PCI bugfixes for 2.6.12-rc5
Subsystems (4)
PATCH 1/1 PCI bugfixes for 2.6.12-rc5
Here are 3 patches for the 2.6.12-rc5 tree that fix some PCI and PCI Hotplug bugs, and 1 patch that adds some new device ids for nvidia ide controllers. All of these patches have been in the past few -mm releases.
Please pull from:
rsync://rsync.kernel.org/pub/scm/linux/kernel/git/gregkh/pci-2.6.git/
Full patches will be sent to the linux-kernel and linux-pci mailing lists, if anyone wants to see them.
thanks,
greg k-h
drivers/ide/pci/amd74xx.c | 3 drivers/pci/hotplug/cpci_hotplug_core.c | 302 ++++++++++++++------------------ drivers/pci/hotplug/cpci_hotplug_pci.c | 144 +++++---------- drivers/pci/hotplug/shpchprm_acpi.c | 4 include/linux/pci_ids.h | 6 5 files changed, 203 insertions(+), 256 deletions(-)
-------------
Andy Currid:
o PCI: amd74xx patch for new NVIDIA device IDs
Kenji Kaneshige:
o PCI Hotplug: SHPCHP driver doesn't enable PERR and SERR properly
o PCI Hotplug: shpchp driver doesn't program _HPP values properly
Scott Murray:
o PCI Hotplug: more CPCI updates
PATCH 1/1 PCI Hotplug: SHPCHP driver doesn't enable PERR and SERR properly
[PATCH] PCI Hotplug: SHPCHP driver doesn't enable PERR and SERR properly
Current shpchp driver doesn't seem to program command register to enable PERR and SERR properly. The following patch fixes this issue.
---
commit 2ac2610b26c9da72820443328ff2c56c7b8c87b8
tree 9ca01e869d0c5a958232b7007e3fb48a27b0c5b3
parent 7a8cb869f31de525bc34095f51f8c8a43ffcb6a9
author Kenji Kaneshige <[email protected]> Fri, 27 May 2005 16:08:14 +0900 committer Greg KH <[email protected]> Tue, 31 May 2005 14:26:37 -0700
drivers/pci/hotplug/shpchprm_acpi.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)
drivers/pci/hotplug/shpchprm_acpi.c +1 -1
@@ -1681,7 +1681,7 @@ void shpchprm_enable_card( | PCI_COMMAND_IO | PCI_COMMAND_MEMORY; bcmd = bcommand = bcommand |PCI_BRIDGE_CTL_NO_ISA ; - ab =find_acpi_bridge_by_bus (acpi_bridges_head, ctrl->seg, ctrl->bus); + ab =find_acpi_bridge_by_bus (acpi_bridges_head, ctrl->seg, ctrl->slot_bus);if (ab) {if (ab->_hpp) {if (ab->_hpp->enable_perr) {
PATCH 1/1 PCI Hotplug: shpchp driver doesn't program _HPP values properly
[PATCH] PCI Hotplug: shpchp driver doesn't program _HPP values properly
Current shpchp driver doesn't seem to program _HPP values properly. The following patch fixes this issue.
---
commit 7a8cb869f31de525bc34095f51f8c8a43ffcb6a9
tree 952c03852730f328c2b500acb4748d2c4298d2b3
parent 2e3e80c2b75e3815a0160cbd23d4fdb767d66b35
author Kenji Kaneshige <[email protected]> Mon, 23 May 2005 19:50:32 +0900 committer Greg KH <[email protected]> Tue, 31 May 2005 14:26:37 -0700
drivers/pci/hotplug/shpchprm_acpi.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)
drivers/pci/hotplug/shpchprm_acpi.c +1 -1
@@ -1626,7 +1626,7 @@ int shpchprm_set_hpp( pci_bus->number = func->bus; devfn =PCI_DEVFN (func->device, func->function); - ab =find_acpi_bridge_by_bus (acpi_bridges_head, ctrl->seg, ctrl->bus); + ab =find_acpi_bridge_by_bus (acpi_bridges_head, ctrl->seg, ctrl->slot_bus);if (ab) {if (ab->_hpp) {
PATCH 1/1 PCI: amd74xx patch for new NVIDIA device IDs
[PATCH] PCI: amd74xx patch for new NVIDIA device IDs
Here's the 2.6 amd74xx patch for NVIDIA MCP51.
