[PATCH] fix for 2.4.21-pre2 and acpi-20021212

[PATCH] fix for 2.4.21-pre2 and acpi-20021212

[Jörg Prante]

[-- Attachment #1: Type: text/plain, Size: 416 bytes --]

Hi,

this patch is needed for acpi-20021212 with 2.4.21-pre2 on a Pentium 4 with 
ACPI IO-APIC. The dynamic MP busses and IRQ_SOURCES in 2.4.21-pre2 introduced 
a conflict that will rise NULL pointer dereferences when combined with the 
ACPI patch. I don't have the hardware to test it, but Margit Schubert-While 
reported a successful boot.

http://infolinux.de/jp/087_acpi-mpparse-fix

Cheers, 

Jörg

[-- Attachment #2: 087_acpi-mpparse-fix --]
[-- Type: text/x-diff, Size: 2842 bytes --]

--- linux-2.4.21-pre2-jp15/arch/i386/kernel/mpparse.c.orig	2003-01-02 21:13:16.000000000 +0100
+++ linux-2.4.21-pre2-jp15/arch/i386/kernel/mpparse.c	2003-01-03 18:14:10.000000000 +0100
@@ -1132,10 +1132,28 @@
 
 void __init mp_config_acpi_legacy_irqs (void)
 {
-	struct mpc_config_intsrc intsrc;
 	int			i = 0;
 	int			ioapic = -1;
 
+	/*
+	 * Initialize mp_irqs for IRQ configuration.
+	 */
+	unsigned char *bus_data;
+	int count;
+
+	count = (MAX_MP_BUSSES * sizeof(int)) * 4;
+	count += (MAX_IRQ_SOURCES * sizeof(int)) * 4;
+	bus_data = alloc_bootmem(count);
+	if (!bus_data) {
+		panic("Fatal: can't allocate bus memory for ACPI legacy IRQ!");
+	}
+	mp_bus_id_to_type = (int *)&bus_data[0];
+	mp_bus_id_to_node = (int *)&bus_data[(MAX_MP_BUSSES * sizeof(int))];
+	mp_bus_id_to_local = (int *)&bus_data[(MAX_MP_BUSSES * sizeof(int)) * 2];
+	mp_bus_id_to_pci_bus = (int *)&bus_data[(MAX_MP_BUSSES * sizeof(int)) * 3];
+	mp_irqs = (struct mpc_config_intsrc *)&bus_data[(MAX_MP_BUSSES * sizeof(int)) * 4];
+	memset(mp_bus_id_to_pci_bus, -1, MAX_MP_BUSSES);
+
 	/* 
 	 * Fabricate the legacy ISA bus (bus #31).
 	 */
@@ -1149,11 +1167,6 @@
 	if (ioapic < 0)
 		return;
 
-	intsrc.mpc_type = MP_INTSRC;
-	intsrc.mpc_irqflag = 0;					/* Conforming */
-	intsrc.mpc_srcbus = MP_ISA_BUS;
-	intsrc.mpc_dstapic = mp_ioapics[ioapic].mpc_apicid;
-
 	/* 
 	 * Use the default configuration for the IRQs 0-15.  These may be
 	 * overriden by (MADT) interrupt source override entries.
@@ -1162,17 +1175,24 @@
 
 		if (i == 2) continue;			/* Don't connect IRQ2 */
 
-		intsrc.mpc_irqtype = i ? mp_INT : mp_ExtINT;   /* 8259A to #0 */
-		intsrc.mpc_srcbusirq = i;		   /* Identity mapped */
-		intsrc.mpc_dstirq = i;
+		mp_irqs[mp_irq_entries].mpc_type = MP_INTSRC;
+		mp_irqs[mp_irq_entries].mpc_irqflag = 0;	/* Conforming */
+		mp_irqs[mp_irq_entries].mpc_srcbus = MP_ISA_BUS;
+		mp_irqs[mp_irq_entries].mpc_dstapic = mp_ioapics[ioapic].mpc_apicid;
+		mp_irqs[mp_irq_entries].mpc_irqtype = i ? mp_INT : mp_ExtINT;   /* 8259A to #0 */
+		mp_irqs[mp_irq_entries].mpc_srcbusirq = i;	   /* Identity mapped */
+		mp_irqs[mp_irq_entries].mpc_dstirq = i;
 
 		Dprintk("Int: type %d, pol %d, trig %d, bus %d, irq %d, "
-			"%d-%d\n", intsrc.mpc_irqtype, intsrc.mpc_irqflag & 3, 
-			(intsrc.mpc_irqflag >> 2) & 3, intsrc.mpc_srcbus, 
-			intsrc.mpc_srcbusirq, intsrc.mpc_dstapic, 
-			intsrc.mpc_dstirq);
+			"%d-%d\n", 
+			mp_irqs[mp_irq_entries].mpc_irqtype, 
+			mp_irqs[mp_irq_entries].mpc_irqflag & 3, 
+			(mp_irqs[mp_irq_entries].mpc_irqflag >> 2) & 3, 
+			mp_irqs[mp_irq_entries].mpc_srcbus, 
+			mp_irqs[mp_irq_entries].mpc_srcbusirq, 
+			mp_irqs[mp_irq_entries].mpc_dstapic, 
+			mp_irqs[mp_irq_entries].mpc_dstirq);
 
