8 files changed, 194 insertions, 137 deletions

diff --git a/arch/s390/pci/pci.c b/arch/s390/pci/pci.c
index b0993e05affe..e7e6788d75a8 100644
--- a/arch/s390/pci/pci.c
+++ b/arch/s390/pci/pci.c
@@ -113,13 +113,16 @@ int zpci_register_ioat(struct zpci_dev *zdev, u8 dmaas,
 {
 	u64 req = ZPCI_CREATE_REQ(zdev->fh, dmaas, ZPCI_MOD_FC_REG_IOAT);
 	struct zpci_fib fib = {0};
-	u8 status;
+	u8 cc, status;
 
 	WARN_ON_ONCE(iota & 0x3fff);
 	fib.pba = base;
 	fib.pal = limit;
 	fib.iota = iota | ZPCI_IOTA_RTTO_FLAG;
-	return zpci_mod_fc(req, &fib, &status) ? -EIO : 0;
+	cc = zpci_mod_fc(req, &fib, &status);
+	if (cc)
+		zpci_dbg(3, "reg ioat fid:%x, cc:%d, status:%d\n", zdev->fid, cc, status);
+	return cc;
 }
 
 /* Modify PCI: Unregister I/O address translation parameters */
@@ -130,9 +133,9 @@ int zpci_unregister_ioat(struct zpci_dev *zdev, u8 dmaas)
 	u8 cc, status;
 
 	cc = zpci_mod_fc(req, &fib, &status);
-	if (cc == 3) /* Function already gone. */
-		cc = 0;
-	return cc ? -EIO : 0;
+	if (cc)
+		zpci_dbg(3, "unreg ioat fid:%x, cc:%d, status:%d\n", zdev->fid, cc, status);
+	return cc;
 }
 
 /* Modify PCI: Set PCI function measurement parameters */
@@ -560,9 +563,12 @@ static void zpci_cleanup_bus_resources(struct zpci_dev *zdev)
 
 int pcibios_add_device(struct pci_dev *pdev)
 {
+	struct zpci_dev *zdev = to_zpci(pdev);
 	struct resource *res;
 	int i;
 
+	/* The pdev has a reference to the zdev via its bus */
+	zpci_zdev_get(zdev);
 	if (pdev->is_physfn)
 		pdev->no_vf_scan = 1;
 
@@ -582,7 +588,10 @@ int pcibios_add_device(struct pci_dev *pdev)
 
 void pcibios_release_device(struct pci_dev *pdev)
 {
+	struct zpci_dev *zdev = to_zpci(pdev);
+
 	zpci_unmap_resources(pdev);
+	zpci_zdev_put(zdev);
 }
 
 int pcibios_enable_device(struct pci_dev *pdev, int mask)
@@ -653,32 +662,37 @@ void zpci_free_domain(int domain)
 
 int zpci_enable_device(struct zpci_dev *zdev)
 {
-	int rc;
-
-	rc = clp_enable_fh(zdev, ZPCI_NR_DMA_SPACES);
-	if (rc)
-		goto out;
-
-	rc = zpci_dma_init_device(zdev);
-	if (rc)
-		goto out_dma;
+	u32 fh = zdev->fh;
+	int rc = 0;
 
-	return 0;
-
-out_dma:
-	clp_disable_fh(zdev);
-out:
+	if (clp_enable_fh(zdev, &fh, ZPCI_NR_DMA_SPACES))
+		rc = -EIO;
+	else
+		zdev->fh = fh;
 	return rc;
 }
 
 int zpci_disable_device(struct zpci_dev *zdev)
 {
-	zpci_dma_exit_device(zdev);
-	/*
-	 * The zPCI function may already be disabled by the platform, this is
-	 * detected in clp_disable_fh() which becomes a no-op.
-	 */
-	return clp_disable_fh(zdev);
+	u32 fh = zdev->fh;
+	int cc, rc = 0;
+
+	cc = clp_disable_fh(zdev, &fh);
+	if (!cc) {
+		zdev->fh = fh;
+	} else if (cc == CLP_RC_SETPCIFN_ALRDY) {
+		pr_info("Disabling PCI function %08x had no effect as it was already disabled\n",
+			zdev->fid);
+		/* Function is already disabled - update handle */
+		rc = clp_refresh_fh(zdev->fid, &fh);
+		if (!rc) {
+			zdev->fh = fh;
+			rc = -EINVAL;
+		}
+	} else {
+		rc = -EIO;
+	}
+	return rc;
 }
 
