Skip to contentSkip to conversation

acpiphp - slots management fix

PATCH In review MUNEDA Takahiro 5 participants 13 messages Jan 20, 2006 – Feb 15, 2006

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

cover acpiphp - slots management fix

Hi,

This series of patches contains slots management fix. Here are summary of changes:

o [PATCH 1/7] - removes unnecesary struct members

o [PATCH 2/7] - removes init_slots(), because acpiphp doesn't register slots when driver is loaded.

o [PATCH 3/7] - changes register_slot() to register hotplug slots when they are found

o [PATCH 4/7] - changes init_bridge_misc() to register hotplug slots when they are found

o [PATCH 5/7] - removes cleanup_slots(), because acpiphp doesn't unregister slots when driver is going to be un-loaded.

o [PATCH 6/7] - changes cleanup_brige() to unregister hotplug slots when acpiphp cleanup the bridge.

o [PATCH 7/7] - adds spin_lock/_unlock to protect slot_list

For more specifc information, please see the header of each patch. These patches are against 2.6.16-rc1. I tested them on Tiger4.

Thanks,
MUNE

Signature · 1 line
MUNEDA Takahiro <[email protected]>
#1 lore raw

Hi!

Quoted 2 lines

This series of patches contains slots management fix. Here are summary of changes:

Quite a lot of development happened in -mm in this area... Won't this clash?
Pavel

Signature · 1 line
Thanks, Sharp!
#2 lore raw

Hi Pavel,

At Fri, 20 Jan 2006 12:30:42 +0100,

Quoted 3 lines · Pavel Machek

Quite a lot of development happened in -mm in this area... Won't this clash?

Thanks for your comments.
I have tested these patches on pure 2.6.16-rc1.
I haven't tested these patches applied with the Kristen's. I guess it might be conflict with.

OK, new -mm was released, so I'll rebase the patch against for 2.6.16-rc1-mm2 and Kristen's patches in a few days

Thanks,
MUNE

Signature · 1 line
MUNEDA Takahiro <[email protected]>
#3 lore raw
In reply to MUNEDA Takahiro
Quoted 20 lines · MUNEDA Takahiro

Hi,

This series of patches contains slots management fix. Here are summary of changes:

o [PATCH 1/7] - removes unnecesary struct members

o [PATCH 2/7] - removes init_slots(), because acpiphp doesn't register slots when driver is loaded.

o [PATCH 3/7] - changes register_slot() to register hotplug slots when they are found

o [PATCH 4/7] - changes init_bridge_misc() to register hotplug slots when they are found

o [PATCH 5/7] - removes cleanup_slots(), because acpiphp doesn't unregister slots when driver is going to be un-loaded.

Quoted 8 lines

o [PATCH 6/7] - changes cleanup_brige() to unregister hotplug slots when acpiphp cleanup the bridge.

o [PATCH 7/7] - adds spin_lock/_unlock to protect slot_list

For more specifc information, please see the header of each patch. These patches are against 2.6.16-rc1. I tested them on Tiger4.

This patch series does not even compile when applied:

CC [M] drivers/pci/hotplug/acpiphp_core.o
drivers/pci/hotplug/acpiphp_core.c: In function `acpiphp_register_hotplug_slot':
drivers/pci/hotplug/acpiphp_core.c:355: error: `hotplug_slot' undeclared (first use in this function) drivers/pci/hotplug/acpiphp_core.c:355: error: (Each undeclared identifier is reported only once drivers/pci/hotplug/acpiphp_core.c:355: error: for each function it appears in.) drivers/pci/hotplug/acpiphp_core.c:359: error: `hotplug_slot_info' undeclared (first use in this function) make[2]: *** [drivers/pci/hotplug/acpiphp_core.o] Error 1

Care to redo them?

thanks,

greg k-h

#4 lore raw

Hi,

At Thu, 26 Jan 2006 20:14:25 -0800,

Quoted 41 lines · Greg KH
Quoted 20 lines · MUNEDA Takahiro

Hi,

This series of patches contains slots management fix. Here are summary of changes:

o [PATCH 1/7] - removes unnecesary struct members

o [PATCH 2/7] - removes init_slots(), because acpiphp doesn't register slots when driver is loaded.

o [PATCH 3/7] - changes register_slot() to register hotplug slots when they are found

o [PATCH 4/7] - changes init_bridge_misc() to register hotplug slots when they are found

o [PATCH 5/7] - removes cleanup_slots(), because acpiphp doesn't unregister slots when driver is going to be un-loaded.

Quoted 8 lines

o [PATCH 6/7] - changes cleanup_brige() to unregister hotplug slots when acpiphp cleanup the bridge.

o [PATCH 7/7] - adds spin_lock/_unlock to protect slot_list

For more specifc information, please see the header of each patch. These patches are against 2.6.16-rc1. I tested them on Tiger4.

This patch series does not even compile when applied:

CC [M] drivers/pci/hotplug/acpiphp_core.o
drivers/pci/hotplug/acpiphp_core.c: In function `acpiphp_register_hotplug_slot':
drivers/pci/hotplug/acpiphp_core.c:355: error: `hotplug_slot' undeclared (first use in this function) drivers/pci/hotplug/acpiphp_core.c:355: error: (Each undeclared identifier is reported only once drivers/pci/hotplug/acpiphp_core.c:355: error: for each function it appears in.) drivers/pci/hotplug/acpiphp_core.c:359: error: `hotplug_slot_info' undeclared (first use in this function) make[2]: *** [drivers/pci/hotplug/acpiphp_core.o] Error 1

Quoted 1 line

Care to redo them?

I'm sorry for my late replying.
And thanks for your taking care of my patch.

At first, please let me introduce the base idea of this patch. I'm trying to support p2p bridge hotplug with acpiphp.

Current acpiphp manages hotplug slot by ID. The ID is incremented when acpiphp founds hotpluggable slots.
If the bridges(with hotpluggable slots) are hotplugged many times, the hotpluggable slots are added many times also. The ID might be overflowed. So this patch removes IDs to manage slots.

And, this patch changes the slot register/unregister timing. Current acpiphp registers the slots in the init_slots() called from acpiphp_init(), and unregisters the slots in the cleanup_slots() called from acpiphp_exit().
acpiphp doesn't assume the increase and decrease of the hotpluggable slots.
This patch removes the slot register/unregister processes from the init/exit phases. Instead, adds the these processes in the bridge add/cleanup phases.

Currently, this change doesn't have any meanings. But these changes are needed to support p2p bridge(with hotplug slot) hotplug.

Here is an updated patch. Old patches were not need to be separated, so I merge them into below patch.

This patch is against 2.6.16-rc1-mm3. I tested them on Tiger4 and it works fine.

Thanks,
MUNE

o This patch removes IDs to manage slots.
o This patch removes the slot register/unregister processes from the init/exit phases. Instead, adds these processes in the bridge add/cleanup phases.

 drivers/pci/hotplug/acpiphp.h      |    5 -
 drivers/pci/hotplug/acpiphp_core.c |  118 ++++++++++++++++++-------------------
 drivers/pci/hotplug/acpiphp_glue.c |   48 +++++++--------
 3 files changed, 85 insertions(+), 86 deletions(-)
drivers/pci/hotplug/acpiphp.h +2 -3
Index: linux-2.6.16-rc1-mm3/drivers/pci/hotplug/acpiphp.h
===================================================================
@@ -60,7 +60,6 @@ struct acpiphp_slot;
  * struct slot - slot information for each *physical* slot
  */
 struct slot {
-	u8 number;
 	struct hotplug_slot	*hotplug_slot;
 	struct list_head	slot_list;
 
@@ -121,7 +120,6 @@ struct acpiphp_slot {
 					   objects (i.e. for each function) */
 	struct mutex crit_sect;
 
-	u32		id;		/* slot id (serial #) for hotplug core */
 	u8		device;		/* pci device# */
 
 	u32		sun;		/* ACPI _SUN (slot unique number) */
@@ -206,12 +204,13 @@ struct acpiphp_attention_info
 /* acpiphp_core.c */
 extern int acpiphp_register_attention(struct acpiphp_attention_info*info);
 extern int acpiphp_unregister_attention(struct acpiphp_attention_info *info);
+extern int acpiphp_register_hotplug_slot(struct acpiphp_slot *slot);
+extern void acpiphp_unregister_hotplug_slot(struct acpiphp_slot *slot);
 
 /* acpiphp_glue.c */
 extern int acpiphp_glue_init (void);
 extern void acpiphp_glue_exit (void);
 extern int acpiphp_get_num_slots (void);
-extern struct acpiphp_slot *get_slot_from_id (int id);
 typedef int (*acpiphp_callback)(struct acpiphp_slot *slot, void *data);
 
 extern int acpiphp_enable_slot (struct acpiphp_slot *slot);
drivers/pci/hotplug/acpiphp_core.c +60 -58

160 lines not on this page · raw message

drivers/pci/hotplug/acpiphp_glue.c +23 -25

99 lines not on this page · raw message

#5 lore raw
Quoted 70 lines · MUNEDA Takahiro

Hi,

At Thu, 26 Jan 2006 20:14:25 -0800,

Quoted 41 lines · Greg KH
Quoted 20 lines · MUNEDA Takahiro

Hi,

This series of patches contains slots management fix. Here are summary of changes:

o [PATCH 1/7] - removes unnecesary struct members

o [PATCH 2/7] - removes init_slots(), because acpiphp doesn't register slots when driver is loaded.

o [PATCH 3/7] - changes register_slot() to register hotplug slots when they are found

o [PATCH 4/7] - changes init_bridge_misc() to register hotplug slots when they are found

o [PATCH 5/7] - removes cleanup_slots(), because acpiphp doesn't unregister slots when driver is going to be un-loaded.

Quoted 8 lines

o [PATCH 6/7] - changes cleanup_brige() to unregister hotplug slots when acpiphp cleanup the bridge.

o [PATCH 7/7] - adds spin_lock/_unlock to protect slot_list

For more specifc information, please see the header of each patch. These patches are against 2.6.16-rc1. I tested them on Tiger4.

This patch series does not even compile when applied:

CC [M] drivers/pci/hotplug/acpiphp_core.o
drivers/pci/hotplug/acpiphp_core.c: In function `acpiphp_register_hotplug_slot':
drivers/pci/hotplug/acpiphp_core.c:355: error: `hotplug_slot' undeclared (first use in this function) drivers/pci/hotplug/acpiphp_core.c:355: error: (Each undeclared identifier is reported only once drivers/pci/hotplug/acpiphp_core.c:355: error: for each function it appears in.) drivers/pci/hotplug/acpiphp_core.c:359: error: `hotplug_slot_info' undeclared (first use in this function) make[2]: *** [drivers/pci/hotplug/acpiphp_core.o] Error 1

Quoted 1 line

Care to redo them?

I'm sorry for my late replying.
And thanks for your taking care of my patch.

At first, please let me introduce the base idea of this patch. I'm trying to support p2p bridge hotplug with acpiphp.

Current acpiphp manages hotplug slot by ID. The ID is incremented when acpiphp founds hotpluggable slots.
If the bridges(with hotpluggable slots) are hotplugged many times, the hotpluggable slots are added many times also. The ID might be overflowed. So this patch removes IDs to manage slots.

And, this patch changes the slot register/unregister timing. Current acpiphp registers the slots in the init_slots() called from acpiphp_init(), and unregisters the slots in the cleanup_slots() called from acpiphp_exit().
acpiphp doesn't assume the increase and decrease of the hotpluggable slots.
This patch removes the slot register/unregister processes from the init/exit phases. Instead, adds the these processes in the bridge add/cleanup phases.

Quoted 8 lines

Currently, this change doesn't have any meanings. But these changes are needed to support p2p bridge(with hotplug slot) hotplug.

Here is an updated patch. Old patches were not need to be separated, so I merge them into below patch.

This patch is against 2.6.16-rc1-mm3. I tested them on Tiger4 and it works fine.

Ok, Kristen, have any objections to this patch? I know it's in much the same area that you are working in.

thanks,

greg k-h

#6 lore raw
Quoted 80 lines · Greg KH
Quoted 70 lines · MUNEDA Takahiro

Hi,

At Thu, 26 Jan 2006 20:14:25 -0800,

Quoted 41 lines · Greg KH
Quoted 20 lines · MUNEDA Takahiro

Hi,

This series of patches contains slots management fix. Here are summary of changes:

o [PATCH 1/7] - removes unnecesary struct members

o [PATCH 2/7] - removes init_slots(), because acpiphp doesn't register slots when driver is loaded.

o [PATCH 3/7] - changes register_slot() to register hotplug slots when they are found

o [PATCH 4/7] - changes init_bridge_misc() to register hotplug slots when they are found

o [PATCH 5/7] - removes cleanup_slots(), because acpiphp doesn't unregister slots when driver is going to be un-loaded.

Quoted 8 lines

o [PATCH 6/7] - changes cleanup_brige() to unregister hotplug slots when acpiphp cleanup the bridge.

o [PATCH 7/7] - adds spin_lock/_unlock to protect slot_list

For more specifc information, please see the header of each patch. These patches are against 2.6.16-rc1. I tested them on Tiger4.

This patch series does not even compile when applied:

CC [M] drivers/pci/hotplug/acpiphp_core.o
drivers/pci/hotplug/acpiphp_core.c: In function `acpiphp_register_hotplug_slot':
drivers/pci/hotplug/acpiphp_core.c:355: error: `hotplug_slot' undeclared (first use in this function) drivers/pci/hotplug/acpiphp_core.c:355: error: (Each undeclared identifier is reported only once drivers/pci/hotplug/acpiphp_core.c:355: error: for each function it appears in.) drivers/pci/hotplug/acpiphp_core.c:359: error: `hotplug_slot_info' undeclared (first use in this function) make[2]: *** [drivers/pci/hotplug/acpiphp_core.o] Error 1

Quoted 1 line

Care to redo them?

I'm sorry for my late replying.
And thanks for your taking care of my patch.

At first, please let me introduce the base idea of this patch. I'm trying to support p2p bridge hotplug with acpiphp.

Current acpiphp manages hotplug slot by ID. The ID is incremented when acpiphp founds hotpluggable slots.
If the bridges(with hotpluggable slots) are hotplugged many times, the hotpluggable slots are added many times also. The ID might be overflowed. So this patch removes IDs to manage slots.

And, this patch changes the slot register/unregister timing. Current acpiphp registers the slots in the init_slots() called from acpiphp_init(), and unregisters the slots in the cleanup_slots() called from acpiphp_exit().
acpiphp doesn't assume the increase and decrease of the hotpluggable slots.
This patch removes the slot register/unregister processes from the init/exit phases. Instead, adds the these processes in the bridge add/cleanup phases.

Quoted 8 lines

Currently, this change doesn't have any meanings. But these changes are needed to support p2p bridge(with hotplug slot) hotplug.

Here is an updated patch. Old patches were not need to be separated, so I merge them into below patch.

This patch is against 2.6.16-rc1-mm3. I tested them on Tiger4 and it works fine.

Quoted 7 lines

Ok, Kristen, have any objections to this patch? I know it's in much the same area that you are working in.

thanks,

greg k-h
-

Just reviewing them, they seem like they will be fine to me.

Thanks,
Kristen

#7 lore raw
Quoted 1 line · Kristen Accardi

Just reviewing them, they seem like they will be fine to me.

This conflicts with your other patch you sent me. I'll keep yours, and drop this one, and let you merge his in with yours.

Is that ok?

thanks,

greg k-h

#8 lore raw
Quoted 11 lines · Greg KH
Quoted 1 line · Kristen Accardi

Just reviewing them, they seem like they will be fine to me.

This conflicts with your other patch you sent me. I'll keep yours, and drop this one, and let you merge his in with yours.

Is that ok?

thanks,

greg k-h

I tried to merge this in against 2.6.16-rc2-mm1, which contains the patch you kept, and there are compile issues in his patch still. I think it would be best if Muneda merged these himself.

Kristen

#9 lore raw

Hi Kristen,

At Fri, 10 Feb 2006 11:28:24 -0800,

Quoted 10 lines · Kristen Accardi
Quoted 8 lines · Greg KH
Quoted 1 line · Kristen Accardi

Just reviewing them, they seem like they will be fine to me.

This conflicts with your other patch you sent me. I'll keep yours, and drop this one, and let you merge his in with yours.

Is that ok?

Quoted 7 lines
Quoted 3 lines

thanks,

greg k-h

I tried to merge this in against 2.6.16-rc2-mm1, which contains the patch you kept, and there are compile issues in his patch still. I think it would be best if Muneda merged these himself.

OK, I understood.
However, I can not insmod acpiphp on 2.6.12-rc2-mm1 because of the dock related error(I reported to list a little while ago), I'll merge it after the error is fixed.

Unfortunately, I don't have any laptop with _DCK method. So I can't fix/test the patch.

Thanks,
MUNE

#10 lore raw

PATCH 1/1 acpiphp - slot management fix - V2

Hi,

At Mon, 13 Feb 2006 11:27:28 +0900,

Quoted 9 lines · MUNEDA Takahiro
Quoted 4 lines

I tried to merge this in against 2.6.16-rc2-mm1, which contains the patch you kept, and there are compile issues in his patch still. I think it would be best if Muneda merged these himself.

OK, I understood.
However, I can not insmod acpiphp on 2.6.12-rc2-mm1 because of the dock related error(I reported to list a little while ago), I'll merge it after the error is fixed.

Kenji fixed the _DCK method problem, so I rebased my patch against 2.6.16-rc2-mm1 + kenji's patch.
I tested them on Tiger4 and it works fine.

Thanks,
MUNE

o This patch removes IDs to manage slots.
o This patch removes the slot register/unregister processes from the init/exit phases. Instead, adds these processes in the bridge add/cleanup phases.
o Currently, this change doesn't have any meanings. But these changes are needed to support p2p bridge(with hotplug slot)

 drivers/pci/hotplug/acpiphp.h      |    5 -
 drivers/pci/hotplug/acpiphp_core.c |  118 ++++++++++++++++++-------------------
 drivers/pci/hotplug/acpiphp_glue.c |   48 +++++++--------
 3 files changed, 85 insertions(+), 86 deletions(-)
drivers/pci/hotplug/acpiphp.h +2 -3
Index: linux-2.6.16-rc2-mm1/drivers/pci/hotplug/acpiphp.h
===================================================================
@@ -60,7 +60,6 @@ struct acpiphp_slot;
  * struct slot - slot information for each *physical* slot
  */
 struct slot {
-	u8 number;
 	struct hotplug_slot	*hotplug_slot;
 	struct list_head	slot_list;
 
@@ -121,7 +120,6 @@ struct acpiphp_slot {
 					   objects (i.e. for each function) */
 	struct mutex crit_sect;
 
-	u32		id;		/* slot id (serial #) for hotplug core */
 	u8		device;		/* pci device# */
 
 	u32		sun;		/* ACPI _SUN (slot unique number) */
@@ -229,12 +227,13 @@ struct acpiphp_dock_station {
 /* acpiphp_core.c */
 extern int acpiphp_register_attention(struct acpiphp_attention_info*info);
 extern int acpiphp_unregister_attention(struct acpiphp_attention_info *info);
+extern int acpiphp_register_hotplug_slot(struct acpiphp_slot *slot);
+extern void acpiphp_unregister_hotplug_slot(struct acpiphp_slot *slot);
 
