RE: [patch 5/5] PNPACPI: clean blacklist

RE: [patch 5/5] PNPACPI: clean blacklist

[Li, Shaohua]

Hi matthieu,
>
>From: matthieu castet <castet.matthieu-GANU6spQydw@public.gmane.org>
>
>Clean the blacklist.  Battery, Button, Fan don't have a CRS and should
be
>removed.  PCI root, PIC, Timer are in pnpbios and are harmless.
>
>Cc: "Brown, Len" <len.brown-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
>Cc: Adam Belay <ambx1-IBH0VoN/3vPQT0dZR+AlfA@public.gmane.org>
>Signed-off-by: Andrew Morton <akpm-3NddpPZAyC0@public.gmane.org>
>---
>
> drivers/pnp/pnpacpi/core.c |   12 ++++++------
> 1 files changed, 6 insertions(+), 6 deletions(-)
>
>diff -puN drivers/pnp/pnpacpi/core.c~pnpacpi-clean-blacklist
>drivers/pnp/pnpacpi/core.c
>--- devel/drivers/pnp/pnpacpi/core.c~pnpacpi-clean-blacklist	2005-09-
>11 03:21:48.000000000 -0700
>+++ devel-akpm/drivers/pnp/pnpacpi/core.c	2005-09-11
03:21:48.000000000 -
>0700
>@@ -27,15 +27,15 @@
>
> static int num = 0;
>
>+/* We need only to blacklist device that have already an acpi driver
that
>+ * can't use pnp layer. We don't need to blacklist device that are
>directly
>+ * used by the kernel (PIC, Timer, ...), as it is harmless and there
were
>+ * already present in pnpbios. Finaly only devices that have a CRS
method
>+ * need to be in this list.
>+ */
> static char __initdata excluded_id_list[] =
>-	"PNP0C0A," /* Battery */
>-	"PNP0C0C,PNP0C0E,PNP0C0D," /* Button */
> 	"PNP0C09," /* EC */
>-	"PNP0C0B," /* Fan */
>-	"PNP0A03," /* PCI root */
> 	"PNP0C0F," /* Link device */
>-	"PNP0000," /* PIC */
>-	"PNP0100," /* Timer */
PNPACPI will call an 'acpi_register_gsi' for device's IRQ. If you remove
PIC/TIMER from the list, their irq will be called. It sounds not safe to
me, especially, in the ioapic mode.
It's meaningless to export PCI root to PNP layer to me. Could we remove
it?
Others look good to me.

Thanks,
Shaohua


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf

Re: [patch 5/5] PNPACPI: clean blacklist

[matthieu castet]

[-- Attachment #1: Type: text/plain, Size: 833 bytes --]

Hi Shaohua,

Li, Shaohua wrote:
> Hi matthieu,
> 
>>From: matthieu castet <castet.matthieu-GANU6spQydw@public.gmane.org>

> PNPACPI will call an 'acpi_register_gsi' for device's IRQ. If you remove
> PIC/TIMER from the list, their irq will be called. It sounds not safe to
> me, especially, in the ioapic mode.
Yes, I remembered this yesterday. I started to try to understand how 
work acpi_register_gsi and if it was safe.

I was wondering if we shouln't call it only when a driver register the 
device (to do something like pci_enable), but it will need pnp layer 
modification.

I updated the patch in order to keep PIC and TIMER.

> It's meaningless to export PCI root to PNP layer to me. Could we remove
> it?
I know, but it is already in pnpbios and I believe we should only put 
device that are problematic.

Thanks,

Matthieu

[-- Attachment #2: pnpacpi_clean_blacklist.patch --]
[-- Type: text/x-patch, Size: 1037 bytes --]

Index: linux-2.6.13rc/drivers/pnp/pnpacpi/core.c
===================================================================
--- linux-2.6.13rc.orig/drivers/pnp/pnpacpi/core.c	2005-08-28 19:26:35.000000000 +0200
+++ linux-2.6.13rc/drivers/pnp/pnpacpi/core.c	2005-09-13 11:31:15.000000000 +0200
@@ -27,12 +27,15 @@
 
 static int num = 0;
 
+/* We need only to blacklist devices that have already an acpi driver that
+ * can't use pnp layer. We don't need to blacklist device that are directly
+ * used by the kernel (PCI root, ...), as it is harmless and there were
+ * already present in pnpbios. But there is an exception for devices that
+ * have irqs (PIC, Timer) because we call acpi_register_gsi.
+ * Finaly only devices that have a CRS method need to be in this list.
+ */
 static char __initdata excluded_id_list[] =
-	"PNP0C0A," /* Battery */
-	"PNP0C0C,PNP0C0E,PNP0C0D," /* Button */
 	"PNP0C09," /* EC */
-	"PNP0C0B," /* Fan */
-	"PNP0A03," /* PCI root */
 	"PNP0C0F," /* Link device */
 	"PNP0000," /* PIC */
 	"PNP0100," /* Timer */

[patch 5/5] PNPACPI: clean blacklist

[akpm-3NddpPZAyC0]

From: matthieu castet <castet.matthieu-GANU6spQydw@public.gmane.org>

Clean the blacklist.  Battery, Button, Fan don't have a CRS and should be
removed.  PCI root, PIC, Timer are in pnpbios and are harmless.

Cc: "Brown, Len" <len.brown-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Cc: Adam Belay <ambx1-IBH0VoN/3vPQT0dZR+AlfA@public.gmane.org>
Signed-off-by: Andrew Morton <akpm-3NddpPZAyC0@public.gmane.org>
---

 drivers/pnp/pnpacpi/core.c |   12 ++++++------
 1 files changed, 6 insertions(+), 6 deletions(-)

diff -puN drivers/pnp/pnpacpi/core.c~pnpacpi-clean-blacklist drivers/pnp/pnpacpi/core.c
--- devel/drivers/pnp/pnpacpi/core.c~pnpacpi-clean-blacklist	2005-09-11 03:21:48.000000000 -0700
+++ devel-akpm/drivers/pnp/pnpacpi/core.c	2005-09-11 03:21:48.000000000 -0700
@@ -27,15 +27,15 @@
 
 static int num = 0;
 
+/* We need only to blacklist device that have already an acpi driver that
+ * can't use pnp layer. We don't need to blacklist device that are directly
+ * used by the kernel (PIC, Timer, ...), as it is harmless and there were
+ * already present in pnpbios. Finaly only devices that have a CRS method
+ * need to be in this list.
+ */
 static char __initdata excluded_id_list[] =
-	"PNP0C0A," /* Battery */
-	"PNP0C0C,PNP0C0E,PNP0C0D," /* Button */
 	"PNP0C09," /* EC */
-	"PNP0C0B," /* Fan */
-	"PNP0A03," /* PCI root */
 	"PNP0C0F," /* Link device */
-	"PNP0000," /* PIC */
-	"PNP0100," /* Timer */
 	;
 static inline int is_exclusive_device(struct acpi_device *dev)
 {
_


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf