm
or under X.


Now, I don't like the idea to have a 
include/acpi/platform/aclinux.h
which say that my OS is a MS one, and I guess nobody want it.


-- 
Ducrot Bruno

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Re: RE: [PATCH] [2/2] Allow override acpi_gbl_pre_defined_names (aka change that _OS).

[Matthew Wilcox]

On Wed, Feb 05, 2003 at 10:55:45AM -0800, Grover, Andrew wrote:
> Dave, can you help me understand how a distro would find this useful, over
> just statically changing the reported _OS string?

Dave's point is that a distro cannot change it, since it would work for
most users as "Linux".  But it offers the opportunity for users to try
(on an individual basis) to pretend to be NT.

-- 
"It's not Hollywood.  War is real, war is primarily not about defeat or
victory, it is about death.  I've seen thousands and thousands of dead bodies.
Do you think I want to have an academic debate on this subject?" -- Robert Fisk


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Re: [PATCH] [2/2] Allow override acpi_gbl_pre_defined_names (aka change that _OS).

[Dave Jones]

On Wed, Feb 05, 2003 at 10:55:45AM -0800, Grover, Andrew wrote:

 > I'm still not clear on why this needs to be changeable at runtime. If we do
 > decide that the BIOS writers are so Windows-centric that we need to
 > masquerade as one, why wouldn't we just change the #define and recompile?
 > 
 > Dave, can you help me understand how a distro would find this useful, over
 > just statically changing the reported _OS string?

Simple. Vendor would only need to ship a single image to support
all possible OS name overrides. CD space is precious, and with SMP,
highmem, P4/K7 optimised, preempt and a number of other such options, the number
of kernel combinations grows and grows.  Having a set of images compiled
saying "NT" and another set saying "95" and another set saying "Linux"
triples the numbers of kernels needed to be shipped, and hence, is
undesired.

Clearer?

        Dave

-- 
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Re: RE: [PATCH] [2/2] Allow override acpi_gbl_pre_defined_names (aka change that _OS).

[Pavel Machek]

Hi!

> > This is the linux part.  It permit to pass a apci_os_name= boot args
> > in order to override the \_OS.
> 
> I'm still not clear on why this needs to be changeable at runtime. If we do
> decide that the BIOS writers are so Windows-centric that we need to
> masquerade as one, why wouldn't we just change the #define and recompile?
> 
> Dave, can you help me understand how a distro would find this useful, over
> just statically changing the reported _OS string?

Well, if we make it runtime-configurable, we can later use blacklist
entries to fool broken bioses... And besides, it is probably not
acceptable for any distribution to ship kernel presenting itself as
"Microsoft Windows".
								Pavel

-- 
Worst form of spam? Adding advertisment signatures ala sourceforge.net.
What goes next? Inserting advertisment *into* email?


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Re: RE: [PATCH] [2/2] Allow override acpi_gbl_pre_defined_names (aka change that _OS).

[Derek Broughton]

From: "Grover, Andrew" <andrew.grover-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

> > From: Ducrot Bruno [mailto:ducrot-kk6yZipjEM5g9hUCZPvPmw@public.gmane.org]
> > This is the linux part.  It permit to pass a apci_os_name= boot args
> > in order to override the \_OS.
>
> I'm still not clear on why this needs to be changeable at runtime. If we do
> decide that the BIOS writers are so Windows-centric that we need to
> masquerade as one, why wouldn't we just change the #define and recompile?

Because many of us have Windows-only BIOSes, and really prefer NOT to recompile.
If not for ACPI, I would be using pre-packaged kernels.

I like Ducrot's idea :-)



-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Re: RE: [PATCH] [2/2] Allow override acpi_gbl_pre_defined_names (aka change that _OS).

[Jean-Pierre Schwickerath]

> I like Ducrot's idea :-)

His first name is Bruno, he is French and French people are used to
write their family name first and then their first name. ;-)


Jean-Pierre
-- 
Powered by Linux From Scratch - http://schwicky.net/
PGP Key ID: 0xEE6F49B4 - AIM/Jabber: Schwicky - ICQ: 4690141

Nothing is impossible... Everything is relative!


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Re: RE: [PATCH] [2/2] Allow override acpi_gbl_pre_defined_names (aka change that _OS).

[Derek Broughton]

>
> > I like Ducrot's idea :-)
>
> His first name is Bruno, he is French and French people are used to
> write their family name first and then their first name. ;-)

Well thanks!  Sorry Bruno.

