[linux-2.6-block.git] / arch / powerpc / platforms / pseries / msi.c

/*
 * Copyright 2006 Jake Moilanen <moilanen@austin.ibm.com>, IBM Corp.
 * Copyright 2006-2007 Michael Ellerman, IBM Corp.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License
 * as published by the Free Software Foundation; version 2 of the
 * License.
 *
 */

#include <linux/device.h>
#include <linux/irq.h>
#include <linux/msi.h>

#include <asm/rtas.h>
#include <asm/hw_irq.h>
#include <asm/ppc-pci.h>
#include <asm/machdep.h>

#include "pseries.h"

static int query_token, change_token;

#define RTAS_QUERY_FN		0
#define RTAS_CHANGE_FN		1
#define RTAS_RESET_FN		2
#define RTAS_CHANGE_MSI_FN	3
#define RTAS_CHANGE_MSIX_FN	4
#define RTAS_CHANGE_32MSI_FN	5

/* RTAS Helpers */

static int rtas_change_msi(struct pci_dn *pdn, u32 func, u32 num_irqs)
{
	u32 addr, seq_num, rtas_ret[3];
	unsigned long buid;
	int rc;

	addr = rtas_config_addr(pdn->busno, pdn->devfn, 0);
	buid = pdn->phb->buid;

	seq_num = 1;
	do {
		if (func == RTAS_CHANGE_MSI_FN || func == RTAS_CHANGE_MSIX_FN ||
		    func == RTAS_CHANGE_32MSI_FN)
			rc = rtas_call(change_token, 6, 4, rtas_ret, addr,
					BUID_HI(buid), BUID_LO(buid),
					func, num_irqs, seq_num);
		else
			rc = rtas_call(change_token, 6, 3, rtas_ret, addr,
					BUID_HI(buid), BUID_LO(buid),
					func, num_irqs, seq_num);

		seq_num = rtas_ret[1];
	} while (rtas_busy_delay(rc));

	/*
	 * If the RTAS call succeeded, return the number of irqs allocated.
	 * If not, make sure we return a negative error code.
	 */
	if (rc == 0)
		rc = rtas_ret[0];
	else if (rc > 0)
		rc = -rc;

	pr_debug("rtas_msi: ibm,change_msi(func=%d,num=%d), got %d rc = %d\n",
		 func, num_irqs, rtas_ret[0], rc);

	return rc;
}

static void rtas_disable_msi(struct pci_dev *pdev)
{
	struct pci_dn *pdn;

	pdn = pci_get_pdn(pdev);
	if (!pdn)
		return;

	/*
	 * disabling MSI with the explicit interface also disables MSI-X
	 */
	if (rtas_change_msi(pdn, RTAS_CHANGE_MSI_FN, 0) != 0) {
		/* 
		 * may have failed because explicit interface is not
		 * present
		 */
		if (rtas_change_msi(pdn, RTAS_CHANGE_FN, 0) != 0) {
			pr_debug("rtas_msi: Setting MSIs to 0 failed!\n");
		}
	}
}

static int rtas_query_irq_number(struct pci_dn *pdn, int offset)
{
	u32 addr, rtas_ret[2];
	unsigned long buid;
	int rc;

	addr = rtas_config_addr(pdn->busno, pdn->devfn, 0);
	buid = pdn->phb->buid;

	do {
		rc = rtas_call(query_token, 4, 3, rtas_ret, addr,
			       BUID_HI(buid), BUID_LO(buid), offset);
	} while (rtas_busy_delay(rc));

	if (rc) {
		pr_debug("rtas_msi: error (%d) querying source number\n", rc);
		return rc;
	}

	return rtas_ret[0];
}

static void rtas_teardown_msi_irqs(struct pci_dev *pdev)
{
	struct msi_desc *entry;

	for_each_pci_msi_entry(entry, pdev) {
		if (entry->irq == NO_IRQ)
			continue;

		irq_set_msi_desc(entry->irq, NULL);
		irq_dispose_mapping(entry->irq);
	}

	rtas_disable_msi(pdev);
}

static int check_req(struct pci_dev *pdev, int nvec, char *prop_name)
{
	struct device_node *dn;
	struct pci_dn *pdn;
	const __be32 *p;
	u32 req_msi;

	pdn = pci_get_pdn(pdev);
	if (!pdn)
		return -ENODEV;

	dn = pdn->node;

	p = of_get_property(dn, prop_name, NULL);
	if (!p) {
		pr_debug("rtas_msi: No %s on %s\n", prop_name, dn->full_name);
		return -ENOENT;
	}

	req_msi = be32_to_cpup(p);
	if (req_msi < nvec) {
		pr_debug("rtas_msi: %s requests < %d MSIs\n", prop_name, nvec);

		if (req_msi == 0) /* Be paranoid */
			return -ENOSPC;

		return req_msi;
	}

	return 0;
}

static int check_req_msi(struct pci_dev *pdev, int nvec)
{
	return check_req(pdev, nvec, "ibm,req#msi");
}

static int check_req_msix(struct pci_dev *pdev, int nvec)
{
	return check_req(pdev, nvec, "ibm,req#msi-x");
}

/* Quota calculation */

static struct device_node *find_pe_total_msi(struct pci_dev *dev, int *total)
{
	struct device_node *dn;
	const __be32 *p;

	dn = of_node_get(pci_device_to_OF_node(dev));
	while (dn) {
		p = of_get_property(dn, "ibm,pe-total-#msi", NULL);
		if (p) {
			pr_debug("rtas_msi: found prop on dn %s\n",
				dn->full_name);
			*total = be32_to_cpup(p);
			return dn;
		}

		dn = of_get_next_parent(dn);
	}

	return NULL;
}

static struct device_node *find_pe_dn(struct pci_dev *dev, int *total)
{
	struct device_node *dn;
	struct pci_dn *pdn;
	struct eeh_dev *edev;

	/* Found our PE and assume 8 at that point. */

	dn = pci_device_to_OF_node(dev);
	if (!dn)
		return NULL;

	/* Get the top level device in the PE */
	edev = pdn_to_eeh_dev(PCI_DN(dn));
	if (edev->pe)
		edev = list_first_entry(&edev->pe->edevs, struct eeh_dev, list);
	pdn = eeh_dev_to_pdn(edev);
	dn = pdn ? pdn->node : NULL;
	if (!dn)
		return NULL;

	/* We actually want the parent */
	dn = of_get_parent(dn);
	if (!dn)
		return NULL;

	/* Hardcode of 8 for old firmwares */
	*total = 8;
	pr_debug("rtas_msi: using PE dn %s\n", dn->full_name);

	return dn;
}

struct msi_counts {
	struct device_node *requestor;
	int num_devices;
	int request;
	int quota;
	int spare;
	int over_quota;
};

static void *count_non_bridge_devices(struct device_node *dn, void *data)
{
	struct msi_counts *counts = data;
	const __be32 *p;
	u32 class;

	pr_debug("rtas_msi: counting %s\n", dn->full_name);

	p = of_get_property(dn, "class-code", NULL);
	class = p ? be32_to_cpup(p) : 0;

	if ((class >> 8) != PCI_CLASS_BRIDGE_PCI)
		counts->num_devices++;

	return NULL;
}

static void *count_spare_msis(struct device_node *dn, void *data)
{
	struct msi_counts *counts = data;
	const __be32 *p;
	int req;

	if (dn == counts->requestor)
		req = counts->request;
	else {
		/* We don't know if a driver will try to use MSI or MSI-X,
		 * so we just have to punt and use the larger of the two. */
		req = 0;
		p = of_get_property(dn, "ibm,req#msi", NULL);
		if (p)
			req = be32_to_cpup(p);

		p = of_get_property(dn, "ibm,req#msi-x", NULL);
		if (p)
			req = max(req, (int)be32_to_cpup(p));
	}

	if (req < counts->quota)
		counts->spare += counts->quota - req;
	else if (req > counts->quota)
		counts->over_quota++;

	return NULL;
}

static int msi_quota_for_device(struct pci_dev *dev, int request)
{
	struct device_node *pe_dn;
	struct msi_counts counts;
	int total;

	pr_debug("rtas_msi: calc quota for %s, request %d\n", pci_name(dev),
		  request);

	pe_dn = find_pe_total_msi(dev, &total);
	if (!pe_dn)
		pe_dn = find_pe_dn(dev, &total);

	if (!pe_dn) {
		pr_err("rtas_msi: couldn't find PE for %s\n", pci_name(dev));
		goto out;
	}

	pr_debug("rtas_msi: found PE %s\n", pe_dn->full_name);

	memset(&counts, 0, sizeof(struct msi_counts));

	/* Work out how many devices we have below this PE */
	traverse_pci_devices(pe_dn, count_non_bridge_devices, &counts);

	if (counts.num_devices == 0) {
		pr_err("rtas_msi: found 0 devices under PE for %s\n",
			pci_name(dev));
		goto out;
	}

	counts.quota = total / counts.num_devices;
	if (request <= counts.quota)
		goto out;

	/* else, we have some more calculating to do */
	counts.requestor = pci_device_to_OF_node(dev);
	counts.request = request;
	traverse_pci_devices(pe_dn, count_spare_msis, &counts);

	/* If the quota isn't an integer multiple of the total, we can
	 * use the remainder as spare MSIs for anyone that wants them. */
	counts.spare += total % counts.num_devices;

	/* Divide any spare by the number of over-quota requestors */
	if (counts.over_quota)
		counts.quota += counts.spare / counts.over_quota;

	/* And finally clamp the request to the possibly adjusted quota */
	request = min(counts.quota, request);

	pr_debug("rtas_msi: request clamped to quota %d\n", request);
out:
	of_node_put(pe_dn);

	return request;
}

static int check_msix_entries(struct pci_dev *pdev)
{
	struct msi_desc *entry;
	int expected;

	/* There's no way for us to express to firmware that we want
	 * a discontiguous, or non-zero based, range of MSI-X entries.
	 * So we must reject such requests. */

	expected = 0;
	for_each_pci_msi_entry(entry, pdev) {
		if (entry->msi_attrib.entry_nr != expected) {
			pr_debug("rtas_msi: bad MSI-X entries.\n");
			return -EINVAL;
		}
		expected++;
	}

	return 0;
}

static void rtas_hack_32bit_msi_gen2(struct pci_dev *pdev)
{
	u32 addr_hi, addr_lo;

	/*
	 * We should only get in here for IODA1 configs. This is based on the
	 * fact that we using RTAS for MSIs, we don't have the 32 bit MSI RTAS
	 * support, and we are in a PCIe Gen2 slot.
	 */
	dev_info(&pdev->dev,
		 "rtas_msi: No 32 bit MSI firmware support, forcing 32 bit MSI\n");
	pci_read_config_dword(pdev, pdev->msi_cap + PCI_MSI_ADDRESS_HI, &addr_hi);
	addr_lo = 0xffff0000 | ((addr_hi >> (48 - 32)) << 4);
	pci_write_config_dword(pdev, pdev->msi_cap + PCI_MSI_ADDRESS_LO, addr_lo);
	pci_write_config_dword(pdev, pdev->msi_cap + PCI_MSI_ADDRESS_HI, 0);
}

static int rtas_setup_msi_irqs(struct pci_dev *pdev, int nvec_in, int type)
{
	struct pci_dn *pdn;
	int hwirq, virq, i, quota, rc;
	struct msi_desc *entry;
	struct msi_msg msg;
	int nvec = nvec_in;
	int use_32bit_msi_hack = 0;

	if (type == PCI_CAP_ID_MSIX)
		rc = check_req_msix(pdev, nvec);
	else
		rc = check_req_msi(pdev, nvec);

	if (rc)
		return rc;

	quota = msi_quota_for_device(pdev, nvec);

	if (quota && quota < nvec)
		return quota;

	if (type == PCI_CAP_ID_MSIX && check_msix_entries(pdev))
		return -EINVAL;

	/*
	 * Firmware currently refuse any non power of two allocation
	 * so we round up if the quota will allow it.
	 */
	if (type == PCI_CAP_ID_MSIX) {
		int m = roundup_pow_of_two(nvec);
		quota = msi_quota_for_device(pdev, m);

		if (quota >= m)
			nvec = m;
	}

	pdn = pci_get_pdn(pdev);

	/*
	 * Try the new more explicit firmware interface, if that fails fall
	 * back to the old interface. The old interface is known to never
	 * return MSI-Xs.
	 */
again:
	if (type == PCI_CAP_ID_MSI) {
		if (pdev->no_64bit_msi) {
			rc = rtas_change_msi(pdn, RTAS_CHANGE_32MSI_FN, nvec);
			if (rc < 0) {
				/*
				 * We only want to run the 32 bit MSI hack below if
				 * the max bus speed is Gen2 speed
				 */
				if (pdev->bus->max_bus_speed != PCIE_SPEED_5_0GT)
					return rc;

				use_32bit_msi_hack = 1;
			}
		} else
			rc = -1;

		if (rc < 0)
			rc = rtas_change_msi(pdn, RTAS_CHANGE_MSI_FN, nvec);

		if (rc < 0) {
			pr_debug("rtas_msi: trying the old firmware call.\n");
			rc = rtas_change_msi(pdn, RTAS_CHANGE_FN, nvec);
		}

		if (use_32bit_msi_hack && rc > 0)
			rtas_hack_32bit_msi_gen2(pdev);
	} else
		rc = rtas_change_msi(pdn, RTAS_CHANGE_MSIX_FN, nvec);

	if (rc != nvec) {
		if (nvec != nvec_in) {
			nvec = nvec_in;
			goto again;
		}
		pr_debug("rtas_msi: rtas_change_msi() failed\n");
		return rc;
	}

	i = 0;
	for_each_pci_msi_entry(entry, pdev) {
		hwirq = rtas_query_irq_number(pdn, i++);
		if (hwirq < 0) {
			pr_debug("rtas_msi: error (%d) getting hwirq\n", rc);
			return hwirq;
		}

		virq = irq_create_mapping(NULL, hwirq);

		if (virq == NO_IRQ) {
			pr_debug("rtas_msi: Failed mapping hwirq %d\n", hwirq);
			return -ENOSPC;
		}

		dev_dbg(&pdev->dev, "rtas_msi: allocated virq %d\n", virq);
		irq_set_msi_desc(virq, entry);