 /**
@@ -788,6 +802,11 @@ int zpci_deconfigure_device(struct zpci_dev *zdev)
 	if (zdev->zbus->bus)
 		zpci_bus_remove_device(zdev, false);
 
+	if (zdev->dma_table) {
+		rc = zpci_dma_exit_device(zdev);
+		if (rc)
+			return rc;
+	}
 	if (zdev_enabled(zdev)) {
 		rc = zpci_disable_device(zdev);
 		if (rc)
@@ -811,6 +830,8 @@ void zpci_release_device(struct kref *kref)
 	if (zdev->zbus->bus)
 		zpci_bus_remove_device(zdev, false);
 
+	if (zdev->dma_table)
+		zpci_dma_exit_device(zdev);
 	if (zdev_enabled(zdev))
 		zpci_disable_device(zdev);
 
@@ -822,7 +843,8 @@ void zpci_release_device(struct kref *kref)
 	case ZPCI_FN_STATE_STANDBY:
 		if (zdev->has_hp_slot)
 			zpci_exit_slot(zdev);
-		zpci_cleanup_bus_resources(zdev);
+		if (zdev->has_resources)
+			zpci_cleanup_bus_resources(zdev);
 		zpci_bus_device_unregister(zdev);
 		zpci_destroy_iommu(zdev);
 		fallthrough;
@@ -886,7 +908,6 @@ static void zpci_mem_exit(void)
 }
 
 static unsigned int s390_pci_probe __initdata = 1;
-static unsigned int s390_pci_no_mio __initdata;
 unsigned int s390_pci_force_floating __initdata;
 static unsigned int s390_pci_initialized;
 
@@ -897,7 +918,7 @@ char * __init pcibios_setup(char *str)
 		return NULL;
 	}
 	if (!strcmp(str, "nomio")) {
-		s390_pci_no_mio = 1;
+		S390_lowcore.machine_flags &= ~MACHINE_FLAG_PCI_MIO;
 		return NULL;
 	}
 	if (!strcmp(str, "force_floating")) {
@@ -928,7 +949,7 @@ static int __init pci_base_init(void)
 		return 0;
 	}
 
-	if (test_facility(153) && !s390_pci_no_mio) {
+	if (MACHINE_HAS_PCI_MIO) {
 		static_branch_enable(&have_mio);
 		ctl_set_bit(2, 5);
 	}
diff --git a/arch/s390/pci/pci_bus.c b/arch/s390/pci/pci_bus.c
index 9629f9779c79..5d77acbd1c87 100644
--- a/arch/s390/pci/pci_bus.c
+++ b/arch/s390/pci/pci_bus.c
@@ -49,6 +49,11 @@ static int zpci_bus_prepare_device(struct zpci_dev *zdev)
 		rc = zpci_enable_device(zdev);
 		if (rc)
 			return rc;
+		rc = zpci_dma_init_device(zdev);
+		if (rc) {
+			zpci_disable_device(zdev);
+			return rc;
+		}
 	}
 
 	if (!zdev->has_resources) {
@@ -343,11 +348,11 @@ static int zpci_bus_add_device(struct zpci_bus *zbus, struct zpci_dev *zdev)
 {
 	int rc = -EINVAL;
 
-	zdev->zbus = zbus;
 	if (zbus->function[zdev->devfn]) {
 		pr_err("devfn %04x is already assigned\n", zdev->devfn);
 		return rc;
 	}
+	zdev->zbus = zbus;
 	zbus->function[zdev->devfn] = zdev;
 	zpci_nb_devices++;
 
@@ -367,6 +372,7 @@ static int zpci_bus_add_device(struct zpci_bus *zbus, struct zpci_dev *zdev)
 
 error:
 	zbus->function[zdev->devfn] = NULL;
+	zdev->zbus = NULL;
 	zpci_nb_devices--;
 	return rc;
 }
diff --git a/arch/s390/pci/pci_bus.h b/arch/s390/pci/pci_bus.h
index b877a97e6745..e359d2686178 100644
--- a/arch/s390/pci/pci_bus.h
+++ b/arch/s390/pci/pci_bus.h
@@ -22,6 +22,11 @@ static inline void zpci_zdev_put(struct zpci_dev *zdev)
 	kref_put(&zdev->kref, zpci_release_device);
 }
 
+static inline void zpci_zdev_get(struct zpci_dev *zdev)
+{
+	kref_get(&zdev->kref);
+}
+
 int zpci_alloc_domain(int domain);
 void zpci_free_domain(int domain);
 int zpci_setup_bus_resources(struct zpci_dev *zdev,
diff --git a/arch/s390/pci/pci_clp.c b/arch/s390/pci/pci_clp.c
index d3331596ddbe..51dc2215a2b7 100644
--- a/arch/s390/pci/pci_clp.c
+++ b/arch/s390/pci/pci_clp.c
@@ -212,17 +212,22 @@ out:
 	return rc;
 }
 
