ISA -> ISA_ (Re: [GIT PATCH] ACPI patches for 2.6.25-rc6)

ISA -> ISA_ (Re: [GIT PATCH] ACPI patches for 2.6.25-rc6)

[Len Brown]

> btw., saw this ACPI-ish commit in rc7:
> 
> | commit b97d4803400a4442b0e4ae14d0bd8e83994b9004
> | Author: Bjorn Helgaas <bjorn.helgaas@hp.com>
> | Date:   Tue Mar 25 11:21:11 2008 -0600
> |
> |    ACPI: fix Medion _PRT quirk (use "ISA_", not "ISA")
> 
> a grep shows that we've got one other case of an "ISA" string in the 
> ACPI code - is the patch below needed too?
> 
> 	Ingo
> 
> -------------------->
> Subject: acpi: fix asus isa
> From: Ingo Molnar <mingo@elte.hu>
> Date: Wed Mar 26 09:05:07 CET 2008
> 
> Signed-off-by: Ingo Molnar <mingo@elte.hu>
> ---
>  drivers/acpi/asus_acpi.c |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> Index: linux-x86.q/drivers/acpi/asus_acpi.c
> ===================================================================
> --- linux-x86.q.orig/drivers/acpi/asus_acpi.c
> +++ linux-x86.q/drivers/acpi/asus_acpi.c
> @@ -149,7 +149,7 @@ struct asus_hotk {
>  };
>  
>  /* Here we go */
> -#define A1x_PREFIX "\\_SB.PCI0.ISA.EC0."
> +#define A1x_PREFIX "\\_SB.PCI0.ISA_.EC0."

>From an AML point of view, ISA is just as legal as "ISA_"
The NameString paramter to Device() can be 1 to 4 characters.

So if this is correct or not depends on what the BIOS programmer
for the Asus A1x choose to write.  Perhaps Corentin 'Iksaif' CHARY
can confirm if the driver is working properly on that system.

-Len

>  #define L3C_PREFIX "\\_SB.PCI0.PX40.ECD0."
>  #define M1A_PREFIX "\\_SB.PCI0.PX40.EC0."
>  #define P30_PREFIX "\\_SB.PCI0.LPCB.EC0."
> 

Re: ISA -> ISA_ (Re: [GIT PATCH] ACPI patches for 2.6.25-rc6)

[Corentin CHARY]

On Wednesday 26 March 2008 16:32:20 Len Brown wrote:
> So if this is correct or not depends on what the BIOS programmer
> for the Asus A1x choose to write.  Perhaps Corentin 'Iksaif' CHARY
> can confirm if the driver is working properly on that system.

Hi,
Sorry I don't have an A1x to test that, and this was here a long time before I start working on acpi4asus.
But, see that grep :

chary_c@foo /usr/src/linux/drivers/misc $ grep ISA *
asus-laptop.c:      "\\_SB.PCI0.ISA.EC0._Q10",  /* A1x */
thinkpad_acpi.c:IBM_HANDLE(ec, root, "\\_SB.PCI0.ISA.EC0",      /* 240, 240x */
thinkpad_acpi.c:           "\\_SB.PCI.ISA.EC",  /* 570 */
thinkpad_acpi.c:           "\\_SB.PCI0.ISA0.EC0",       /* 600e/x, 770e, 770x */
thinkpad_acpi.c:           "\\_SB.PCI0.ISA.EC", /* A21e, A2xm/p, T20-22, X20-21 */
thinkpad_acpi.c:           "\\_SB.PCI.ISA.SLCE",        /* 570 */

So it seems to work for thinkpads.
I think this patch isn't needed.

-- 
Corentin 'Iksaif' CHARY
http://xf.iksaif.net
--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: ISA -> ISA_ (Re: [GIT PATCH] ACPI patches for 2.6.25-rc6)

[Bjorn Helgaas]

On Wednesday 26 March 2008 09:32:20 am Len Brown wrote:
> >  /* Here we go */
> > -#define A1x_PREFIX "\\_SB.PCI0.ISA.EC0."
> > +#define A1x_PREFIX "\\_SB.PCI0.ISA_.EC0."
> 
> From an AML point of view, ISA is just as legal as "ISA_"
> The NameString paramter to Device() can be 1 to 4 characters.
> 
> So if this is correct or not depends on what the BIOS programmer
> for the Asus A1x choose to write.  Perhaps Corentin 'Iksaif' CHARY
> can confirm if the driver is working properly on that system.

I'm curious about how this works.  I disassembled the DSDT from
http://bugzilla.kernel.org/show_bug.cgi?id=4773 (I attached the disassembly
at http://bugzilla.kernel.org/attachment.cgi?id=15448&action=view), and
the _PRT contains "_SB" and "ISA" (no trailing underscores):

                Package (0x04)
                {
                    0x000DFFFF, 
                    0x00, 
                    \_SB.PCI0.ISA.LNKA, 
                    0x00
                }

But by the time we get to acpi_pci_irq_add_entry(), we've added the
underscores somewhere (see
http://bugzilla.kernel.org/attachment.cgi?id=15423&action=view):

  0000:00:0d[A] -> \_SB_.PCI0.ISA_.LNKA

I don't know where this happens, but it certainly confused me, and
it seems like it could lead to other bugs.

Bjorn

Re: ISA -> ISA_ (Re: [GIT PATCH] ACPI patches for 2.6.25-rc6)

[Henrique de Moraes Holschuh]

On Wed, 26 Mar 2008, Corentin CHARY wrote:
> thinkpad_acpi.c:IBM_HANDLE(ec, root, "\\_SB.PCI0.ISA.EC0",      /* 240, 240x */
> thinkpad_acpi.c:           "\\_SB.PCI.ISA.EC",  /* 570 */
> thinkpad_acpi.c:           "\\_SB.PCI0.ISA0.EC0",       /* 600e/x, 770e, 770x */
> thinkpad_acpi.c:           "\\_SB.PCI0.ISA.EC", /* A21e, A2xm/p, T20-22, X20-21 */
> thinkpad_acpi.c:           "\\_SB.PCI.ISA.SLCE",        /* 570 */
> 
> So it seems to work for thinkpads.

ThinkPads use "ISA" as the node (not ISA_ or anything else like that),
and the ACPICA node match routine works properly.

So please don't touch thinkpad-acpi's ".ISA." stuff.

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh

Re: ISA -> ISA_ (Re: [GIT PATCH] ACPI patches for 2.6.25-rc6)

[Alexey Starikovskiy]

Bjorn Helgaas wrote:
> I'm curious about how this works.  I disassembled the DSDT from
> http://bugzilla.kernel.org/show_bug.cgi?id=4773 (I attached the disassembly
> at http://bugzilla.kernel.org/attachment.cgi?id=15448&action=view), and
> the _PRT contains "_SB" and "ISA" (no trailing underscores):
>
>                 Package (0x04)
>                 {
>                     0x000DFFFF, 
>                     0x00, 
>                     \_SB.PCI0.ISA.LNKA, 
>                     0x00
>                 }
>
> But by the time we get to acpi_pci_irq_add_entry(), we've added the
> underscores somewhere (see
> http://bugzilla.kernel.org/attachment.cgi?id=15423&action=view):
>
>   0000:00:0d[A] -> \_SB_.PCI0.ISA_.LNKA
>
> I don't know where this happens, but it certainly confused me, and
> it seems like it could lead to other bugs.
>   
ACPI is supposed to add trailing underscore for pad all names to 4 bytes.
So ISA will be padded to ISA_.

Regards,
Alex.