---
commit af00f9811e0ccbd3db84ddc4cffb0da942653393
tree 5a9c3b7f7d61d96d3624ad130b173a761cb7dac2
parent 2ac2610b26c9da72820443328ff2c56c7b8c87b8
author Andy Currid <[email protected]> Mon, 23 May 2005 08:55:45 -0700 committer Greg KH <[email protected]> Tue, 31 May 2005 14:26:38 -0700
drivers/ide/pci/amd74xx.c | 3 +++ include/linux/pci_ids.h | 6 ++++++ 2 files changed, 9 insertions(+), 0 deletions(-)
drivers/ide/pci/amd74xx.c +3 -0
@@ -72,6 +72,7 @@ static struct amd_ide_chip { { PCI_DEVICE_ID_NVIDIA_NFORCE3S_SATA2, 0x50, AMD_UDMA_133 }, { PCI_DEVICE_ID_NVIDIA_NFORCE_CK804_IDE, 0x50, AMD_UDMA_133 }, { PCI_DEVICE_ID_NVIDIA_NFORCE_MCP04_IDE, 0x50, AMD_UDMA_133 }, + { PCI_DEVICE_ID_NVIDIA_NFORCE_MCP51_IDE, 0x50, AMD_UDMA_133 }, { 0 } };
@@ -487,6 +488,7 @@ static ide_pci_device_t amd74xx_chipsets/* 12 */ DECLARE_NV_DEV ("NFORCE3-250-SATA2" ),/* 13 */ DECLARE_NV_DEV ("NFORCE-CK804" ),/* 14 */ DECLARE_NV_DEV ("NFORCE-MCP04" ), + /* 15 */ DECLARE_NV_DEV("NFORCE-MCP51"), };static int __devinit amd74xx_probe (struct pci_dev *dev,const struct pci_device_id *id)
@@ -521,6 +523,7 @@ static struct pci_device_id amd74xx_pci_#endif { PCI_VENDOR_ID_NVIDIA, PCI_DEVICE_ID_NVIDIA_NFORCE_CK804_IDE, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 13 }, { PCI_VENDOR_ID_NVIDIA, PCI_DEVICE_ID_NVIDIA_NFORCE_MCP04_IDE, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 14 }, + { PCI_VENDOR_ID_NVIDIA, PCI_DEVICE_ID_NVIDIA_NFORCE_MCP51_IDE, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 15 }, { 0, }, };MODULE_DEVICE_TABLE (pci, amd74xx_pci_tbl);
include/linux/pci_ids.h +6 -0
@@ -1230,6 +1230,12 @@#define PCI_DEVICE_ID_NVIDIA_QUADRO4_900XGL 0x0258#define PCI_DEVICE_ID_NVIDIA_QUADRO4_750XGL 0x0259#define PCI_DEVICE_ID_NVIDIA_QUADRO4_700XGL 0x025B +#define PCI_DEVICE_ID_NVIDIA_NFORCE_MCP51_IDE 0x0265 +#define PCI_DEVICE_ID_NVIDIA_NFORCE_MCP51_SATA 0x0266 +#define PCI_DEVICE_ID_NVIDIA_NFORCE_MCP51_SATA2 0x0267 +#define PCI_DEVICE_ID_NVIDIA_NVENET_12 0x0268 +#define PCI_DEVICE_ID_NVIDIA_NVENET_13 0x0269 +#define PCI_DEVICE_ID_NVIDIA_MCP51_AUDIO 0x026B#define PCI_DEVICE_ID_NVIDIA_GEFORCE4_TI_4800 0x0280#define PCI_DEVICE_ID_NVIDIA_GEFORCE4_TI_4800_8X 0x0281#define PCI_DEVICE_ID_NVIDIA_GEFORCE4_TI_4800SE 0x0282
PATCH 1/1 PCI Hotplug: more CPCI updates
[PATCH] PCI Hotplug: more CPCI updates
Here is my third attempt at a patch to further update the CompactPCI hotplug driver infrastructure to address the pci_enable_device issue discussed on the list as well as a few other issues I discovered during some more testing. This version addresses a few more issues pointed out by Prarit Bhargava. Changes include:
- cpci_enable_device and its recursive calling of pci_enable_device on new devices removed.
- Use list_rwsem to avoid slot status change races between disable_slot and check_slots.
- Fixed oopsing in cpci_hp_unregister_bus caused by calling list_del on a slot after calling pci_hp_deregister.
- Removed kfree calls in cleanup_slots since release_slot will have done it already.
- Reworked init_slots a bit to fix latch and adapter file updating on subsequent calls to cpci_hp_start.
- Improved sanity checking in cpci_hp_register_controller.
- Now shut things down correctly in cpci_hotplug_exit.
- Switch to pci_get_slot instead of deprecated pci_find_slot.
- A bunch of CodingStyle fixes.
