Skip to contentSkip to conversation

linux-next: pci merge conflict

PATCH New Stephen Rothwell 3 participants 11 messages May 28, 2008

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

Hi Jesse,

Today's linux-next mege of the pci tree got conflicts in
drivers/pci/hotplug/pciehp_hpc.c between commits
5808639bfa98d69f77a481d759570d85f164fea0 ("pciehp: fix slow probing") and 6592e02ae4bd7b277230aa0c5821588a13b9d8e3 ("pciehp: poll cmd completion if hotplug interrupt is disabled") from the pci-current tree and commit
2b26196717939221ccc06073780b73d0f4ee3203 ("pciehp: fix waiting
command"). I fixed it up as best I can, but please have a look.

Signature · 3 lines
Cheers,
Stephen Rothwell                    [email protected]
http://www.canb.auug.org.au/~sfr/
#1 lore raw

Hi Stephen-san, Jesse-san,

I have already have updated patches. I'll send them if necessary.

Thanks,
Kenji Kaneshige

Quoted 9 lines · Stephen Rothwell

Hi Jesse,

Today's linux-next mege of the pci tree got conflicts in
drivers/pci/hotplug/pciehp_hpc.c between commits
5808639bfa98d69f77a481d759570d85f164fea0 ("pciehp: fix slow probing") and 6592e02ae4bd7b277230aa0c5821588a13b9d8e3 ("pciehp: poll cmd completion if hotplug interrupt is disabled") from the pci-current tree and commit
2b26196717939221ccc06073780b73d0f4ee3203 ("pciehp: fix waiting
command"). I fixed it up as best I can, but please have a look.

#2 lore raw

Hi Kenji,

Quoted 2 lines · On Wed, 28 May 2008 11:50:27 +0900 Kenji Kaneshige…

I have already have updated patches. I'll send them if necessary.

That would be good for tomorrow as the pci-current tree has now gone into Linus' tree.

Thanks

Signature · 3 lines
Cheers,
Stephen Rothwell                    [email protected]
http://www.canb.auug.org.au/~sfr/
#3 lore raw
In reply to Kenji Kaneshige
Quoted 17 lines · Kenji Kaneshige

Hi Stephen-san, Jesse-san,

I have already have updated patches. I'll send them if necessary.

Thanks,
Kenji Kaneshige

Quoted 9 lines · Stephen Rothwell

Hi Jesse,

Today's linux-next mege of the pci tree got conflicts in drivers/pci/hotplug/pciehp_hpc.c between commits 5808639bfa98d69f77a481d759570d85f164fea0 ("pciehp: fix slow probing") and 6592e02ae4bd7b277230aa0c5821588a13b9d8e3 ("pciehp: poll cmd completion if hotplug interrupt is disabled") from the pci-current tree and commit 2b26196717939221ccc06073780b73d0f4ee3203 ("pciehp: fix waiting command"). I fixed it up as best I can, but please have a look.

#4 lore raw
In reply to Kenji Kaneshige
Quoted 3 lines · Kenji Kaneshige

Hi Stephen-san, Jesse-san,

I have already have updated patches. I'll send them if necessary.

Oops, meant to actually write a message. :)

I'll merge the Linus tree into linux-next tomorrow. Kenji-san, I'll probably ask for updated patches then for anything that's in linux-next but not in Linus' tree yet.

Thanks,
Jesse

#5 lore raw
Quoted 11 lines · Jesse Barnes
Quoted 3 lines · Kenji Kaneshige

Hi Stephen-san, Jesse-san,

I have already have updated patches. I'll send them if necessary.

Oops, meant to actually write a message. :)

I'll merge the Linus tree into linux-next tomorrow. Kenji-san, I'll probably ask for updated patches then for anything that's in linux-next but not in Linus' tree yet.

Ok, just in case, I send patches related to pciehp now. I'll send the flowing patches followed by this e-mail.