Re: ISA -> ISA_ (Re: [GIT PATCH] ACPI patches for 2.6.25-rc6)

[Bjorn Helgaas]

On Wednesday 26 March 2008 11:24:41 am Alexey Starikovskiy wrote:
> Bjorn Helgaas wrote:
> > I'm curious about how this works.  I disassembled the DSDT from
> > http://bugzilla.kernel.org/show_bug.cgi?id=4773 (I attached the disassembly
> > at http://bugzilla.kernel.org/attachment.cgi?id=15448&action=view), and
> > the _PRT contains "_SB" and "ISA" (no trailing underscores):
> >
> >                 Package (0x04)
> >                 {
> >                     0x000DFFFF, 
> >                     0x00, 
> >                     \_SB.PCI0.ISA.LNKA, 
> >                     0x00
> >                 }
> >
> > But by the time we get to acpi_pci_irq_add_entry(), we've added the
> > underscores somewhere (see
> > http://bugzilla.kernel.org/attachment.cgi?id=15423&action=view):
> >
> >   0000:00:0d[A] -> \_SB_.PCI0.ISA_.LNKA
> >
> > I don't know where this happens, but it certainly confused me, and
> > it seems like it could lead to other bugs.
> >   
> ACPI is supposed to add trailing underscore for pad all names to 4 bytes.
> So ISA will be padded to ISA_.

The questions are "where does this happen?" and "why does it not
apply to other things like asus_acpi.c?"

Re: ISA -> ISA_ (Re: [GIT PATCH] ACPI patches for 2.6.25-rc6)

[Alexey Starikovskiy]

Bjorn Helgaas wrote:
> On Wednesday 26 March 2008 11:24:41 am Alexey Starikovskiy wrote:
>   
>> Bjorn Helgaas wrote:
>>     
>>> I'm curious about how this works.  I disassembled the DSDT from
>>> http://bugzilla.kernel.org/show_bug.cgi?id=4773 (I attached the disassembly
>>> at http://bugzilla.kernel.org/attachment.cgi?id=15448&action=view), and
>>> the _PRT contains "_SB" and "ISA" (no trailing underscores):
>>>
>>>                 Package (0x04)
>>>                 {
>>>                     0x000DFFFF, 
>>>                     0x00, 
>>>                     \_SB.PCI0.ISA.LNKA, 
>>>                     0x00
>>>                 }
>>>
>>> But by the time we get to acpi_pci_irq_add_entry(), we've added the
>>> underscores somewhere (see
>>> http://bugzilla.kernel.org/attachment.cgi?id=15423&action=view):
>>>
>>>   0000:00:0d[A] -> \_SB_.PCI0.ISA_.LNKA
>>>
>>> I don't know where this happens, but it certainly confused me, and
>>> it seems like it could lead to other bugs.
>>>   
>>>       
>> ACPI is supposed to add trailing underscore for pad all names to 4 bytes.
>> So ISA will be padded to ISA_.
>>     
>
> The questions are "where does this happen?" and "why does it not
> apply to other things like asus_acpi.c?"
>   
1. drivers/acpi/namespace/nsutils.c:419

Re: ISA -> ISA_ (Re: [GIT PATCH] ACPI patches for 2.6.25-rc6)

[Len Brown]

Bob,
Wouldn't it be less confusing if the dis-assembler
also adding the trailing underscores to match what
is actually done by the interpreter?