-static int clp_refresh_fh(u32 fid);
-/*
- * Enable/Disable a given PCI function and update its function handle if
- * necessary
+/**
+ * clp_set_pci_fn() - Execute a command on a PCI function
+ * @zdev: Function that will be affected
+ * @fh: Out parameter for updated function handle
+ * @nr_dma_as: DMA address space number
+ * @command: The command code to execute
+ *
+ * Returns: 0 on success, < 0 for Linux errors (e.g. -ENOMEM), and
+ * > 0 for non-success platform responses
  */
-static int clp_set_pci_fn(struct zpci_dev *zdev, u8 nr_dma_as, u8 command)
+static int clp_set_pci_fn(struct zpci_dev *zdev, u32 *fh, u8 nr_dma_as, u8 command)
 {
 	struct clp_req_rsp_set_pci *rrb;
 	int rc, retries = 100;
-	u32 fid = zdev->fid;
 
+	*fh = 0;
 	rrb = clp_alloc_block(GFP_KERNEL);
 	if (!rrb)
 		return -ENOMEM;
@@ -245,17 +250,13 @@ static int clp_set_pci_fn(struct zpci_dev *zdev, u8 nr_dma_as, u8 command)
 		}
 	} while (rrb->response.hdr.rsp == CLP_RC_SETPCIFN_BUSY);
 
-	if (rc || rrb->response.hdr.rsp != CLP_RC_OK) {
+	if (!rc && rrb->response.hdr.rsp == CLP_RC_OK) {
+		*fh = rrb->response.fh;
+	} else {
 		zpci_err("Set PCI FN:\n");
 		zpci_err_clp(rrb->response.hdr.rsp, rc);
-	}
-
-	if (!rc && rrb->response.hdr.rsp == CLP_RC_OK) {
-		zdev->fh = rrb->response.fh;
-	} else if (!rc && rrb->response.hdr.rsp == CLP_RC_SETPCIFN_ALRDY &&
-			rrb->response.fh == 0) {
-		/* Function is already in desired state - update handle */
-		rc = clp_refresh_fh(fid);
+		if (!rc)
+			rc = rrb->response.hdr.rsp;
 	}
 	clp_free_block(rrb);
 	return rc;
@@ -295,35 +296,62 @@ int clp_setup_writeback_mio(void)
 	return rc;
 }
 
-int clp_enable_fh(struct zpci_dev *zdev, u8 nr_dma_as)
+int clp_enable_fh(struct zpci_dev *zdev, u32 *fh, u8 nr_dma_as)
 {
 	int rc;
 
-	rc = clp_set_pci_fn(zdev, nr_dma_as, CLP_SET_ENABLE_PCI_FN);
-	zpci_dbg(3, "ena fid:%x, fh:%x, rc:%d\n", zdev->fid, zdev->fh, rc);
-	if (rc)
-		goto out;
-
-	if (zpci_use_mio(zdev)) {
-		rc = clp_set_pci_fn(zdev, nr_dma_as, CLP_SET_ENABLE_MIO);
+	rc = clp_set_pci_fn(zdev, fh, nr_dma_as, CLP_SET_ENABLE_PCI_FN);
+	zpci_dbg(3, "ena fid:%x, fh:%x, rc:%d\n", zdev->fid, *fh, rc);
+	if (!rc && zpci_use_mio(zdev)) {
+		rc = clp_set_pci_fn(zdev, fh, nr_dma_as, CLP_SET_ENABLE_MIO);
 		zpci_dbg(3, "ena mio fid:%x, fh:%x, rc:%d\n",
-				zdev->fid, zdev->fh, rc);
+				zdev->fid, *fh, rc);
 		if (rc)
-			clp_disable_fh(zdev);
+			clp_disable_fh(zdev, fh);
 	}
-out:
 	return rc;
 }
 
-int clp_disable_fh(struct zpci_dev *zdev)
+int clp_disable_fh(struct zpci_dev *zdev, u32 *fh)
 {
 	int rc;
 
 	if (!zdev_enabled(zdev))
 		return 0;
 