- [PATCH 1/4] Updated version of 72c4928058175e889b2f83d1a62ac6f2eb0498a0
- [PATCH 2/4] Updated version of 4c8cc2d54b1d35cf24ede6057e05f103cfa28852
- [PATCH 3/4] Updated version of 2b26196717939221ccc06073780b73d0f4ee3203
- [PATCH 4/4] Updated version of 9d06d9963852e7818157994c645c640239cb5137

I made these patches against 2.6.24-rc4 with recent pciehp regression fixes applied.

Thanks,
Kenji Kaneshige

#6 lore raw

PATCH 1/4 Updated version of 72c4928058175e889b2f83d1a62ac6f2eb0498a0

Remove the redundant initialization of pci_dev member of struct controller in pciehp_probe(). It is initialized in pcie_init().

---
 drivers/pci/hotplug/pciehp_core.c |    1 -
 1 file changed, 1 deletion(-)
drivers/pci/hotplug/pciehp_core.c +0 -1
Index: linux-2.6.26-rc4/drivers/pci/hotplug/pciehp_core.c
===================================================================
@@ -454,7 +454,6 @@ static int pciehp_probe(struct pcie_devi
 	INIT_LIST_HEAD(&ctrl->slot_list);
 
 	pdev = dev->port;
-	ctrl->pci_dev = pdev;
 
 	rc = pcie_init(ctrl, dev);
 	if (rc) {
#7 lore raw

PATCH 2/4 Updated version of 4c8cc2d54b1d35cf24ede6057e05f103cfa28852

In reply to Kenji Kaneshige

Current pciehp evaluates _OSC/OSHP method after some controller initialization is done. So if evaluating _OSC/OSHP is failed, we need to cleanup already initialized data structures or hardware. This clearly is not robust way. With this patch, _OSC/OSHP evaluation is done first.

---
 drivers/pci/hotplug/pciehp.h      |    1 +
 drivers/pci/hotplug/pciehp_core.c |   10 +++++++---
 drivers/pci/hotplug/pciehp_hpc.c  |   17 ++---------------
 3 files changed, 10 insertions(+), 18 deletions(-)
drivers/pci/hotplug/pciehp_core.c +7 -3
Index: linux-2.6.26-rc4/drivers/pci/hotplug/pciehp_core.c
===================================================================
@@ -444,7 +444,13 @@ static int pciehp_probe(struct pcie_devi
 	struct controller *ctrl;
 	struct slot *t_slot;
 	u8 value;
-	struct pci_dev *pdev;
+	struct pci_dev *pdev = dev->port;
+
+	if (pciehp_force)
+		dbg("Bypassing BIOS check for pciehp use on %s\n",
+		    pci_name(pdev));
+	else if (pciehp_get_hp_hw_control_from_firmware(pdev))
+		goto err_out_none;
 
 	ctrl = kzalloc(sizeof(*ctrl), GFP_KERNEL);
 	if (!ctrl) {
@@ -453,8 +459,6 @@ static int pciehp_probe(struct pcie_devi
 	}
 	INIT_LIST_HEAD(&ctrl->slot_list);
 
-	pdev = dev->port;
-
 	rc = pcie_init(ctrl, dev);
 	if (rc) {
 		dbg("%s: controller initialization failed\n", PCIE_MODULE_NAME);
drivers/pci/hotplug/pciehp_hpc.c +2 -15
Index: linux-2.6.26-rc4/drivers/pci/hotplug/pciehp_hpc.c
===================================================================
@@ -1018,7 +1018,7 @@ static struct hpc_ops pciehp_hpc_ops = {
 };
 
 #ifdef CONFIG_ACPI
-static int pciehp_acpi_get_hp_hw_control_from_firmware(struct pci_dev *dev)
+int pciehp_acpi_get_hp_hw_control_from_firmware(struct pci_dev *dev)
 {
 	acpi_status status;
 	acpi_handle chandle, handle = DEVICE_ACPI_HANDLE(&(dev->dev));
@@ -1122,23 +1122,10 @@ int pcie_init_hardware_part2(struct cont
 
 	if (pcie_write_cmd(ctrl, cmd, mask)) {
 		err("%s: Cannot enable software notification\n", __func__);
-		goto abort;
+		return -1;
 	}
 