---
commit bcc488ab02254a6e60d749187a632dc3d642d4f8
tree 4cd45b4d546cf85c14c442b27f611e048a4d9938
parent af00f9811e0ccbd3db84ddc4cffb0da942653393
author Scott Murray <[email protected]> Fri, 27 May 2005 16:48:52 -0400 committer Greg KH <[email protected]> Tue, 31 May 2005 14:26:38 -0700
drivers/pci/hotplug/cpci_hotplug_core.c | 302 ++++++++++++++----------------- drivers/pci/hotplug/cpci_hotplug_pci.c | 144 ++++++--------- 2 files changed, 192 insertions(+), 254 deletions(-)
drivers/pci/hotplug/cpci_hotplug_core.c +138 -164
@@ -1,7 +1,7 @@/* * CompactPCI Hot Plug Driver * -* Copyright (C) 2002 SOMA Networks, Inc. +* Copyright (C) 2002,2005 SOMA Networks, Inc. * Copyright (C) 2001 Greg Kroah-Hartman ([email protected]) * Copyright (C) 2001 IBM Corp. *
@@ -45,10 +45,10 @@ #define dbg(format, arg...) \ do { \ - if(cpci_debug) \ + if (cpci_debug) \ printk (KERN_DEBUG "%s: " format "\n", \ MY_NAME , ## arg); \ - } while(0) + } while (0) #define err(format, arg...) printk(KERN_ERR "%s: " format "\n", MY_NAME , ## arg) #define info(format, arg...) printk(KERN_INFO "%s: " format "\n", MY_NAME , ## arg) #define warn(format, arg...) printk(KERN_WARNING "%s: " format "\n", MY_NAME , ## arg)
@@ -111,10 +111,8 @@ enable_slot(struct hotplug_slot *hotplugdbg ("%s - physical_slot = %s" , __FUNCTION__, hotplug_slot->name); -if (controller->ops->set_power) { +if (controller->ops->set_power) retval = controller->ops->set_power (slot,1 ); - } -return retval; }
@@ -126,37 +124,41 @@ disable_slot(struct hotplug_slot *hotpludbg ("%s - physical_slot = %s" , __FUNCTION__, hotplug_slot->name); +down_write (&list_rwsem); +/* Unconfigure device */ dbg ("%s - unconfiguring slot %s" , __FUNCTION__, slot->hotplug_slot->name); -if ((retval =cpci_unconfigure_slot (slot))) { +if ((retval =cpci_unconfigure_slot (slot))) {err ("%s - could not unconfigure slot %s" , __FUNCTION__, slot->hotplug_slot->name); -return retval; + goto disable_error; }dbg ("%s - finished unconfiguring slot %s" , __FUNCTION__, slot->hotplug_slot->name);/* Clear EXT (by setting it) */ -if (cpci_clear_ext (slot)) { +if (cpci_clear_ext (slot)) {err ("%s - could not clear EXT for slot %s" , __FUNCTION__, slot->hotplug_slot->name); retval = -ENODEV ; + goto disable_error; }cpci_led_on (slot); -if (controller->ops->set_power) { - retval = controller->ops->set_power (slot,0 ); - } +if (controller->ops->set_power) +if ((retval = controller->ops->set_power (slot,0 ))) + goto disable_error; -if (update_adapter_status (slot->hotplug_slot,0 )) { +if (update_adapter_status (slot->hotplug_slot,0 ))warn ("failure to update adapter file" ); - } -if (slot->extracting) { +if (slot->extracting) { slot->extracting =0 ;atomic_dec (&extracting); } +disable_error: +up_write (&list_rwsem);return retval; }
@@ -165,9 +167,8 @@ cpci_get_power_status(struct slot *slot) {u8 power =1 ; -if (controller->ops->get_power) { +if (controller->ops->get_power) power = controller->ops->get_power (slot); - }return power; }
@@ -237,9 +238,8 @@ cpci_hp_register_bus(struct pci_bus *busint status = -ENOMEM ;int i; -if (!(controller && bus)) { +if (!(controller && bus))return -ENODEV ; - }/* * Create a structure for each slot, and register that slot
@@ -316,32 +316,30 @@ int cpci_hp_unregister_bus(struct pci_bus *bus) {struct slot *slot; -struct list_head *tmp; -struct list_head *next; -int status; +struct slot *tmp; +int status =0 ;down_write (&list_rwsem); -if (!slots) { +if (!slots) {up_write (&list_rwsem);return -1 ; } -list_for_each_safe (tmp, next, &slot_list) { - slot = list_entry(tmp, struct slot, slot_list); -if (slot->bus == bus) { +list_for_each_entry_safe (slot, tmp, &slot_list, slot_list) { +if (slot->bus == bus) { +list_del (&slot->slot_list); + slots--; +dbg ("deregistering slot %s" , slot->hotplug_slot->name); status =pci_hp_deregister (slot->hotplug_slot); -if (status) { +if (status) {err ("pci_hp_deregister failed with error %d" , status); -return status; +break ; } - -list_del (&slot->slot_list); - slots--; } }up_write (&list_rwsem); -return 0 ; +return status; }/* This is the interrupt mode interrupt handler */