		/* Read config space back so we can restore after reset */
		__pci_read_msi_msg(entry, &msg);
		entry->msg = msg;
	}

	return 0;
}

static void rtas_msi_pci_irq_fixup(struct pci_dev *pdev)
{
	/* No LSI -> leave MSIs (if any) configured */
	if (pdev->irq == NO_IRQ) {
		dev_dbg(&pdev->dev, "rtas_msi: no LSI, nothing to do.\n");
		return;
	}

	/* No MSI -> MSIs can't have been assigned by fw, leave LSI */
	if (check_req_msi(pdev, 1) && check_req_msix(pdev, 1)) {
		dev_dbg(&pdev->dev, "rtas_msi: no req#msi/x, nothing to do.\n");
		return;
	}

	dev_dbg(&pdev->dev, "rtas_msi: disabling existing MSI.\n");
	rtas_disable_msi(pdev);
}

static int rtas_msi_init(void)
{
	struct pci_controller *phb;

	query_token  = rtas_token("ibm,query-interrupt-source-number");
	change_token = rtas_token("ibm,change-msi");

	if ((query_token == RTAS_UNKNOWN_SERVICE) ||
			(change_token == RTAS_UNKNOWN_SERVICE)) {
		pr_debug("rtas_msi: no RTAS tokens, no MSI support.\n");
		return -1;
	}

	pr_debug("rtas_msi: Registering RTAS MSI callbacks.\n");

	WARN_ON(pseries_pci_controller_ops.setup_msi_irqs);
	pseries_pci_controller_ops.setup_msi_irqs = rtas_setup_msi_irqs;
	pseries_pci_controller_ops.teardown_msi_irqs = rtas_teardown_msi_irqs;

	list_for_each_entry(phb, &hose_list, list_node) {
		WARN_ON(phb->controller_ops.setup_msi_irqs);
		phb->controller_ops.setup_msi_irqs = rtas_setup_msi_irqs;
		phb->controller_ops.teardown_msi_irqs = rtas_teardown_msi_irqs;
	}

	WARN_ON(ppc_md.pci_irq_fixup);
	ppc_md.pci_irq_fixup = rtas_msi_pci_irq_fixup;