-	if (pciehp_force)
-		dbg("Bypassing BIOS check for pciehp use on %s\n",
-				pci_name(ctrl->pci_dev));
-	else if (pciehp_get_hp_hw_control_from_firmware(ctrl->pci_dev))
-		goto abort_disable_intr;
-
 	return 0;
-
-	/* We end up here for the many possible ways to fail this API. */
-abort_disable_intr:
-	if (pcie_write_cmd(ctrl, 0, HP_INTR_ENABLE))
-		err("%s : disabling interrupts failed\n", __func__);
-abort:
-	return -1;
 }
 
 static inline void dbg_ctrl(struct controller *ctrl)
drivers/pci/hotplug/pciehp.h +1 -0
Index: linux-2.6.26-rc4/drivers/pci/hotplug/pciehp.h
===================================================================
@@ -202,6 +202,7 @@ struct hpc_ops {
 #include <acpi/actypes.h>
 #include <linux/pci-acpi.h>
 
+extern int pciehp_acpi_get_hp_hw_control_from_firmware(struct pci_dev *dev);
 #define pciehp_get_hp_hw_control_from_firmware(dev)			\
 	pciehp_acpi_get_hp_hw_control_from_firmware(dev)
 static inline int pciehp_get_hp_params_from_firmware(struct pci_dev *dev,
#8 lore raw

PATCH 3/4 Updated version of 2b26196717939221ccc06073780b73d0f4ee3203

In reply to Kenji Kaneshige

Since we need to wait for command completion for muximum 1sec, waiting command should not be interrupted by a signal.

---
 drivers/pci/hotplug/pciehp_hpc.c |   16 ++++------------
 1 file changed, 4 insertions(+), 12 deletions(-)
drivers/pci/hotplug/pciehp_hpc.c +4 -12
Index: linux-2.6.26-rc4/drivers/pci/hotplug/pciehp_hpc.c
===================================================================
@@ -268,9 +268,8 @@ completed:
 	return timeout;
 }
 
-static inline int pcie_wait_cmd(struct controller *ctrl, int poll)
+static inline void pcie_wait_cmd(struct controller *ctrl, int poll)
 {
-	int retval = 0;
 	unsigned int msecs = pciehp_poll_mode ? 2500 : 1000;
 	unsigned long timeout = msecs_to_jiffies(msecs);
 	int rc;
@@ -278,16 +277,9 @@ static inline int pcie_wait_cmd(struct c
 	if (poll)
 		rc = pcie_poll_cmd(ctrl);
 	else
-		rc = wait_event_interruptible_timeout(ctrl->queue,
-					      !ctrl->cmd_busy, timeout);
+		rc = wait_event_timeout(ctrl->queue, !ctrl->cmd_busy, timeout);
 	if (!rc)
 		dbg("Command not completed in 1000 msec\n");
-	else if (rc < 0) {
-		retval = -EINTR;
-		info("Command was interrupted by a signal\n");
-	}
-
-	return retval;
 }
 
 /**
@@ -365,7 +357,7 @@ static int pcie_write_cmd(struct control
 		if (!(slot_ctrl & HP_INTR_ENABLE) ||
 		    !(slot_ctrl & CMD_CMPL_INTR_ENABLE))
 			poll = 1;
-                retval = pcie_wait_cmd(ctrl, poll);
+                pcie_wait_cmd(ctrl, poll);
 	}
  out:
 	mutex_unlock(&ctrl->ctrl_lock);
@@ -797,7 +789,7 @@ static irqreturn_t pcie_isr(int irq, voi
 	if (intr_loc & CMD_COMPLETED) {
 		ctrl->cmd_busy = 0;
 		smp_mb();
-		wake_up_interruptible(&ctrl->queue);
+		wake_up(&ctrl->queue);
 	}
 
 	if (!(intr_loc & ~CMD_COMPLETED))
#9 lore raw

PATCH 4/4 Updated version of 9d06d9963852e7818157994c645c640239cb5137

In reply to Kenji Kaneshige

Fix the following problems of shpchp driver about getting hotplug control from firmware.