That's one problem with the Internet - you never do know which parts of names
are personal and which are family names.  Usually with European names I have a
good idea but I still try to avoid using them until I'm certain - or I slip up.
"M. Ducrot" would have worked better :-).  Here, French-Canadians don't use that
convention.

derek



-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Re: RE: [PATCH] [2/2] Allow override acpi_gbl_pre_defined_names (aka change that _OS).

[Ducrot Bruno]

Hi Derek,

On Thu, Feb 06, 2003 at 09:57:26AM -0400, Derek Broughton wrote:
> >
> > > I like Ducrot's idea :-)
> >
> > His first name is Bruno, he is French and French people are used to
> > write their family name first and then their first name. ;-)
> 
> Well thanks!  Sorry Bruno.

Why?  Using the family name as if it where the firstname is ok
at least in my country.

Cheers,

-- 
Ducrot Bruno

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Intel ACPI

[Margit Schubert-While]

Hi,
	So, What do you think about this (D845):

Method (OSFL, 0, NotSerialized)
     {
         Store (0x00, Local0)
         If (MCTH (\_OS, "Microsoft Windows"))
         {
             Store (0x01, Local0)
         }

         If (MCTH (\_OS, "Microsoft WindowsME: Millennium Edition"))
         {
             Store (0x01, Local0)
         }

         Return (Local0)
     }

Method (_PS3, 0, NotSerialized)
                     {
                         If (MCTH (\_OS, "Microsoft Windows"))
                         {
                             Store (_GTM (), TIM1)
                         }
                     }

Method (_PS0, 0, NotSerialized)
                     {
                         If (MCTH (\_OS, "Microsoft Windows"))
                         {
                             If (BOOT)
                             {
                                 If (And (DRVI, 0x20))
                                 {
                                     PRGT (MPIO, MDMA, FLGS, 0xA0)
                                 }

                                 If (And (DRVI, 0x80))
                                 {
                                     PRGT (SPIO, SDMA, FLGS, 0xB0)
                                 }
                             }

                             Store (0x01, BOOT)
                         }
                     }


Method (_PS3, 0, NotSerialized)
                     {
                         If (MCTH (\_OS, "Microsoft Windows"))
                         {
                             Store (_GTM (), TIM1)
                         }
                     }


                    Method (_PS0, 0, NotSerialized)
                     {
                         If (MCTH (\_OS, "Microsoft Windows"))
                         {
                             If (BOOT)
                             {
                                 If (And (DRVI, 0x20))
                                 {
                                     PRGT (MPIO, MDMA, FLGS, 0xA0)
                                 }

                                 If (And (DRVI, 0x80))
                                 {
                                     PRGT (SPIO, SDMA, FLGS, 0xB0)
                                 }
                             }

                             Store (0x01, BOOT)
                         }
                     }



-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Re: Intel ACPI

[Ducrot Bruno]

On Tue, Feb 04, 2003 at 12:14:13AM +0100, Margit Schubert-While wrote:
> Hi,
> 	So, What do you think about this (D845):
> 

Nothing, apart that it will not work reliability with Linux or
any flavour of free *BSD.

-- 
Ducrot Bruno

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Re: Intel ACPI

[Dave Jones]

On Tue, Feb 04, 2003 at 10:45:40AM +0100, Ducrot Bruno wrote:
 > > 	So, What do you think about this (D845):
 > Nothing, apart that it will not work reliability with Linux or
 > any flavour of free *BSD.

I wonder if it's worth allowing the user to override \_OS at boottime.

        Dave

-- 
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Re: Intel ACPI

[Ducrot Bruno]

On Tue, Feb 04, 2003 at 01:22:51PM +0100, Dave Jones wrote:
> On Tue, Feb 04, 2003 at 10:45:40AM +0100, Ducrot Bruno wrote:
>  > > 	So, What do you think about this (D845):
>  > Nothing, apart that it will not work reliability with Linux or
>  > any flavour of free *BSD.
> 
> I wonder if it's worth allowing the user to override \_OS at boottime.

In fact, I *always* change my \_OS string.

-- 
Ducrot Bruno

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Re: Intel ACPI

[Dave Jones]

On Tue, Feb 04, 2003 at 04:43:59PM +0100, Ducrot Bruno wrote:

 > > > Nothing, apart that it will not work reliability with Linux or
 > > > any flavour of free *BSD.
 > > I wonder if it's worth allowing the user to override \_OS at boottime.
 > In fact, I *always* change my \_OS string.