 /* acpiphp_glue.c */
 extern int acpiphp_glue_init (void);
 extern void acpiphp_glue_exit (void);
 extern int acpiphp_get_num_slots (void);
-extern struct acpiphp_slot *get_slot_from_id (int id);
 typedef int (*acpiphp_callback)(struct acpiphp_slot *slot, void *data);
 void handle_hotplug_event_func(acpi_handle, u32, void*);
 
drivers/pci/hotplug/acpiphp_core.c +60 -58

160 lines not on this page · raw message

drivers/pci/hotplug/acpiphp_glue.c +23 -25

99 lines not on this page · raw message

#11 lore raw

Re: [Pcihpd-discuss] [PATCH] acpiphp - slot management fix - V2

Quoted 14 lines · MUNEDA Takahiro
(quoted diff) +6 -1
-static void __exit cleanup_slots (void)
+void acpiphp_unregister_hotplug_slot(struct acpiphp_slot *acpiphp_slot)
 {
 	struct list_head *tmp, *n;
 	struct slot *slot;
+	int retval = 0;
 
 	list_for_each_safe (tmp, n, &slot_list) {
 		/* memory will be freed in release_slot callback */
 		slot = list_entry(tmp, struct slot, slot_list);
+		if (slot->acpi_slot->sun != acpiphp_slot->sun)
+			continue;
+
+		spin_lock(&list_lock);

Review

Quoted 10 lines
(quoted diff) +6 -1
 		list_del(&slot->slot_list);
-		pci_hp_deregister(slot->hotplug_slot);
+		spin_unlock(&list_lock);
+		info ("Slot [%s] unregistered\n", slot->hotplug_slot->name);
+
+		retval = pci_hp_deregister(slot->hotplug_slot);
+		if (retval)
+			err("pci_hp_deregister failed with error %d\n", retval);
 	}
 }

With your patch, I think this list_for_each_safe() loop becomes needless and it should be removed. In addition, slot_list is no longer needed.

Here is a sample patch against your patch. Note that I have not tested it very much.

Thanks,
Kenji Kaneshige

drivers/pci/hotplug/acpiphp.h | 3 +--
drivers/pci/hotplug/acpiphp_core.c | 32 ++++++++------------------------

2 files changed, 9 insertions(+), 26 deletions(-)
drivers/pci/hotplug/acpiphp.h +0 -0
Index: linux-2.6.16-rc2-mm1/drivers/pci/hotplug/acpiphp.h
===================================================================
@@ -61,8 +61,6 @@
 */
struct slot {
	struct hotplug_slot	*hotplug_slot;
-	struct list_head	slot_list;
-
	struct acpiphp_slot	*acpi_slot;
};

@@ -118,6 +116,7 @@
struct acpiphp_bridge *bridge; /* parent */
struct list_head funcs; /* one slot may have different
objects (i.e. for each function) */

+	struct slot *slot;
	struct mutex crit_sect;

	u8		device;		/* pci device# */
drivers/pci/hotplug/acpiphp_core.c +0 -0
Index: linux-2.6.16-rc2-mm1/drivers/pci/hotplug/acpiphp_core.c
===================================================================
@@ -44,9 +44,6 @@

#include "pci_hotplug.h"
#include "acpiphp.h"

-static LIST_HEAD(slot_list);
-static DEFINE_SPINLOCK(list_lock);
-
#define MY_NAME "acpiphp"

static int debug;
@@ -379,6 +376,8 @@

	slot->hotplug_slot->info->max_bus_speed = PCI_SPEED_UNKNOWN;
	slot->hotplug_slot->info->cur_bus_speed = PCI_SPEED_UNKNOWN;

+ acpiphp_slot->slot = slot;
+

	make_slot_name(slot);

	retval = pci_hp_register(slot->hotplug_slot);
@@ -387,10 +386,6 @@
		goto error_name;

}

- /* add slot to our internal list */
- spin_lock(&list_lock);
- list_add(&slot->slot_list, &slot_list);

-	spin_unlock(&list_lock);
	info("Slot [%s] registered\n", slot->hotplug_slot->name);

	return 0;

@@ -409,25 +404,14 @@

void acpiphp_unregister_hotplug_slot(struct acpiphp_slot *acpiphp_slot) {
- struct list_head *tmp, *n;
- struct slot *slot;

+	struct slot *slot = acpiphp_slot->slot;
	int retval = 0;