- The shpchp driver must not control the hotplug controller if it fails to get control from the firmware. But current shpchp controls the hotplug controller regardless the result, because it doesn't check the return value of get_hp_hw_control_from_firmware().

- Current shpchp driver doesn't support _OSC.

The pciehp driver already have the code for evaluating _OSC and OSHP and shpchp and pciehp can share it. So this patch move that code from pciehp to acpi_pcihp.c.

---
 drivers/pci/hotplug/acpi_pcihp.c  |   85 ++++++++++++++++++++++++++++++++++++--
 drivers/pci/hotplug/pciehp.h      |   10 +++-
 drivers/pci/hotplug/pciehp_hpc.c  |   69 ------------------------------
 drivers/pci/hotplug/shpchp.h      |   14 +++---
 drivers/pci/hotplug/shpchp_core.c |   15 +++---
 drivers/pci/hotplug/shpchp_hpc.c  |    1 
 include/linux/pci_hotplug.h       |    2 
 7 files changed, 105 insertions(+), 91 deletions(-)
drivers/pci/hotplug/acpi_pcihp.c +81 -4
Index: linux-2.6.26-rc4/drivers/pci/hotplug/acpi_pcihp.c
===================================================================
@@ -30,6 +30,7 @@
 #include <linux/types.h>
 #include <linux/pci.h>
 #include <linux/pci_hotplug.h>
+#include <linux/pci-acpi.h>
 #include <acpi/acpi.h>
 #include <acpi/acpi_bus.h>
 #include <acpi/actypes.h>
@@ -299,7 +300,7 @@ free_and_return:
  *
  * @handle - the handle of the hotplug controller.
  */
-acpi_status acpi_run_oshp(acpi_handle handle)
+static acpi_status acpi_run_oshp(acpi_handle handle)
 {
 	acpi_status		status;
 	struct acpi_buffer	string = { ACPI_ALLOCATE_BUFFER, NULL };
@@ -322,9 +323,6 @@ acpi_status acpi_run_oshp(acpi_handle ha
 	kfree(string.pointer);
 	return status;
 }
-EXPORT_SYMBOL_GPL(acpi_run_oshp);
-
-
 
 /* acpi_get_hp_params_from_firmware
  *
@@ -374,6 +372,85 @@ acpi_status acpi_get_hp_params_from_firm
 }
 EXPORT_SYMBOL_GPL(acpi_get_hp_params_from_firmware);
 