@@ -351,7 +349,7 @@ cpci_hp_intr(int irq, void *data, structdbg ("entered cpci_hp_intr" );/* Check to see if it was our interrupt */ -if ((controller->irq_flags &SA_SHIRQ ) && +if ((controller->irq_flags &SA_SHIRQ ) && !controller->ops->check_irq (controller->dev_id)) {dbg ("exited cpci_hp_intr, not our interrupt" );return IRQ_NONE ;
@@ -373,38 +371,30 @@ cpci_hp_intr(int irq, void *data, struct* INS bits of the cold-inserted devices. */ static int -init_slots (void ) +init_slots (int clear_ins) {struct slot *slot; -struct list_head *tmp;struct pci_dev * dev;dbg ("%s - enter" , __FUNCTION__);down_read (&list_rwsem); -if (!slots) { +if (!slots) {up_read (&list_rwsem);return -1 ; } -list_for_each (tmp, &slot_list) { - slot = list_entry(tmp, struct slot, slot_list); +list_for_each_entry (slot, &slot_list, slot_list) {dbg ("%s - looking at slot %s" , __FUNCTION__, slot->hotplug_slot->name); -if (cpci_check_and_clear_ins (slot)) { +if (clear_ins &&cpci_check_and_clear_ins (slot))dbg ("%s - cleared INS for slot %s" , __FUNCTION__, slot->hotplug_slot->name); - dev =pci_find_slot (slot->bus->number,PCI_DEVFN (slot->number,0 )); -if (dev) { -if (update_adapter_status (slot->hotplug_slot,1 )) { -warn ("failure to update adapter file" ); - } -if (update_latch_status (slot->hotplug_slot,1 )) { -warn ("failure to update latch file" ); - } - slot->dev = dev; - }else { -err ("%s - no driver attached to device in slot %s" , - __FUNCTION__, slot->hotplug_slot->name); - } + dev =pci_get_slot (slot->bus,PCI_DEVFN (slot->number,0 )); +if (dev) { +if (update_adapter_status (slot->hotplug_slot,1 )) +warn ("failure to update adapter file" ); +if (update_latch_status (slot->hotplug_slot,1 )) +warn ("failure to update latch file" ); + slot->dev = dev; } }up_read (&list_rwsem);
@@ -416,26 +406,28 @@ static intcheck_slots (void) {struct slot *slot; -struct list_head *tmp;int extracted;int inserted;u16 hs_csr;down_read (&list_rwsem); -if (!slots) { +if (!slots) {up_read (&list_rwsem);err ("no slots registered, shutting down" );return -1 ; } extracted = inserted =0 ; -list_for_each (tmp, &slot_list) { - slot = list_entry(tmp, struct slot, slot_list); +list_for_each_entry (slot, &slot_list, slot_list) {dbg ("%s - looking at slot %s" , __FUNCTION__, slot->hotplug_slot->name); -if (cpci_check_and_clear_ins (slot)) { -/* Some broken hardware (e.g. PLX 9054AB) asserts ENUM# twice... */ -if (slot->dev) { -warn ("slot %s already inserted" , slot->hotplug_slot->name); +if (cpci_check_and_clear_ins (slot)) { +/* +* Some broken hardware (e.g. PLX 9054AB) asserts +* ENUM# twice... +*/ +if (slot->dev) { +warn ("slot %s already inserted" , + slot->hotplug_slot->name); inserted++;continue ; }
@@ -452,7 +444,7 @@ check_slots(void)/* Configure device */ dbg ("%s - configuring slot %s" , __FUNCTION__, slot->hotplug_slot->name); -if (cpci_configure_slot (slot)) { +if (cpci_configure_slot (slot)) {err ("%s - could not configure slot %s" , __FUNCTION__, slot->hotplug_slot->name);continue ;
@@ -465,13 +457,11 @@ check_slots(void)dbg ("%s - slot %s HS_CSR (2) = %04x" , __FUNCTION__, slot->hotplug_slot->name, hs_csr); -if (update_latch_status (slot->hotplug_slot,1 )) { +if (update_latch_status (slot->hotplug_slot,1 ))warn ("failure to update latch file" ); - } -if (update_adapter_status (slot->hotplug_slot,1 )) { +if (update_adapter_status (slot->hotplug_slot,1 ))warn ("failure to update adapter file" ); - }cpci_led_off (slot);
@@ -481,7 +471,7 @@ check_slots(void) __FUNCTION__, slot->hotplug_slot->name, hs_csr); inserted++; - }else if (cpci_check_ext (slot)) { + }else if (cpci_check_ext (slot)) {/* Process extraction request */ dbg ("%s - slot %s extracted" , __FUNCTION__, slot->hotplug_slot->name);
