Skip to contentSkip to conversation

x86/pci: add pci=skip_isa_align command lines.

PATCH In review Yinghai Lu 5 participants 7 messages Mar 27, 2008

Subsystems (4)
Documentation x86 ACPI PCI
lore
Share & tools
lore

Yinghai Lu

author
#0 lore raw

PATCH 1/1 x86/pci: add pci=skip_isa_align command lines.

[PATCH] x86/pci: add pci=skip_isa_align command lines.

so we don't align the io port start address for pci cards.

also move out dmi check out acpi.c, because it has nothing to do with acpi. it could spare some calling when we have several peer root buses.

Signed-off-by Yinghai Lu <[email protected]>
Documentation/kernel-parameters.txt +2 -0
index 1970479..d25db3b 100644
@@ -1497,6 +1497,8 @@ and is between 256 and 4096 characters. It is defined in the file
 				This is normally done in pci_enable_device(),
 				so this option is a temporary workaround
 				for broken drivers that don't call it.
+		skip_isa_align	[X86] do not align io start addr, so can
+				handle more pci cards
 		firmware	[ARM] Do not re-enumerate the bus but instead
 				just use the configuration from the
 				bootloader. This is currently used on
arch/x86/pci/acpi.c +0 -41

53 lines not on this page · raw message

arch/x86/pci/common.c +47 -0

59 lines not on this page · raw message

arch/x86/pci/init.c +2 -0
index 343c363..49bb1cf 100644
@@ -29,6 +29,8 @@ static __init int pci_access_init(void)
 		printk(KERN_ERR
 		"PCI: Fatal: No config space access function found\n");
 
+	dmi_check_skip_isa_align();
+
 	return 0;
 }
 arch_initcall(pci_access_init);
arch/x86/pci/pci.h +2 -0
index c9fb637..d338935 100644
@@ -39,6 +39,8 @@ enum pci_bf_sort_state {
 	pci_dmi_bf,
 };
 
+extern void __init dmi_check_skip_isa_align(void);
+
 /* pci-i386.c */
 
 extern unsigned int pcibios_max_latency;
#1 lore raw
Quoted 7 lines · Yinghai Lu

[PATCH] x86/pci: add pci=skip_isa_align command lines.

so we don't align the io port start address for pci cards.

also move out dmi check out acpi.c, because it has nothing to do with acpi. it could spare some calling when we have several peer root buses.

i like this feature, and i've applied your patch to x86.git for testing, but i'd like to hear what the ACPI and PCI guys think about this.

Also, we should try as hard as possible to make it a blacklist instead of a whitelist? It would be cool to support more PCI cards/devices on all new(-ish) systems by default and if we didnt have to maintain the DMI whitelist for eternity. (a whitelist will always be incomplete and will lag behind reality)

Ingo

#2 lore raw
Quoted 11 lines · Ingo Molnar
Quoted 7 lines · Yinghai Lu

[PATCH] x86/pci: add pci=skip_isa_align command lines.

so we don't align the io port start address for pci cards.

also move out dmi check out acpi.c, because it has nothing to do with acpi. it could spare some calling when we have several peer root buses.

Quoted 8 lines

i like this feature, and i've applied your patch to x86.git for testing, but i'd like to hear what the ACPI and PCI guys think about this.

Also, we should try as hard as possible to make it a blacklist instead of a whitelist? It would be cool to support more PCI cards/devices on all new(-ish) systems by default and if we didnt have to maintain the DMI whitelist for eternity. (a whitelist will always be incomplete and will lag behind reality)

yeah, blacklist is more good. that limitation is quite old...

YH

#3 lore raw
In reply to Ingo Molnar
Quoted 11 lines · Ingo Molnar
Quoted 7 lines · Yinghai Lu

[PATCH] x86/pci: add pci=skip_isa_align command lines.

so we don't align the io port start address for pci cards.

also move out dmi check out acpi.c, because it has nothing to do with acpi. it could spare some calling when we have several peer root buses.

Quoted 8 lines

i like this feature, and i've applied your patch to x86.git for testing, but i'd like to hear what the ACPI and PCI guys think about this.

Also, we should try as hard as possible to make it a blacklist instead of a whitelist? It would be cool to support more PCI cards/devices on all new(-ish) systems by default and if we didnt have to maintain the DMI whitelist for eternity. (a whitelist will always be incomplete and will lag behind reality)

Ingo, This is a great idea. I was the guy that added the whitelist and ISA alignment avoidance code but have also been concerned about the headache of keeping whitelist current in mainline and Distro releases as new systems are introduced. When I made the change I assumed (appearently incorrectly) that there were way too many existing systems requiring the alignment to even consider the blacklist approach. Do you have any suggestions on how to identify systems to include in the blacklist? ...or would we just boldly make non-alignment the default, provide an empty blacklist, and let breakage identify those systems that need to be blacklisted?

Thanks.

Gary

Signature · 6 lines
Gary Hade
System x Enablement
IBM Linux Technology Center
503-578-4503  IBM T/L: 775-4503
[email protected]
http://www.ibm.com/linux/ltc
#4 lore raw
Quoted 22 lines · Gary Hade
Quoted 11 lines · Ingo Molnar
Quoted 7 lines · Yinghai Lu

[PATCH] x86/pci: add pci=skip_isa_align command lines.

so we don't align the io port start address for pci cards.

also move out dmi check out acpi.c, because it has nothing to do with acpi. it could spare some calling when we have several peer root buses.

Quoted 8 lines

i like this feature, and i've applied your patch to x86.git for testing, but i'd like to hear what the ACPI and PCI guys think about this.

Also, we should try as hard as possible to make it a blacklist instead of a whitelist? It would be cool to support more PCI cards/devices on all new(-ish) systems by default and if we didnt have to maintain the DMI whitelist for eternity. (a whitelist will always be incomplete and will lag behind reality)

Quoted 10 lines

Ingo, This is a great idea. I was the guy that added the whitelist and ISA alignment avoidance code but have also been concerned about the headache of keeping whitelist current in mainline and Distro releases as new systems are introduced. When I made the change I assumed (appearently incorrectly) that there were way too many existing systems requiring the alignment to even consider the blacklist approach. Do you have any suggestions on how to identify systems to include in the blacklist? ...or would we just boldly make non-alignment the default, provide an empty blacklist, and let breakage identify those systems that need to be blacklisted?

at least to use blacklist with x86_64

YH

reviewer
#5 lore raw
Quoted 3 lines · Yinghai Lu

at least to use blacklist with x86_64

No, please. The last thing we need is an arbitrary behaviour difference based on 32 vs 64 bits.

-hpa

reviewer
#6 lore raw
In reply to Yinghai Lu
Quoted 3 lines · Yinghai Lu

at least to use blacklist with x86_64

I propose that we should use a rule like:

if ( DMI_date >= cutoff && !(CONFIG_ISA_SLOTS && ISA slots in DMI) )

-hpa