+/**
+ * acpi_get_hp_hw_control_from_firmware
+ * @dev: the pci_dev of the bridge that has a hotplug controller
+ * @flags: requested control bits for _OSC
+ *
+ * Attempt to take hotplug control from firmware.
+ */
+int acpi_get_hp_hw_control_from_firmware(struct pci_dev *dev, u32 flags)
+{
+	acpi_status status;
+	acpi_handle chandle, handle = DEVICE_ACPI_HANDLE(&(dev->dev));
+	struct pci_dev *pdev = dev;
+	struct pci_bus *parent;
+	struct acpi_buffer string = { ACPI_ALLOCATE_BUFFER, NULL };
+
+	flags &= (OSC_PCI_EXPRESS_NATIVE_HP_CONTROL |
+		  OSC_SHPC_NATIVE_HP_CONTROL |
+		  OSC_PCI_EXPRESS_CAP_STRUCTURE_CONTROL);
+	if (!flags) {
+		err("Invalid flags %u specified!\n", flags);
+		return -EINVAL;
+	}
+
+	/*
+	 * Per PCI firmware specification, we should run the ACPI _OSC
+	 * method to get control of hotplug hardware before using it. If
+	 * an _OSC is missing, we look for an OSHP to do the same thing.
+	 * To handle different BIOS behavior, we look for _OSC and OSHP
+	 * within the scope of the hotplug controller and its parents,
+	 * upto the host bridge under which this controller exists.
+	 */
+	while (!handle) {
+		/*
+		 * This hotplug controller was not listed in the ACPI name
+		 * space at all. Try to get acpi handle of parent pci bus.
+		 */
+		if (!pdev || !pdev->bus->parent)
+			break;
+		parent = pdev->bus->parent;
+		dbg("Could not find %s in acpi namespace, trying parent\n",
+		    pci_name(pdev));
+		if (!parent->self)
+			/* Parent must be a host bridge */
+			handle = acpi_get_pci_rootbridge_handle(
+					pci_domain_nr(parent),
+					parent->number);
+		else
+			handle = DEVICE_ACPI_HANDLE(&(parent->self->dev));
+		pdev = parent->self;
+	}
+
+	while (handle) {
+		acpi_get_name(handle, ACPI_FULL_PATHNAME, &string);
+		dbg("Trying to get hotplug control for %s \n",
+		    (char *)string.pointer);
+		status = pci_osc_control_set(handle, flags);
+		if (status == AE_NOT_FOUND)
+			status = acpi_run_oshp(handle);
+		if (ACPI_SUCCESS(status)) {
+			dbg("Gained control for hotplug HW for pci %s (%s)\n",
+			    pci_name(dev), (char *)string.pointer);
+			kfree(string.pointer);
+			return 0;
+		}
+		if (acpi_root_bridge(handle))
+			break;
+		chandle = handle;
+		status = acpi_get_parent(chandle, &handle);
+		if (ACPI_FAILURE(status))
+			break;
+	}
+
+	dbg("Cannot get control of hotplug hardware for pci %s\n",
+	    pci_name(dev));
+
+	kfree(string.pointer);
+	return -ENODEV;
+}
+EXPORT_SYMBOL(acpi_get_hp_hw_control_from_firmware);
 
 /* acpi_root_bridge - check to see if this acpi object is a root bridge
  *
drivers/pci/hotplug/pciehp.h +7 -3
Index: linux-2.6.26-rc4/drivers/pci/hotplug/pciehp.h
===================================================================
@@ -202,9 +202,13 @@ struct hpc_ops {
 #include <acpi/actypes.h>
 #include <linux/pci-acpi.h>
 
-extern int pciehp_acpi_get_hp_hw_control_from_firmware(struct pci_dev *dev);
-#define pciehp_get_hp_hw_control_from_firmware(dev)			\
-	pciehp_acpi_get_hp_hw_control_from_firmware(dev)
+static inline int pciehp_get_hp_hw_control_from_firmware(struct pci_dev *dev)
+{
+	u32 flags = (OSC_PCI_EXPRESS_NATIVE_HP_CONTROL |
+		     OSC_PCI_EXPRESS_CAP_STRUCTURE_CONTROL);
+	return acpi_get_hp_hw_control_from_firmware(dev, flags);
+}
+
 static inline int pciehp_get_hp_params_from_firmware(struct pci_dev *dev,
 			struct hotplug_params *hpp)
 {
drivers/pci/hotplug/pciehp_hpc.c +0 -69

75 lines not on this page · raw message

include/linux/pci_hotplug.h +1 -1
Index: linux-2.6.26-rc4/include/linux/pci_hotplug.h
===================================================================
@@ -227,9 +227,9 @@ struct hotplug_params {
 #include <acpi/acpi.h>
 #include <acpi/acpi_bus.h>
 #include <acpi/actypes.h>
-extern acpi_status acpi_run_oshp(acpi_handle handle);
 extern acpi_status acpi_get_hp_params_from_firmware(struct pci_bus *bus,
 				struct hotplug_params *hpp);
+int acpi_get_hp_hw_control_from_firmware(struct pci_dev *dev, u32 flags);
 int acpi_root_bridge(acpi_handle handle);
 #endif
 #endif
drivers/pci/hotplug/shpchp.h +8 -6
Index: linux-2.6.26-rc4/drivers/pci/hotplug/shpchp.h
===================================================================
@@ -170,6 +170,7 @@ extern void shpchp_queue_pushbutton_work
 extern int shpc_init( struct controller *ctrl, struct pci_dev *pdev);
 