@@ -491,27 +481,25 @@ check_slots(void)dbg ("%s - slot %s HS_CSR = %04x" , __FUNCTION__, slot->hotplug_slot->name, hs_csr); -if (!slot->extracting) { -if (update_latch_status (slot->hotplug_slot,0 )) { +if (!slot->extracting) { +if (update_latch_status (slot->hotplug_slot,0 )) {warn ("failure to update latch file" ); - } -atomic_inc (&extracting); slot->extracting =1 ; +atomic_inc (&extracting); } extracted++; - }else if (slot->extracting) { + }else if (slot->extracting) { hs_csr =cpci_get_hs_csr (slot); -if (hs_csr ==0xffff ) { +if (hs_csr ==0xffff ) {/* * Hmmm, we're likely hosed at this point, should we * bother trying to tell the driver or not? */ err ("card in slot %s was improperly removed" , slot->hotplug_slot->name); -if (update_adapter_status (slot->hotplug_slot,0 )) { +if (update_adapter_status (slot->hotplug_slot,0 ))warn ("failure to update adapter file" ); - } slot->extracting =0 ;atomic_dec (&extracting); }
@@ -520,10 +508,9 @@ check_slots(void)up_read (&list_rwsem);dbg ("inserted=%d, extracted=%d, extracting=%d" , inserted, extracted,atomic_read (&extracting)); -if (inserted || extracted) { +if (inserted || extracted)return extracted; - } -else if (!atomic_read (&extracting)) { +else if (!atomic_read (&extracting)) {err ("cannot find ENUM# source, shutting down" );return -1 ; }
@@ -541,12 +528,12 @@ event_thread(void *data)unlock_kernel ();dbg ("%s - event thread started" , __FUNCTION__); -while (1 ) { +while (1 ) {dbg ("event thread sleeping" );down_interruptible (&event_semaphore);dbg ("event thread woken, thread_finished = %d" , thread_finished); -if (thread_finished ||signal_pending (current)) +if (thread_finished ||signal_pending (current))break ; do { rc = check_slots();
@@ -558,7 +545,9 @@ event_thread(void *data) thread_finished =1 ;break ; } - }while (atomic_read (&extracting) !=0 ); + }while (atomic_read (&extracting) && !thread_finished); +if (thread_finished) +break ;/* Re-enable ENUM# interrupt */ dbg ("%s - re-enabling irq" , __FUNCTION__);
@@ -579,21 +568,21 @@ poll_thread(void *data)daemonize ("cpci_hp_polld" );unlock_kernel (); -while (1 ) { -if (thread_finished ||signal_pending (current)) +while (1 ) { +if (thread_finished ||signal_pending (current))break ; -if (controller->ops->query_enum ()) { +if (controller->ops->query_enum ()) {do { rc =check_slots (); -if (rc >0 ) { +if (rc >0 ) {/* Give userspace a chance to handle extraction */ msleep (500 ); - }else if (rc <0 ) { + }else if (rc <0 ) {dbg ("%s - error checking slots" , __FUNCTION__); thread_finished =1 ;break ; } - }while (atomic_read (&extracting) !=0 ); + }while (atomic_read (&extracting) && !thread_finished); }msleep (100 ); }
@@ -612,12 +601,11 @@ cpci_start_thread(void)init_MUTEX_LOCKED (&thread_exit); thread_finished =0 ; -if (controller->irq) { +if (controller->irq) pid =kernel_thread (event_thread,NULL ,0 ); - }else { +else pid =kernel_thread (poll_thread,NULL ,0 ); - } -if (pid <0 ) { +if (pid <0 ) {err ("Can't start up our thread" );return -1 ; }
@@ -630,9 +618,8 @@ cpci_stop_thread(void) { thread_finished =1 ;dbg ("thread finish command given" ); -if (controller->irq) { +if (controller->irq)up (&event_semaphore); - }dbg ("wait for thread to exit" );down (&thread_exit); }
@@ -642,45 +629,67 @@ cpci_hp_register_controller(struct cpci_ {int status =0 ; -if (!controller) { - controller = new_controller; -if (controller->irq) { -if (request_irq (controller->irq, - cpci_hp_intr, - controller->irq_flags, -MY_NAME , controller->dev_id)) { -err ("Can't get irq %d for the hotplug cPCI controller" , controller->irq); - status = -ENODEV ; - } -dbg ("%s - acquired controller irq %d" , __FUNCTION__, - controller->irq); +if (controller) +return -1 ; +if (!(new_controller && new_controller->ops)) +return -EINVAL ; +if (new_controller->irq) { +if (!