thanks,
-Len

On Wednesday 26 March 2008, Moore, Robert wrote:
> References, ACPI specification 3.0b:
> 
> 
> 5.3  ACPI Namespace 
> 
> By convention, when an ASL compiler pads a name shorter than 4
> characters, it is done so with trailing underscores ('_'). See the
> language definition for AML NameSeg in Section 16, "ACPI Source Language
> Reference."
> 
> (Reference should point to section 18, AML reference)
> 
>  
> 
> 18.2.2   Name Objects Encoding
> 
> NameSeg            := <LeadNameChar NameChar NameChar NameChar>
> // Notice that NameSegs shorter than 4 characters are filled with
> // trailing underscores ('_'s).
> 
> 
> 
> 
> 
> -----Original Message-----
> From: Moore, Robert 
> Sent: Wednesday, March 26, 2008 6:59 PM
> To: 'Len Brown'
> Cc: 'aystarik@gmail.com'; Lin, Ming M
> Subject: RE: Re: ISA -> ISA_ (Re: [GIT PATCH] ACPI patches for
> 2.6.25-rc6)
> 
> Whenever an ACPI namepath is parsed (each 4 character name within a
> namepath is called a "nameseg"), each nameseg is padded out with
> underscores if it is shorter than 4 characters. This happens whenever a
> named object is created and whenever a reference to a name is
> encountered.
> 
> Thus, internally, ALL namesegs are exactly 4 characters, no need to add
> code to handle variable-length namesegs.
> 
> Thus, smaller code and fewer bugs. This is also mentioned in the ACPI
> spec somewhere.
> 
> Bob
> 
> 
> Example:
> 
> Name (I, 0x1234)
> 
> Will be entered into the namespace as
> 
> I___
> 
> A reference to I such as
> 
> Store (\DEV0\I, Local0)
> 
> Will convert "I" to "I___" during AML parsing.
> 
> 
> 
> -----Original Message-----
> From: Len Brown [mailto:lenb@kernel.org] 
> Sent: Wednesday, March 26, 2008 11:00 AM
> To: Moore, Robert
> Subject: Fwd: Re: ISA -> ISA_ (Re: [GIT PATCH] ACPI patches for
> 2.6.25-rc6)
> 
> Bob,
> can you comment on this thread?
> 
> thanks,
> -Len
> 
> ----------  Forwarded Message  ----------
> 
> Subject: Re: ISA -> ISA_ (Re: [GIT PATCH] ACPI patches for 2.6.25-rc6)
> Date: Wednesday 26 March 2008
> From: Alexey Starikovskiy < aystarik@gmail.com >
> To: Bjorn Helgaas <bjorn.helgaas@hp.com>
> 
> Bjorn Helgaas wrote:
> > I'm curious about how this works.  I disassembled the DSDT from
> > http://bugzilla.kernel.org/show_bug.cgi?id=4773 (I attached the
> disassembly
> > at http://bugzilla.kernel.org/attachment.cgi?id=15448&action=view),
> and
> > the _PRT contains "_SB" and "ISA" (no trailing underscores):
> >
> >                 Package (0x04)
> >                 {
> >                     0x000DFFFF, 
> >                     0x00, 
> >                     \_SB.PCI0.ISA.LNKA, 
> >                     0x00
> >                 }
> >
> > But by the time we get to acpi_pci_irq_add_entry(), we've added the
> > underscores somewhere (see
> > http://bugzilla.kernel.org/attachment.cgi?id=15423&action=view):
> >
> >   0000:00:0d[A] -> \_SB_.PCI0.ISA_.LNKA
> >
> > I don't know where this happens, but it certainly confused me, and
> > it seems like it could lead to other bugs.
> >   
> ACPI is supposed to add trailing underscore for pad all names to 4
> bytes.
> So ISA will be padded to ISA_.
> 
> Regards,
> Alex.
> 
> -------------------------------------------------------
> 

RE: ISA -> ISA_ (Re: [GIT PATCH] ACPI patches for 2.6.25-rc6)

[Moore, Robert]

The disassembler dumps out the actual code -- including namesegs and
namepaths -- as they appear in the AML. I don't want the disassembler
changing or embellishing anything, I want to see the original ASL.