 #ifdef CONFIG_ACPI
+#include <linux/pci-acpi.h>
 static inline int get_hp_params_from_firmware(struct pci_dev *dev,
 					      struct hotplug_params *hpp)
 {
@@ -177,14 +178,15 @@ static inline int get_hp_params_from_fir
 			return -ENODEV;
 	return 0;
 }
-#define get_hp_hw_control_from_firmware(pdev)				\
-	do {								\
-		if (DEVICE_ACPI_HANDLE(&(pdev->dev)))			\
-			acpi_run_oshp(DEVICE_ACPI_HANDLE(&(pdev->dev)));\
-	} while (0)
+
+static inline int get_hp_hw_control_from_firmware(struct pci_dev *dev)
+{
+	u32 flags = OSC_SHPC_NATIVE_HP_CONTROL;
+	return acpi_get_hp_hw_control_from_firmware(dev, flags);
+}
 #else
 #define get_hp_params_from_firmware(dev, hpp) (-ENODEV)
-#define get_hp_hw_control_from_firmware(dev) do { } while (0)
+#define get_hp_hw_control_from_firmware(dev) (0)
 #endif
 
 struct ctrl_reg {
drivers/pci/hotplug/shpchp_core.c +8 -7
Index: linux-2.6.26-rc4/drivers/pci/hotplug/shpchp_core.c
===================================================================
@@ -330,13 +330,14 @@ static int get_cur_bus_speed (struct hot
 
 static int is_shpc_capable(struct pci_dev *dev)
 {
-       if ((dev->vendor == PCI_VENDOR_ID_AMD) || (dev->device ==
-                               PCI_DEVICE_ID_AMD_GOLAM_7450))
-               return 1;
-       if (pci_find_capability(dev, PCI_CAP_ID_SHPC))
-               return 1;
-
-       return 0;
+	if ((dev->vendor == PCI_VENDOR_ID_AMD) || (dev->device ==
+						PCI_DEVICE_ID_AMD_GOLAM_7450))
+		return 1;
+	if (!pci_find_capability(dev, PCI_CAP_ID_SHPC))
+		return 0;
+	if (get_hp_hw_control_from_firmware(dev))
+		return 0;
+	return 1;
 }
 
 static int shpc_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
drivers/pci/hotplug/shpchp_hpc.c +0 -1
Index: linux-2.6.26-rc4/drivers/pci/hotplug/shpchp_hpc.c
===================================================================
@@ -1084,7 +1084,6 @@ int shpc_init(struct controller *ctrl, s
 	dbg("%s: HPC at b:d:f:irq=0x%x:%x:%x:%x\n", __func__,
 			pdev->bus->number, PCI_SLOT(pdev->devfn),
 			PCI_FUNC(pdev->devfn), pdev->irq);
-	get_hp_hw_control_from_firmware(pdev);
 
 	/*
 	 * If this is the first controller to be initialized,
#10 lore raw
In reply to Kenji Kaneshige
Quoted 15 lines · Kenji Kaneshige
Quoted 10 lines · Jesse Barnes
Quoted 3 lines · Kenji Kaneshige

Hi Stephen-san, Jesse-san,

I have already have updated patches. I'll send them if necessary.

Oops, meant to actually write a message. :)

I'll merge the Linus tree into linux-next tomorrow. Kenji-san, I'll probably ask for updated patches then for anything that's in linux-next but not in Linus' tree yet.

Ok, just in case, I send patches related to pciehp now. I'll send the flowing patches followed by this e-mail.

Quoted 7 lines

- [PATCH 1/4] Updated version of 72c4928058175e889b2f83d1a62ac6f2eb0498a0
- [PATCH 2/4] Updated version of 4c8cc2d54b1d35cf24ede6057e05f103cfa28852
- [PATCH 3/4] Updated version of 2b26196717939221ccc06073780b73d0f4ee3203
- [PATCH 4/4] Updated version of 9d06d9963852e7818157994c645c640239cb5137

I made these patches against 2.6.24-rc4 with recent pciehp regression fixes applied.

Great, thanks a lot. I just pushed these into linux-next, please verify things when you get a chance.

Thanks,
Jesse