Interesting. For any noticable gain ?

        Dave

-- 
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Re: Intel ACPI

[Ducrot Bruno]

On Tue, Feb 04, 2003 at 04:46:42PM +0100, Dave Jones wrote:
> On Tue, Feb 04, 2003 at 04:43:59PM +0100, Ducrot Bruno wrote:
> 
>  > > > Nothing, apart that it will not work reliability with Linux or
>  > > > any flavour of free *BSD.
>  > > I wonder if it's worth allowing the user to override \_OS at boottime.
>  > In fact, I *always* change my \_OS string.
> 
> Interesting. For any noticable gain ?

S1 and S4bios work better, especially in 2.4 series, if \_OS is set as
'Microsoft Windows NT", because this OS have the same kind of support
for suspending/resuming devices as Linux 2.4 in general...

2.5 is another story.  I am not sure for now what is 'best'.

-- 
Ducrot Bruno

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Re: Intel ACPI

[Zdeněk OGAR Skalák]

Hi,

Ducrot Bruno wrote:
> 
> On Tue, Feb 04, 2003 at 04:46:42PM +0100, Dave Jones wrote:
> > On Tue, Feb 04, 2003 at 04:43:59PM +0100, Ducrot Bruno wrote:
> >
> >  > > > Nothing, apart that it will not work reliability with Linux or
> >  > > > any flavour of free *BSD.
> >  > > I wonder if it's worth allowing the user to override \_OS at boottime.
> >  > In fact, I *always* change my \_OS string.
> >
> > Interesting. For any noticable gain ?
> 
> S1 and S4bios work better, especially in 2.4 series, if \_OS is set as
> 'Microsoft Windows NT", because this OS have the same kind of support
> for suspending/resuming devices as Linux 2.4 in general...
> 
> 2.5 is another story.  I am not sure for now what is 'best'.

	So could we add kernel boot argument to allow changig/overriding the \_OS
string? Eg. acpi_os="Microsoft Windows NT" to allow us to play with it without
rekompiling the kernel ?

	Zdenek OGAR Skalak

-- 
Ing. Zdeněk OGAR Skalák
Monet+ a.s.
Zámecká 365
763 14 Zlín - Štípa, CZ
Tel: +420 / 577 110 411,  Fax: +420 / 577 914 557


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Re: Intel ACPI

[Dave Jones]

On Wed, Feb 05, 2003 at 07:29:06AM +0100, Zdenìk OGAR Skalák wrote:
 > 	So could we add kernel boot argument to allow changig/overriding the \_OS
 > string? Eg. acpi_os="Microsoft Windows NT" to allow us to play with it without
 > rekompiling the kernel ?

A bootarg does seem to be the most sensible way to do it. You need
to set it early enough so that you catch everything. Doing it later
through a sysctl or the like would be too late, and doing it statically
at compile time would make it useless for vendor kernels.

        Dave

-- 
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

[PATCH] [1/2] Allow override acpi_gbl_pre_defined_names (aka change that _OS).

[Ducrot Bruno]

Hi!

This patch allow an OS vendor to be able to change some values of the
acpi_gbl_pre_defined_names global variable.

The main purpose actually is to override the \_OS part, in order
to test/try etc. when a DSDT was more specifically written for
some Redmond OSes.

In order to get it to work, OS must provide an acpi_os_predefined_override()
(or AcpiOsPredefinedOverride) function.
One example is at the end of this patch.

A patch using a bootarg in order to change the \_OS string for Linux follow.



--- linux/drivers/acpi/namespace/nsaccess.c	2003/02/05 14:25:40	1.1
+++ linux/drivers/acpi/namespace/nsaccess.c	2003/02/05 15:23:59
@@ -100,6 +100,16 @@
 		 * initial value, create the initial value.
 		 */
 		if (init_val->val) {
+			acpi_string val;
+
+			status = acpi_os_predefined_override(init_val, &val);
+			if (ACPI_FAILURE (status)) {
+				ACPI_DEBUG_PRINT ((ACPI_DB_ERROR,
+					"Could not override predefined %s\n", init_val->name));
+			}
+			if (val == NULL)
+				val = init_val->val;
+
 			/*
 			 * Entry requests an initial value, allocate a
 			 * descriptor for it.
@@ -118,7 +128,7 @@
 			switch (init_val->type) {
 			case ACPI_TYPE_METHOD:
 				obj_desc->method.param_count =
-						(u8) ACPI_STRTOUL (init_val->val, NULL, 10);
+						(u8) ACPI_STRTOUL (val, NULL, 10);
 				obj_desc->common.flags |= AOPOBJ_DATA_VALID;
 