Bob


>-----Original Message-----
>From: Len Brown [mailto:lenb@kernel.org]
>Sent: Wednesday, March 26, 2008 10:29 PM
>To: Moore, Robert; linux-acpi@vger.kernel.org
>Cc: aystarik@gmail.com; Lin, Ming M
>Subject: Re: ISA -> ISA_ (Re: [GIT PATCH] ACPI patches for 2.6.25-rc6)
>
>Bob,
>Wouldn't it be less confusing if the dis-assembler
>also adding the trailing underscores to match what
>is actually done by the interpreter?
>
>thanks,
>-Len
>
>On Wednesday 26 March 2008, Moore, Robert wrote:
>> References, ACPI specification 3.0b:
>>
>>
>> 5.3  ACPI Namespace
>>
>> By convention, when an ASL compiler pads a name shorter than 4
>> characters, it is done so with trailing underscores ('_'). See the
>> language definition for AML NameSeg in Section 16, "ACPI Source
Language
>> Reference."
>>
>> (Reference should point to section 18, AML reference)
>>
>>
>>
>> 18.2.2   Name Objects Encoding
>>
>> NameSeg            := <LeadNameChar NameChar NameChar NameChar>
>> // Notice that NameSegs shorter than 4 characters are filled with
>> // trailing underscores ('_'s).
>>
>>
>>
>>
>>
>> -----Original Message-----
>> From: Moore, Robert
>> Sent: Wednesday, March 26, 2008 6:59 PM
>> To: 'Len Brown'
>> Cc: 'aystarik@gmail.com'; Lin, Ming M
>> Subject: RE: Re: ISA -> ISA_ (Re: [GIT PATCH] ACPI patches for
>> 2.6.25-rc6)
>>
>> Whenever an ACPI namepath is parsed (each 4 character name within a
>> namepath is called a "nameseg"), each nameseg is padded out with
>> underscores if it is shorter than 4 characters. This happens whenever
a
>> named object is created and whenever a reference to a name is
>> encountered.
>>
>> Thus, internally, ALL namesegs are exactly 4 characters, no need to
add
>> code to handle variable-length namesegs.
>>
>> Thus, smaller code and fewer bugs. This is also mentioned in the ACPI
>> spec somewhere.
>>
>> Bob
>>
>>
>> Example:
>>
>> Name (I, 0x1234)
>>
>> Will be entered into the namespace as
>>
>> I___
>>
>> A reference to I such as
>>
>> Store (\DEV0\I, Local0)
>>
>> Will convert "I" to "I___" during AML parsing.
>>
>>
>>
>> -----Original Message-----
>> From: Len Brown [mailto:lenb@kernel.org]
>> Sent: Wednesday, March 26, 2008 11:00 AM
>> To: Moore, Robert
>> Subject: Fwd: Re: ISA -> ISA_ (Re: [GIT PATCH] ACPI patches for
>> 2.6.25-rc6)
>>
>> Bob,
>> can you comment on this thread?
>>
>> thanks,
>> -Len
>>
>> ----------  Forwarded Message  ----------
>>
>> Subject: Re: ISA -> ISA_ (Re: [GIT PATCH] ACPI patches for
2.6.25-rc6)
>> Date: Wednesday 26 March 2008
>> From: Alexey Starikovskiy < aystarik@gmail.com >
>> To: Bjorn Helgaas <bjorn.helgaas@hp.com>
>>
>> Bjorn Helgaas wrote:
>> > I'm curious about how this works.  I disassembled the DSDT from
>> > http://bugzilla.kernel.org/show_bug.cgi?id=4773 (I attached the
>> disassembly
>> > at http://bugzilla.kernel.org/attachment.cgi?id=15448&action=view),
>> and
>> > the _PRT contains "_SB" and "ISA" (no trailing underscores):
>> >
>> >                 Package (0x04)
>> >                 {
>> >                     0x000DFFFF,
>> >                     0x00,
>> >                     \_SB.PCI0.ISA.LNKA,
>> >                     0x00
>> >                 }
>> >
>> > But by the time we get to acpi_pci_irq_add_entry(), we've added the
>> > underscores somewhere (see
>> > http://bugzilla.kernel.org/attachment.cgi?id=15423&action=view):
>> >
>> >   0000:00:0d[A] -> \_SB_.PCI0.ISA_.LNKA
>> >
>> > I don't know where this happens, but it certainly confused me, and
>> > it seems like it could lead to other bugs.
>> >
>> ACPI is supposed to add trailing underscore for pad all names to 4
>> bytes.
>> So ISA will be padded to ISA_.
>>
>> Regards,
>> Alex.
>>
>> -------------------------------------------------------
>>
>