- list_for_each_safe (tmp, n, &slot_list) {
- /* memory will be freed in release_slot callback */
- slot = list_entry(tmp, struct slot, slot_list);
- if (slot->acpi_slot->sun != acpiphp_slot->sun)
- continue;
-
- spin_lock(&list_lock);
- list_del(&slot->slot_list);
- spin_unlock(&list_lock);
- info ("Slot [%s] unregistered\n", slot->hotplug_slot->name);
-
- retval = pci_hp_deregister(slot->hotplug_slot);
- if (retval)
- err("pci_hp_deregister failed with error %d\n", retval);
- }
+ info ("Slot [%s] unregistered\n", slot->hotplug_slot->name);
+
+ retval = pci_hp_deregister(slot->hotplug_slot);
+ if (retval)
+ err("pci_hp_deregister failed with error %d\n", retval);
}

#12 lore raw

PATCH 1/1 acpiphp - slot management fix - V3

Hi,

At Tue, 14 Feb 2006 18:17:20 +0900,

Quoted 7 lines · Kenji Kaneshige

With your patch, I think this list_for_each_safe() loop becomes needless and it should be removed. In addition, slot_list is no longer needed.

Here is a sample patch against your patch. Note that I have not tested it very much.

Thank you for your comments. Your proposal looks good to me. So I've merged your patch.
This patch is against for 2.6.16-rc3-mm1.
I tested it on my Tiger4 and it works well.

Thanks,
MUNE

o This patch removes IDs (for slots management).
o This patch removes the slot register/unregister processes from the init/exit phases. Instead, adds these processes in the bridge add/cleanup phases.
o Currently, this change doesn't have any meanings. But these changes are needed to support p2p bridge(with hotplug slot)

 drivers/pci/hotplug/acpiphp.h      |    8 --
 drivers/pci/hotplug/acpiphp_core.c |  119 ++++++++++++++++---------------------
 drivers/pci/hotplug/acpiphp_glue.c |   48 +++++++-------
 3 files changed, 78 insertions(+), 97 deletions(-)
drivers/pci/hotplug/acpiphp.h +3 -5
Index: linux-2.6.16-rc3-mm1/drivers/pci/hotplug/acpiphp.h
===================================================================
@@ -60,10 +60,7 @@ struct acpiphp_slot;
  * struct slot - slot information for each *physical* slot
  */
 struct slot {
-	u8 number;
 	struct hotplug_slot	*hotplug_slot;
-	struct list_head	slot_list;
-
 	struct acpiphp_slot	*acpi_slot;
 };
 
@@ -119,9 +116,9 @@ struct acpiphp_slot {
 	struct acpiphp_bridge *bridge;	/* parent */
 	struct list_head funcs;		/* one slot may have different
 					   objects (i.e. for each function) */
+	struct slot *slot;
 	struct mutex crit_sect;
 
-	u32		id;		/* slot id (serial #) for hotplug core */
 	u8		device;		/* pci device# */
 
 	u32		sun;		/* ACPI _SUN (slot unique number) */
@@ -229,12 +226,13 @@ struct acpiphp_dock_station {
 /* acpiphp_core.c */
 extern int acpiphp_register_attention(struct acpiphp_attention_info*info);
 extern int acpiphp_unregister_attention(struct acpiphp_attention_info *info);
+extern int acpiphp_register_hotplug_slot(struct acpiphp_slot *slot);
+extern void acpiphp_unregister_hotplug_slot(struct acpiphp_slot *slot);
 
 /* acpiphp_glue.c */
 extern int acpiphp_glue_init (void);
 extern void acpiphp_glue_exit (void);
 extern int acpiphp_get_num_slots (void);
-extern struct acpiphp_slot *get_slot_from_id (int id);
 typedef int (*acpiphp_callback)(struct acpiphp_slot *slot, void *data);
 void handle_hotplug_event_func(acpi_handle, u32, void*);
 
drivers/pci/hotplug/acpiphp_core.c +52 -67

155 lines not on this page · raw message

drivers/pci/hotplug/acpiphp_glue.c +23 -25

99 lines not on this page · raw message