	return 0;
}
machine_arch_initcall(pseries, rtas_msi_init);
Commit	Line	Data
85f2bf9f ME	1	/*
	2	* Copyright 2006 Jake Moilanen <moilanen@austin.ibm.com>, IBM Corp.
	3	* Copyright 2006-2007 Michael Ellerman, IBM Corp.
	4	*
	5	* This program is free software; you can redistribute it and/or
	6	* modify it under the terms of the GNU General Public License
	7	* as published by the Free Software Foundation; version 2 of the
	8	* License.
	9	*
	10	*/
	11
	12	#include <linux/device.h>
	13	#include <linux/irq.h>
	14	#include <linux/msi.h>
	15
	16	#include <asm/rtas.h>
	17	#include <asm/hw_irq.h>
	18	#include <asm/ppc-pci.h>
8e83e905	19	#include <asm/machdep.h>
85f2bf9f	20
1d14b875 DA	21	#include "pseries.h"
1d14b875 DA	22
85f2bf9f ME	23	static int query_token, change_token;
	24
	25	#define RTAS_QUERY_FN 0
	26	#define RTAS_CHANGE_FN 1
	27	#define RTAS_RESET_FN 2
	28	#define RTAS_CHANGE_MSI_FN 3
	29	#define RTAS_CHANGE_MSIX_FN 4
e61133dd	30	#define RTAS_CHANGE_32MSI_FN 5
85f2bf9f	31
85f2bf9f ME	32	/* RTAS Helpers */
	33
	34	static int rtas_change_msi(struct pci_dn *pdn, u32 func, u32 num_irqs)
	35	{
	36	u32 addr, seq_num, rtas_ret[3];
	37	unsigned long buid;
	38	int rc;
	39
	40	addr = rtas_config_addr(pdn->busno, pdn->devfn, 0);
	41	buid = pdn->phb->buid;
	42
	43	seq_num = 1;
	44	do {
e61133dd BK	45	if (func == RTAS_CHANGE_MSI_FN \|\| func == RTAS_CHANGE_MSIX_FN \|\|
e61133dd BK	46	func == RTAS_CHANGE_32MSI_FN)
85f2bf9f ME	47	rc = rtas_call(change_token, 6, 4, rtas_ret, addr,
	48	BUID_HI(buid), BUID_LO(buid),
	49	func, num_irqs, seq_num);
	50	else
	51	rc = rtas_call(change_token, 6, 3, rtas_ret, addr,
	52	BUID_HI(buid), BUID_LO(buid),
	53	func, num_irqs, seq_num);
	54
	55	seq_num = rtas_ret[1];
	56	} while (rtas_busy_delay(rc));
	57
d385366a	58	/*
6071ed04 ME	59	* If the RTAS call succeeded, return the number of irqs allocated.
6071ed04 ME	60	* If not, make sure we return a negative error code.
d385366a	61	*/
6071ed04 ME	62	if (rc == 0)
	63	rc = rtas_ret[0];
	64	else if (rc > 0)
	65	rc = -rc;
85f2bf9f	66
d385366a ME	67	pr_debug("rtas_msi: ibm,change_msi(func=%d,num=%d), got %d rc = %d\n",
d385366a ME	68	func, num_irqs, rtas_ret[0], rc);
85f2bf9f ME	69
	70	return rc;
	71	}
	72
	73	static void rtas_disable_msi(struct pci_dev *pdev)
	74	{
	75	struct pci_dn *pdn;
	76
b72c1f65	77	pdn = pci_get_pdn(pdev);
85f2bf9f ME	78	if (!pdn)
	79	return;
	80
964a2996 NA	81	/*
	82	* disabling MSI with the explicit interface also disables MSI-X
	83	*/
	84	if (rtas_change_msi(pdn, RTAS_CHANGE_MSI_FN, 0) != 0) {
	85	/*
	86	* may have failed because explicit interface is not
	87	* present
	88	*/
	89	if (rtas_change_msi(pdn, RTAS_CHANGE_FN, 0) != 0) {
	90	pr_debug("rtas_msi: Setting MSIs to 0 failed!\n");
	91	}
	92	}
85f2bf9f ME	93	}
	94
	95	static int rtas_query_irq_number(struct pci_dn *pdn, int offset)
	96	{
	97	u32 addr, rtas_ret[2];
	98	unsigned long buid;
	99	int rc;
	100
	101	addr = rtas_config_addr(pdn->busno, pdn->devfn, 0);
	102	buid = pdn->phb->buid;
	103
	104	do {
	105	rc = rtas_call(query_token, 4, 3, rtas_ret, addr,
	106	BUID_HI(buid), BUID_LO(buid), offset);
	107	} while (rtas_busy_delay(rc));
	108
	109	if (rc) {
	110	pr_debug("rtas_msi: error (%d) querying source number\n", rc);
	111	return rc;
	112	}
	113
	114	return rtas_ret[0];
	115	}
	116
	117	static void rtas_teardown_msi_irqs(struct pci_dev *pdev)
	118	{
	119	struct msi_desc *entry;
	120
2921d179	121	for_each_pci_msi_entry(entry, pdev) {
85f2bf9f ME	122	if (entry->irq == NO_IRQ)
	123	continue;
	124
ec775d0e	125	irq_set_msi_desc(entry->irq, NULL);
85f2bf9f ME	126	irq_dispose_mapping(entry->irq);
	127	}
	128
	129	rtas_disable_msi(pdev);
	130	}
	131
3a51c0cb	132	static int check_req(struct pci_dev pdev, int nvec, char prop_name)
85f2bf9f ME	133	{
	134	struct device_node *dn;
	135	struct pci_dn *pdn;
8d153155 AB	136	const __be32 *p;
8d153155 AB	137	u32 req_msi;
85f2bf9f	138
b72c1f65	139	pdn = pci_get_pdn(pdev);
85f2bf9f ME	140	if (!pdn)
	141	return -ENODEV;