(new_controller->ops->enable_irq && + new_controller->ops->disable_irq)) + status = -EINVAL ; +if (request_irq (new_controller->irq, + cpci_hp_intr, + new_controller->irq_flags, +MY_NAME , + new_controller->dev_id)) { +err ("Can't get irq %d for the hotplug cPCI controller" , + new_controller->irq); + status = -ENODEV ; } - }else { -err ("cPCI hotplug controller already registered" ); - status =-1 ; +dbg ("%s - acquired controller irq %d" , + __FUNCTION__, new_controller->irq); } +if (!status) + controller = new_controller;return status; } +static void +cleanup_slots (void ) +{ +struct slot *slot; +struct slot *tmp; + +/* +* Unregister all of our slots with the pci_hotplug subsystem, +* and free up all memory that we had allocated. +*/ +down_write (&list_rwsem); +if (!slots) + goto cleanup_null; +list_for_each_entry_safe (slot, tmp, &slot_list, slot_list) { +list_del (&slot->slot_list); +pci_hp_deregister (slot->hotplug_slot); + } +cleanup_null: +up_write (&list_rwsem); +return ; +} +int cpci_hp_unregister_controller (struct cpci_hp_controller *old_controller) {int status =0 ; -if (controller) { -if (atomic_read (&extracting) !=0 ) { -return -EBUSY ; - } -if (!thread_finished) { +if (controller) { +if (!thread_finished)cpci_stop_thread (); - } -if (controller->irq) { +if (controller->irq)free_irq (controller->irq, controller->dev_id); - } controller =NULL ; - }else { +cleanup_slots (); + }else status = -ENODEV ; - }return status; }
@@ -691,32 +700,28 @@ cpci_hp_start(void)int status;dbg ("%s - enter" , __FUNCTION__); -if (!controller) { +if (!controller)return -ENODEV ; - }down_read (&list_rwsem); -if (list_empty (&slot_list)) { +if (list_empty (&slot_list)) {up_read (&list_rwsem);return -ENODEV ; }up_read (&list_rwsem); -if (first) { - status =init_slots (); -if (status) { -return status; - } + status =init_slots (first); +if (first) first =0 ; - } +if (status) +return status; status =cpci_start_thread (); -if (status) { +if (status)return status; - }dbg ("%s - thread started" , __FUNCTION__); -if (controller->irq) { +if (controller->irq) {/* Start enum interrupt processing */ dbg ("%s - enabling irq" , __FUNCTION__); controller->ops->enable_irq ();
@@ -728,13 +733,9 @@ cpci_hp_start(void)int cpci_hp_stop (void ) { -if (!controller) { +if (!controller)return -ENODEV ; - } -if (atomic_read (&extracting) !=0 ) { -return -EBUSY ; - } -if (controller->irq) { +if (controller->irq) {/* Stop enum interrupt processing */ dbg ("%s - disabling irq" , __FUNCTION__); controller->ops->disable_irq ();
@@ -743,34 +744,6 @@ cpci_hp_stop(void)return 0 ; } -static void __exit -cleanup_slots (void ) -{ -struct list_head *tmp; -struct slot *slot; - -/* -* Unregister all of our slots with the pci_hotplug subsystem, -* and free up all memory that we had allocated. -*/ -down_write (&list_rwsem); -if (!slots) { - goto null_cleanup; - } -list_for_each (tmp, &slot_list) { - slot = list_entry(tmp, struct slot, slot_list); -list_del (&slot->slot_list); -pci_hp_deregister (slot->hotplug_slot); -kfree (slot->hotplug_slot->info); -kfree (slot->hotplug_slot->name); -kfree (slot->hotplug_slot); -kfree (slot); - } - null_cleanup: -up_write (&list_rwsem); -return ; -} -int __init cpci_hotplug_init (int debug) {
@@ -784,7 +757,8 @@ cpci_hotplug_exit(void)/* * Clean everything up. */ -cleanup_slots (); +cpci_hp_stop (); +cpci_hp_unregister_controller (controller); }EXPORT_SYMBOL_GPL (cpci_hp_register_controller);
drivers/pci/hotplug/cpci_hotplug_pci.c +54 -90
297 lines not on this page · raw message
Re: [PATCH] PCI Hotplug: more CPCI updates
Quoted 5 lines · Greg KH
[PATCH] PCI Hotplug: more CPCI updates
- Switch to
pci_get_slotinstead of deprecatedpci_find_slot.
- A bunch of CodingStyle fixes.
drivers/pci/hotplug/cpci_hotplug_core.c cpci_hp_intr(int irq, void *data, struct · line 399
Quoted 10 lines
drivers/pci/hotplug/cpci_hotplug_core.c +7
-1- } + dev =pci_get_slot (slot->bus,PCI_DEVFN (slot->number,0 )); +if (dev) { +if (update_adapter_status (slot->hotplug_slot,1 )) +warn ("failure to update adapter file" ); +if (update_latch_status (slot->hotplug_slot,1 )) +warn ("failure to update latch file" ); + slot->dev = dev; } }
I don't claim to know the code as well as Scott or Greg does, but I don't see a pci_put_dev for the slot->dev to clean up the usage count?