-		mp_irqs[mp_irq_entries] = intsrc;
 		if (++mp_irq_entries == MAX_IRQ_SOURCES)
 			panic("Max # of irq sources exceeded!\n");
 	}

RE: [PATCH] fix for 2.4.21-pre2 and acpi-20021212

[Grover, Andrew]

> From: Jörg Prante [mailto:joergprante-UIqTAjE4tzc5WgrkcBd8vg@public.gmane.org] 
> this patch is needed for acpi-20021212 with 2.4.21-pre2 on a 
> Pentium 4 with 
> ACPI IO-APIC. The dynamic MP busses and IRQ_SOURCES in 
> 2.4.21-pre2 introduced 
> a conflict that will rise NULL pointer dereferences when 
> combined with the 
> ACPI patch. I don't have the hardware to test it, but Margit 
> Schubert-While 
> reported a successful boot.
> 
> http://infolinux.de/jp/087_acpi-mpparse-fix

Excellent! Thanks, applied.

Regards -- Andy


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

RE: [PATCH] fix for 2.4.21-pre2 and acpi-20021212

[Margit Schubert-While]

At 11:58 03.01.03 -0800, you wrote:
> > From: Jörg Prante [mailto:joergprante-UIqTAjE4tzc5WgrkcBd8vg@public.gmane.org]
> > this patch is needed for acpi-20021212 with 2.4.21-pre2 on a
> > Pentium 4 with
> > ACPI IO-APIC. The dynamic MP busses and IRQ_SOURCES in
> > 2.4.21-pre2 introduced
> > a conflict that will rise NULL pointer dereferences when
> > combined with the
> > ACPI patch. I don't have the hardware to test it, but Margit
> > Schubert-While
> > reported a successful boot.
> >
> > http://infolinux.de/jp/087_acpi-mpparse-fix
>
>Excellent! Thanks, applied.
>
>Regards -- Andy

Hi Andy,
         Now that I have done my good deed for the day(s) ,
         perhaps you can answer this.  :-)
         M/B is a D845PESVL with latest BIOS.
         Nothing shows up in "fan" and "thermal_zone".
         Looking on the Intel site, however shows a Windows
         tool for getting this info. Hmmm.
         A quick squint at the DSDT shows some suspicious
         "if windows" tests. Hmmm.
         Is the ACPI table only releasing this info for Windows ?
         I really would like to know how my P4 is cooking :-)

         Margit



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [PATCH] fix for 2.4.21-pre2 and acpi-20021212

[Ducrot Bruno]

On Fri, Jan 03, 2003 at 10:42:52PM +0100, Margit Schubert-While wrote:
> At 11:58 03.01.03 -0800, you wrote:
> >> From: Jörg Prante [mailto:joergprante-UIqTAjE4tzc5WgrkcBd8vg@public.gmane.org]
> >> this patch is needed for acpi-20021212 with 2.4.21-pre2 on a
> >> Pentium 4 with
> >> ACPI IO-APIC. The dynamic MP busses and IRQ_SOURCES in
> >> 2.4.21-pre2 introduced
> >> a conflict that will rise NULL pointer dereferences when
> >> combined with the
> >> ACPI patch. I don't have the hardware to test it, but Margit
> >> Schubert-While
> >> reported a successful boot.
> >>
> >> http://infolinux.de/jp/087_acpi-mpparse-fix
> >
> >Excellent! Thanks, applied.
> >
> >Regards -- Andy
> 
> Hi Andy,
>         Now that I have done my good deed for the day(s) ,
>         perhaps you can answer this.  :-)
>         M/B is a D845PESVL with latest BIOS.
>         Nothing shows up in "fan" and "thermal_zone".
>         Looking on the Intel site, however shows a Windows
>         tool for getting this info. Hmmm.
>         A quick squint at the DSDT shows some suspicious
>         "if windows" tests. Hmmm.
>         Is the ACPI table only releasing this info for Windows ?
>         I really would like to know how my P4 is cooking :-)
> 
>         Margit
> 

I guess an EC SMB driver that (still) need to be written for Linux.
That should be for the lm-sensors team, I think.

The only trouble I think is to know where is the (first) EC adress for
that stuff, perhaps in the registry?

-- 
Ducrot Bruno

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


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [PATCH] fix for 2.4.21-pre2 and acpi-20021212

[Pavel Machek]

Hi!

> >         Looking on the Intel site, however shows a Windows
> >         tool for getting this info. Hmmm.
> >         A quick squint at the DSDT shows some suspicious
> >         "if windows" tests. Hmmm.
> >         Is the ACPI table only releasing this info for Windows ?
> >         I really would like to know how my P4 is cooking :-)
> 
> I guess an EC SMB driver that (still) need to be written for Linux.
> That should be for the lm-sensors team, I think.

I guess relying on lm-sensors team is bad idea.
They have enough work, already; and basic lm-sensors
support is in 2.5.54, so you should be able to
develop against Linus' tree.

				Pavel

-- 
				Pavel
Written on sharp zaurus, because my Velo1 broke. If you have Velo you don't need...



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