	142
	143	dn = pdn->node;
	144
8d153155 AB	145	p = of_get_property(dn, prop_name, NULL);
8d153155 AB	146	if (!p) {
3a51c0cb	147	pr_debug("rtas_msi: No %s on %s\n", prop_name, dn->full_name);
85f2bf9f ME	148	return -ENOENT;
	149	}
	150
8d153155 AB	151	req_msi = be32_to_cpup(p);
8d153155 AB	152	if (req_msi < nvec) {
3a51c0cb	153	pr_debug("rtas_msi: %s requests < %d MSIs\n", prop_name, nvec);
d523cc37	154
8d153155	155	if (req_msi == 0) /* Be paranoid */
d523cc37 ME	156	return -ENOSPC;
d523cc37 ME	157
8d153155	158	return req_msi;
85f2bf9f ME	159	}
	160
	161	return 0;
	162	}
	163
3a51c0cb ME	164	static int check_req_msi(struct pci_dev *pdev, int nvec)
	165	{
	166	return check_req(pdev, nvec, "ibm,req#msi");
	167	}
	168
	169	static int check_req_msix(struct pci_dev *pdev, int nvec)
	170	{
	171	return check_req(pdev, nvec, "ibm,req#msi-x");
	172	}
	173
448e2ca0 ME	174	/* Quota calculation */
	175
	176	static struct device_node find_pe_total_msi(struct pci_dev dev, int *total)
	177	{
	178	struct device_node *dn;
8d153155	179	const __be32 *p;
448e2ca0 ME	180
	181	dn = of_node_get(pci_device_to_OF_node(dev));
	182	while (dn) {
	183	p = of_get_property(dn, "ibm,pe-total-#msi", NULL);
	184	if (p) {
	185	pr_debug("rtas_msi: found prop on dn %s\n",
	186	dn->full_name);
8d153155	187	*total = be32_to_cpup(p);
448e2ca0 ME	188	return dn;
	189	}
	190
	191	dn = of_get_next_parent(dn);
	192	}
	193
	194	return NULL;
	195	}
	196
	197	static struct device_node find_pe_dn(struct pci_dev dev, int *total)
	198	{
	199	struct device_node *dn;
c6406d8f	200	struct pci_dn *pdn;
66523d9f	201	struct eeh_dev *edev;
448e2ca0 ME	202
	203	/* Found our PE and assume 8 at that point. */
	204
	205	dn = pci_device_to_OF_node(dev);
	206	if (!dn)
	207	return NULL;
	208
66523d9f	209	/* Get the top level device in the PE */
c6406d8f	210	edev = pdn_to_eeh_dev(PCI_DN(dn));
bb461882 AK	211	if (edev->pe)
bb461882 AK	212	edev = list_first_entry(&edev->pe->edevs, struct eeh_dev, list);
c6406d8f GS	213	pdn = eeh_dev_to_pdn(edev);
c6406d8f GS	214	dn = pdn ? pdn->node : NULL;
448e2ca0 ME	215	if (!dn)
	216	return NULL;
	217
	218	/* We actually want the parent */
	219	dn = of_get_parent(dn);
	220	if (!dn)
	221	return NULL;
	222
	223	/* Hardcode of 8 for old firmwares */
	224	*total = 8;
	225	pr_debug("rtas_msi: using PE dn %s\n", dn->full_name);
	226
	227	return dn;
	228	}
	229
	230	struct msi_counts {
	231	struct device_node *requestor;
	232	int num_devices;
	233	int request;
	234	int quota;
	235	int spare;
	236	int over_quota;
	237	};
	238
	239	static void count_non_bridge_devices(struct device_node dn, void *data)
	240	{
	241	struct msi_counts *counts = data;
8d153155	242	const __be32 *p;
448e2ca0 ME	243	u32 class;
	244
	245	pr_debug("rtas_msi: counting %s\n", dn->full_name);
	246
	247	p = of_get_property(dn, "class-code", NULL);
8d153155	248	class = p ? be32_to_cpup(p) : 0;
448e2ca0 ME	249
	250	if ((class >> 8) != PCI_CLASS_BRIDGE_PCI)
	251	counts->num_devices++;
	252
	253	return NULL;
	254	}
	255
	256	static void count_spare_msis(struct device_node dn, void *data)
	257	{
	258	struct msi_counts *counts = data;
8d153155	259	const __be32 *p;
448e2ca0 ME	260	int req;
	261
	262	if (dn == counts->requestor)
	263	req = counts->request;
	264	else {
	265	/* We don't know if a driver will try to use MSI or MSI-X,
	266	* so we just have to punt and use the larger of the two. */
	267	req = 0;
	268	p = of_get_property(dn, "ibm,req#msi", NULL);
	269	if (p)
8d153155	270	req = be32_to_cpup(p);
448e2ca0 ME	271
	272	p = of_get_property(dn, "ibm,req#msi-x", NULL);
	273	if (p)
8d153155	274	req = max(req, (int)be32_to_cpup(p));
448e2ca0 ME	275	}
	276
	277	if (req < counts->quota)
	278	counts->spare += counts->quota - req;
	279	else if (req > counts->quota)
	280	counts->over_quota++;
	281
	282	return NULL;
	283	}
	284
	285	static int msi_quota_for_device(struct pci_dev *dev, int request)
	286	{
	287	struct device_node *pe_dn;
	288	struct msi_counts counts;