Quoted 4 lines
drivers/pci/hotplug/cpci_hotplug_pci.c +0
-0@@ -311,10 +277,10 @@ int cpci_configure_slot(struct slot* slo*/ n =pci_scan_slot (slot->bus, slot->devfn);dbg ("%s: pci_scan_slot returned %d" , __FUNCTION__, n);
drivers/pci/hotplug/cpci_hotplug_pci.c int cpci_configure_slot(struct slot* slo · line 286
Quoted 10 lines
drivers/pci/hotplug/cpci_hotplug_pci.c +3
-3-if (n >0 ) +if (n >0 )pci_bus_add_devices (slot->bus); - slot->dev =pci_find_slot (slot->bus->number, slot->devfn); -if (slot->dev ==NULL ) { + slot->dev =pci_get_slot (slot->bus, slot->devfn); +if (slot->dev ==NULL ) {err ("Could not find PCI device for slot %02x" , slot->number);return 1 ; }
Same here -- I don't see a pci_put_dev for the slot->dev.
Quoted 9 lines
drivers/pci/hotplug/cpci_hotplug_pci.c +1
-1@@ -341,15 +305,15 @@ int cpci_unconfigure_slot(struct slot* sstruct pci_dev *dev;dbg ("%s - enter" , __FUNCTION__); -if (!slot->dev) { +if (!slot->dev) {err ("No device for slot %02x\n" , slot->number);return -ENODEV ; }
drivers/pci/hotplug/cpci_hotplug_pci.c int cpci_unconfigure_slot(struct slot* s · line 319
Quoted 10 lines
drivers/pci/hotplug/cpci_hotplug_pci.c +2
-2for (i =0 ; i <8 ; i++) { - dev =pci_find_slot (slot->bus->number, + dev =pci_get_slot (slot->bus,PCI_DEVFN (PCI_SLOT (slot->devfn), i)); -if (dev) { +if (dev) {pci_remove_bus_device (dev); slot->dev =NULL ; }
And again here...
P.
Re: [PATCH] PCI Hotplug: more CPCI updates
Quoted 21 lines · Prarit Bhargava
Quoted 7 lines · Greg KH
[PATCH] PCI Hotplug: more CPCI updates
- Switch to
pci_get_slotinstead of deprecatedpci_find_slot.
- A bunch of CodingStyle fixes.drivers/pci/hotplug/cpci_hotplug_core.c cpci_hp_intr(int irq, void *data, struct · line 399
Quoted 10 lines
drivers/pci/hotplug/cpci_hotplug_core.c +7
-1- } + dev =pci_get_slot (slot->bus,PCI_DEVFN (slot->number,0 )); +if (dev) { +if (update_adapter_status (slot->hotplug_slot,1 )) +warn ("failure to update adapter file" ); +if (update_latch_status (slot->hotplug_slot,1 )) +warn ("failure to update latch file" ); + slot->dev = dev; } }
Quoted 2 lines
I don't claim to know the code as well as Scott or Greg does, but I don't see a
pci_put_devfor theslot->devto clean up the usage count?
s/pci_put_dev/pci_dev_put/g
:)
P.
Re: [PATCH] PCI Hotplug: more CPCI updates
Quoted 22 lines · Prarit Bhargava
Quoted 21 lines · Prarit Bhargava
Quoted 17 lines · Greg KH
[PATCH] PCI Hotplug: more CPCI updates
- Switch to
pci_get_slotinstead of deprecatedpci_find_slot.
- A bunch of CodingStyle fixes.drivers/pci/hotplug/cpci_hotplug_core.c +7
-1- } + dev =pci_get_slot (slot->bus,PCI_DEVFN (slot->number,0 )); +if (dev) { +if (update_adapter_status (slot->hotplug_slot,1 )) +warn ("failure to update adapter file" ); +if (update_latch_status (slot->hotplug_slot,1 )) +warn ("failure to update latch file" ); + slot->dev = dev; } }
Quoted 4 lines
Quoted 2 lines
I don't claim to know the code as well as Scott or Greg does, but I don't see a
pci_put_devfor theslot->devto clean up the usage count?s/pci_put_dev/pci_dev_put/g
Sorry Prarit, when you suggested I switch over to pci_get_slot in your previous comments to me, I didn't look that closely and missed the reference counting. Greg, I think the required fix is just a couple of lines in my hotplug slot release function, I'll code it up and test it ASAP tomorrow with an eye on getting a patch off by early afternoon EDT.