 #if defined (ACPI_NO_METHOD_EXECUTION) || defined (ACPI_CONSTANT_EVAL_ONLY)
@@ -132,7 +142,7 @@
 			case ACPI_TYPE_INTEGER:
 
 				obj_desc->integer.value =
-						(acpi_integer) ACPI_STRTOUL (init_val->val, NULL, 10);
+						(acpi_integer) ACPI_STRTOUL (val, NULL, 10);
 				break;
 
 
@@ -141,8 +151,8 @@
 				/*
 				 * Build an object around the static string
 				 */
-				obj_desc->string.length = (u32) ACPI_STRLEN (init_val->val);
-				obj_desc->string.pointer = init_val->val;
+				obj_desc->string.length = (u32) ACPI_STRLEN (val);
+				obj_desc->string.pointer = val;
 				obj_desc->common.flags |= AOPOBJ_STATIC_POINTER;
 				break;
 
@@ -151,7 +161,7 @@
 
 				obj_desc->mutex.node = new_node;
 				obj_desc->mutex.sync_level =
-						 (u16) ACPI_STRTOUL (init_val->val, NULL, 10);
+						 (u16) ACPI_STRTOUL (val, NULL, 10);
 
 				if (ACPI_STRCMP (init_val->name, "_GL_") == 0) {
 					/*
--- linux/include/acpi/acpiosxf.h	2003/02/05 15:15:11	1.1
+++ linux/include/acpi/acpiosxf.h	2003/02/05 15:25:08
@@ -93,6 +93,11 @@
 	struct acpi_pointer             *address);
 
 acpi_status
+acpi_os_predefined_override (
+	const struct acpi_predefined_names *init_val,
+	acpi_string                     *new_val);
+
+acpi_status
 acpi_os_table_override (
 	struct acpi_table_header        *existing_table,
 	struct acpi_table_header        **new_table);





--- linux/drivers/acpi/osl.c	2003/01/31 18:12:52	1.1
+++ linux/drivers/acpi/osl.c	2003/02/05 15:42:43
@@ -198,6 +198,17 @@
 
 	*phys = virt_to_phys(virt);
 
+	return AE_OK;
+}
+
+acpi_status
+acpi_os_predefined_override (const struct acpi_predefined_names *init_val,
+		             acpi_string *new_val)
+{
+	if (!init_val || !new_val)
+		return AE_BAD_PARAMETER;
+
+	*new_val = NULL;
 	return AE_OK;
 }






-- 
Ducrot Bruno

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

[PATCH] [2/2] Allow override acpi_gbl_pre_defined_names (aka change that _OS).

[Ducrot Bruno]

This is the linux part.  It permit to pass a apci_os_name= boot args
in order to override the \_OS.



--- linux/drivers/acpi/osl.c	2003/02/05 15:51:26	1.5
+++ linux/drivers/acpi/osl.c	2003/02/05 18:09:21
@@ -201,6 +201,8 @@
 	return AE_OK;
 }
 
+static char acpi_os_name[200] = ACPI_OS_NAME;
+
 acpi_status
 acpi_os_predefined_override (const struct acpi_predefined_names *init_val,
 		             acpi_string *new_val)
@@ -209,6 +211,9 @@
 		return AE_BAD_PARAMETER;
 
 	*new_val = NULL;
+	if (!memcmp (init_val->name, "_OS_", 4))
+		*new_val = acpi_os_name;
+
 	return AE_OK;
 }
 
@@ -848,3 +853,28 @@
 
 	return AE_OK;
 }
+
+int __init
+acpi_os_name_setup(char *str)
+{
+	char *p = acpi_os_name;
+	int count = 199;
+
+	if (!str || !*str)
+		return 0;
+
+	for (; count-- && str && *str; str++) {
+		if (isalnum(*str) || *str == ' ')
+			*p++ = *str;
+		else if (*str == '\'' || *str == '"')
+			continue;
+		else
+			break;
+	}
+	*p = 0;
+
+	return 1;
+		
+}
+
+__setup("acpi_os_name=", acpi_os_name_setup);




Cheers,
-- 
Ducrot Bruno

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com