	289	int total;
	290
	291	pr_debug("rtas_msi: calc quota for %s, request %d\n", pci_name(dev),
	292	request);
	293
	294	pe_dn = find_pe_total_msi(dev, &total);
	295	if (!pe_dn)
	296	pe_dn = find_pe_dn(dev, &total);
	297
	298	if (!pe_dn) {
	299	pr_err("rtas_msi: couldn't find PE for %s\n", pci_name(dev));
	300	goto out;
	301	}
	302
	303	pr_debug("rtas_msi: found PE %s\n", pe_dn->full_name);
	304
	305	memset(&counts, 0, sizeof(struct msi_counts));
	306
	307	/* Work out how many devices we have below this PE */
	308	traverse_pci_devices(pe_dn, count_non_bridge_devices, &counts);
	309
	310	if (counts.num_devices == 0) {
	311	pr_err("rtas_msi: found 0 devices under PE for %s\n",
	312	pci_name(dev));
	313	goto out;
	314	}
	315
	316	counts.quota = total / counts.num_devices;
	317	if (request <= counts.quota)
	318	goto out;
	319
	320	/* else, we have some more calculating to do */
	321	counts.requestor = pci_device_to_OF_node(dev);
	322	counts.request = request;
	323	traverse_pci_devices(pe_dn, count_spare_msis, &counts);
	324
	325	/* If the quota isn't an integer multiple of the total, we can
	326	* use the remainder as spare MSIs for anyone that wants them. */
	327	counts.spare += total % counts.num_devices;
	328
	329	/* Divide any spare by the number of over-quota requestors */
	330	if (counts.over_quota)
	331	counts.quota += counts.spare / counts.over_quota;
	332
	333	/* And finally clamp the request to the possibly adjusted quota */
	334	request = min(counts.quota, request);
	335
	336	pr_debug("rtas_msi: request clamped to quota %d\n", request);
	337	out:
	338	of_node_put(pe_dn);
339
340	return request;
341	}
342
94afa5a5 ME	343	static int check_msix_entries(struct pci_dev *pdev)
	344	{
	345	struct msi_desc *entry;
	346	int expected;
	347
	348	/* There's no way for us to express to firmware that we want
	349	* a discontiguous, or non-zero based, range of MSI-X entries.
	350	* So we must reject such requests. */
	351
	352	expected = 0;
2921d179	353	for_each_pci_msi_entry(entry, pdev) {
94afa5a5 ME	354	if (entry->msi_attrib.entry_nr != expected) {
	355	pr_debug("rtas_msi: bad MSI-X entries.\n");
	356	return -EINVAL;
	357	}
	358	expected++;
	359	}
	360
	361	return 0;
	362	}
	363
f1dd1531 BK	364	static void rtas_hack_32bit_msi_gen2(struct pci_dev *pdev)
	365	{
	366	u32 addr_hi, addr_lo;
	367
	368	/*
	369	* We should only get in here for IODA1 configs. This is based on the
	370	* fact that we using RTAS for MSIs, we don't have the 32 bit MSI RTAS
	371	* support, and we are in a PCIe Gen2 slot.
	372	*/
	373	dev_info(&pdev->dev,
	374	"rtas_msi: No 32 bit MSI firmware support, forcing 32 bit MSI\n");
	375	pci_read_config_dword(pdev, pdev->msi_cap + PCI_MSI_ADDRESS_HI, &addr_hi);
	376	addr_lo = 0xffff0000 \| ((addr_hi >> (48 - 32)) << 4);
	377	pci_write_config_dword(pdev, pdev->msi_cap + PCI_MSI_ADDRESS_LO, addr_lo);
	378	pci_write_config_dword(pdev, pdev->msi_cap + PCI_MSI_ADDRESS_HI, 0);
	379	}
	380
752f5216	381	static int rtas_setup_msi_irqs(struct pci_dev *pdev, int nvec_in, int type)
85f2bf9f ME	382	{
85f2bf9f ME	383	struct pci_dn *pdn;
6b2fd7ef	384	int hwirq, virq, i, quota, rc;
85f2bf9f	385	struct msi_desc *entry;
1db3e890	386	struct msi_msg msg;
752f5216	387	int nvec = nvec_in;
f1dd1531	388	int use_32bit_msi_hack = 0;
85f2bf9f	389
6b2fd7ef AG	390	if (type == PCI_CAP_ID_MSIX)
	391	rc = check_req_msix(pdev, nvec);
	392	else
	393	rc = check_req_msi(pdev, nvec);
	394
	395	if (rc)
	396	return rc;
	397
	398	quota = msi_quota_for_device(pdev, nvec);
	399
	400	if (quota && quota < nvec)
	401	return quota;
85f2bf9f	402
94afa5a5 ME	403	if (type == PCI_CAP_ID_MSIX && check_msix_entries(pdev))
	404	return -EINVAL;
	405
752f5216 AB	406	/*
	407	* Firmware currently refuse any non power of two allocation
	408	* so we round up if the quota will allow it.
	409	*/
	410	if (type == PCI_CAP_ID_MSIX) {
	411	int m = roundup_pow_of_two(nvec);
6b2fd7ef	412	quota = msi_quota_for_device(pdev, m);
752f5216 AB	413