Scott
Signature · 4 lines
Scott Murray SOMA Networks, Inc. Toronto, Ontario e-mail: [email protected]
Re: [PATCH] PCI Hotplug: more CPCI updates
Quoted 27 lines · Scott Murray
Quoted 25 lines · Prarit Bhargava
Quoted 23 lines · Prarit Bhargava
Quoted 7 lines · Greg KH
[PATCH] PCI Hotplug: more CPCI updates
- Switch to
pci_get_slotinstead of deprecatedpci_find_slot.
- A bunch of CodingStyle fixes.Review
Quoted 10 lines
(quoted diff) +7
-1- } + dev = pci_get_slot(slot->bus, PCI_DEVFN(slot->number, 0)); + if (dev) { + if (update_adapter_status(slot->hotplug_slot, 1)) + warn("failure to update adapter file"); + if (update_latch_status(slot->hotplug_slot, 1)) + warn("failure to update latch file"); + slot->dev = dev; } }I don't claim to know the code as well as Scott or Greg does, but I don't see a
pci_put_devfor theslot->devto clean up the usage count?
Quoted 7 lines
Quoted 1 line
s/pci_put_dev/pci_dev_put/g
Sorry Prarit, when you suggested I switch over to
pci_get_slotin your previous comments to me, I didn't look that closely and missed the reference counting. Greg, I think the required fix is just a couple of lines in my hotplug slot release function, I'll code it up and test it ASAP tomorrow with an eye on getting a patch off by early afternoon EDT.
Greg, sorry for the delay, here's a patch that fixes up the pci_dev refcounting in the CPCI code. I've done some testing against it and it seems fine here.
cpci_hotplug_core.c | 2 ++ cpci_hotplug_pci.c | 5 ++++- 2 files changed, 6 insertions(+), 1 deletion(-)
drivers/pci/hotplug/cpci_hotplug_core.c +2 -0
@@ -217,6 +217,8 @@kfree (slot->hotplug_slot->info);kfree (slot->hotplug_slot->name);kfree (slot->hotplug_slot); +if (slot->dev) +pci_dev_put (slot->dev);kfree (slot); }
drivers/pci/hotplug/cpci_hotplug_pci.c +4 -1
@@ -315,9 +315,12 @@ PCI_DEVFN(PCI_SLOT(slot->devfn), i));if (dev) {pci_remove_bus_device (dev); - slot->dev =NULL ; +pci_dev_put (dev); } } +pci_dev_put (slot->dev); + slot->dev =NULL ; +dbg ("%s - exit" , __FUNCTION__);return 0 ; }
Signature · 4 lines
Scott Murray SOMA Networks, Inc. Toronto, Ontario e-mail: [email protected]
Re: [PATCH] PCI: amd74xx patch for new NVIDIA device IDs
In reply to Greg KHQuoted 14 lines · Greg KH
[PATCH] PCI: amd74xx patch for new NVIDIA device IDs
Here's the 2.6 amd74xx patch for NVIDIA MCP51.
Signed-off-by Andy Currid <[email protected]>Signed-off-by Greg Kroah-Hartman <[email protected]>---
commitaf00f9811e0ccbd3db84ddc4cffb0da942653393
tree5a9c3b7f7d61d96d3624ad130b173a761cb7dac2
parent2ac2610b26c9da72820443328ff2c56c7b8c87b8
author Andy Currid <[email protected]> Mon, 23 May 2005 08:55:45 -0700 committer Greg KH <[email protected]> Tue, 31 May 2005 14:26:38 -0700
Quoted 2 lines
drivers/ide/pci/amd74xx.c | 3 +++ include/linux/pci_ids.h | 6 ++++++
This is hardly a PCI patch.
Has Bart, the IDE maintainer, acked this?
Jeff
Re: [PATCH] PCI: amd74xx patch for new NVIDIA device IDs
Quoted 19 lines · Jeff Garzik
Quoted 14 lines · Greg KH
[PATCH] PCI: amd74xx patch for new NVIDIA device IDs
Here's the 2.6 amd74xx patch for NVIDIA MCP51.
Signed-off-by Andy Currid <[email protected]>Signed-off-by Greg Kroah-Hartman <[email protected]>---
commitaf00f9811e0ccbd3db84ddc4cffb0da942653393
tree5a9c3b7f7d61d96d3624ad130b173a761cb7dac2
parent2ac2610b26c9da72820443328ff2c56c7b8c87b8
author Andy Currid <[email protected]> Mon, 23 May 2005 08:55:45 -0700 committer Greg KH <[email protected]> Tue, 31 May 2005 14:26:38 -0700Quoted 2 lines
drivers/ide/pci/amd74xx.c | 3 +++ include/linux/pci_ids.h | 6 ++++++
Quoted 6 lines
This is hardly a PCI patch.
Has Bart, the IDE maintainer, acked this?
Jeff
Thanks Jeff.
Ack.