	414	if (quota >= m)
	415	nvec = m;
	416	}
	417
6b2fd7ef AG	418	pdn = pci_get_pdn(pdev);
6b2fd7ef AG	419
85f2bf9f ME	420	/*
	421	* Try the new more explicit firmware interface, if that fails fall
	422	* back to the old interface. The old interface is known to never
	423	* return MSI-Xs.
	424	*/
752f5216	425	again:
85f2bf9f	426	if (type == PCI_CAP_ID_MSI) {
415072a0	427	if (pdev->no_64bit_msi) {
e61133dd	428	rc = rtas_change_msi(pdn, RTAS_CHANGE_32MSI_FN, nvec);
f1dd1531 BK	429	if (rc < 0) {
	430	/*
	431	* We only want to run the 32 bit MSI hack below if
	432	* the max bus speed is Gen2 speed
	433	*/
	434	if (pdev->bus->max_bus_speed != PCIE_SPEED_5_0GT)
	435	return rc;
	436
	437	use_32bit_msi_hack = 1;
	438	}
	439	} else
	440	rc = -1;
	441
	442	if (rc < 0)
e61133dd	443	rc = rtas_change_msi(pdn, RTAS_CHANGE_MSI_FN, nvec);
85f2bf9f	444
f1dd1531	445	if (rc < 0) {
85f2bf9f ME	446	pr_debug("rtas_msi: trying the old firmware call.\n");
	447	rc = rtas_change_msi(pdn, RTAS_CHANGE_FN, nvec);
	448	}
f1dd1531 BK	449
	450	if (use_32bit_msi_hack && rc > 0)
	451	rtas_hack_32bit_msi_gen2(pdev);
85f2bf9f ME	452	} else
	453	rc = rtas_change_msi(pdn, RTAS_CHANGE_MSIX_FN, nvec);
	454
6071ed04	455	if (rc != nvec) {
752f5216 AB	456	if (nvec != nvec_in) {
	457	nvec = nvec_in;
	458	goto again;
	459	}
85f2bf9f	460	pr_debug("rtas_msi: rtas_change_msi() failed\n");
fcbe8090	461	return rc;
85f2bf9f ME	462	}
	463
	464	i = 0;
2921d179	465	for_each_pci_msi_entry(entry, pdev) {
e27ed698	466	hwirq = rtas_query_irq_number(pdn, i++);
85f2bf9f	467	if (hwirq < 0) {
85f2bf9f	468	pr_debug("rtas_msi: error (%d) getting hwirq\n", rc);
fcbe8090	469	return hwirq;
85f2bf9f ME	470	}
	471
	472	virq = irq_create_mapping(NULL, hwirq);
	473
	474	if (virq == NO_IRQ) {
	475	pr_debug("rtas_msi: Failed mapping hwirq %d\n", hwirq);
fcbe8090	476	return -ENOSPC;
85f2bf9f ME	477	}
	478
	479	dev_dbg(&pdev->dev, "rtas_msi: allocated virq %d\n", virq);
ec775d0e	480	irq_set_msi_desc(virq, entry);
1db3e890 ME	481
1db3e890 ME	482	/* Read config space back so we can restore after reset */
891d4a48	483	__pci_read_msi_msg(entry, &msg);
1db3e890	484	entry->msg = msg;
85f2bf9f ME	485	}
	486
	487	return 0;
85f2bf9f ME	488	}
	489
	490	static void rtas_msi_pci_irq_fixup(struct pci_dev *pdev)
	491	{
	492	/* No LSI -> leave MSIs (if any) configured */
	493	if (pdev->irq == NO_IRQ) {
	494	dev_dbg(&pdev->dev, "rtas_msi: no LSI, nothing to do.\n");
	495	return;
	496	}
	497
	498	/* No MSI -> MSIs can't have been assigned by fw, leave LSI */
649781f8 ME	499	if (check_req_msi(pdev, 1) && check_req_msix(pdev, 1)) {
649781f8 ME	500	dev_dbg(&pdev->dev, "rtas_msi: no req#msi/x, nothing to do.\n");
85f2bf9f ME	501	return;
	502	}
	503
	504	dev_dbg(&pdev->dev, "rtas_msi: disabling existing MSI.\n");
	505	rtas_disable_msi(pdev);
	506	}
	507
	508	static int rtas_msi_init(void)
	509	{
1d14b875 DA	510	struct pci_controller *phb;
1d14b875 DA	511
85f2bf9f ME	512	query_token = rtas_token("ibm,query-interrupt-source-number");
	513	change_token = rtas_token("ibm,change-msi");
	514
	515	if ((query_token == RTAS_UNKNOWN_SERVICE) \|\|
	516	(change_token == RTAS_UNKNOWN_SERVICE)) {
	517	pr_debug("rtas_msi: no RTAS tokens, no MSI support.\n");
	518	return -1;
	519	}
	520
	521	pr_debug("rtas_msi: Registering RTAS MSI callbacks.\n");
	522
1d14b875 DA	523	WARN_ON(pseries_pci_controller_ops.setup_msi_irqs);
	524	pseries_pci_controller_ops.setup_msi_irqs = rtas_setup_msi_irqs;
	525	pseries_pci_controller_ops.teardown_msi_irqs = rtas_teardown_msi_irqs;
	526
	527	list_for_each_entry(phb, &hose_list, list_node) {
	528	WARN_ON(phb->controller_ops.setup_msi_irqs);
	529	phb->controller_ops.setup_msi_irqs = rtas_setup_msi_irqs;
	530	phb->controller_ops.teardown_msi_irqs = rtas_teardown_msi_irqs;
	531	}
85f2bf9f ME	532
	533	WARN_ON(ppc_md.pci_irq_fixup);
	534	ppc_md.pci_irq_fixup = rtas_msi_pci_irq_fixup;
	535
	536	return 0;
	537	}
8e83e905	538	machine_arch_initcall(pseries, rtas_msi_init);