-	rc = clp_set_pci_fn(zdev, 0, CLP_SET_DISABLE_PCI_FN);
-	zpci_dbg(3, "dis fid:%x, fh:%x, rc:%d\n", zdev->fid, zdev->fh, rc);
+	rc = clp_set_pci_fn(zdev, fh, 0, CLP_SET_DISABLE_PCI_FN);
+	zpci_dbg(3, "dis fid:%x, fh:%x, rc:%d\n", zdev->fid, *fh, rc);
+	return rc;
+}
+
+static int clp_list_pci_req(struct clp_req_rsp_list_pci *rrb,
+			    u64 *resume_token, int *nentries)
+{
+	int rc;
+
+	memset(rrb, 0, sizeof(*rrb));
+	rrb->request.hdr.len = sizeof(rrb->request);
+	rrb->request.hdr.cmd = CLP_LIST_PCI;
+	/* store as many entries as possible */
+	rrb->response.hdr.len = CLP_BLK_SIZE - LIST_PCI_HDR_LEN;
+	rrb->request.resume_token = *resume_token;
+
+	/* Get PCI function handle list */
+	rc = clp_req(rrb, CLP_LPS_PCI);
+	if (rc || rrb->response.hdr.rsp != CLP_RC_OK) {
+		zpci_err("List PCI FN:\n");
+		zpci_err_clp(rrb->response.hdr.rsp, rc);
+		return -EIO;
+	}
+
+	update_uid_checking(rrb->response.uid_checking);
+	WARN_ON_ONCE(rrb->response.entry_size !=
+		sizeof(struct clp_fh_list_entry));
+
+	*nentries = (rrb->response.hdr.len - LIST_PCI_HDR_LEN) /
+		rrb->response.entry_size;
+	*resume_token = rrb->response.resume_token;
+
 	return rc;
 }
 
@@ -331,38 +359,40 @@ static int clp_list_pci(struct clp_req_rsp_list_pci *rrb, void *data,
 			void (*cb)(struct clp_fh_list_entry *, void *))
 {
 	u64 resume_token = 0;
-	int entries, i, rc;
+	int nentries, i, rc;
 
 	do {
-		memset(rrb, 0, sizeof(*rrb));
-		rrb->request.hdr.len = sizeof(rrb->request);
-		rrb->request.hdr.cmd = CLP_LIST_PCI;
-		/* store as many entries as possible */
-		rrb->response.hdr.len = CLP_BLK_SIZE - LIST_PCI_HDR_LEN;
-		rrb->request.resume_token = resume_token;
-
-		/* Get PCI function handle list */
-		rc = clp_req(rrb, CLP_LPS_PCI);
-		if (rc || rrb->response.hdr.rsp != CLP_RC_OK) {
-			zpci_err("List PCI FN:\n");
-			zpci_err_clp(rrb->response.hdr.rsp, rc);
-			rc = -EIO;
-			goto out;
-		}
+		rc = clp_list_pci_req(rrb, &resume_token, &nentries);
+		if (rc)
+			return rc;
+		for (i = 0; i < nentries; i++)
+			cb(&rrb->response.fh_list[i], data);
+	} while (resume_token);
 
-		update_uid_checking(rrb->response.uid_checking);
-		WARN_ON_ONCE(rrb->response.entry_size !=
-			sizeof(struct clp_fh_list_entry));
+	return rc;
+}
 
-		entries = (rrb->response.hdr.len - LIST_PCI_HDR_LEN) /
-			rrb->response.entry_size;
+static int clp_find_pci(struct clp_req_rsp_list_pci *rrb, u32 fid,
+			struct clp_fh_list_entry *entry)
+{
+	struct clp_fh_list_entry *fh_list;
+	u64 resume_token = 0;
+	int nentries, i, rc;
 
-		resume_token = rrb->response.resume_token;
-		for (i = 0; i < entries; i++)
-			cb(&rrb->response.fh_list[i], data);
+	do {
+		rc = clp_list_pci_req(rrb, &resume_token, &nentries);
+		if (rc)
+			return rc;
+		for (i = 0; i < nentries; i++) {
+			fh_list = rrb->response.fh_list;
+			if (fh_list[i].fid == fid) {
+				*entry = fh_list[i];
+				return 0;
+			}
+		}
 	} while (resume_token);
-out:
-	return rc;
+
+	return -ENODEV;
 }
 
 static void __clp_add(struct clp_fh_list_entry *entry, void *data)
@@ -392,67 +422,41 @@ int clp_scan_pci_devices(void)
 	return rc;
 }
 
-static void __clp_refresh_fh(struct clp_fh_list_entry *entry, void *data)
-{
-	struct zpci_dev *zdev;
-	u32 fid = *((u32 *)data);
-
-	if (!entry->vendor_id || fid != entry->fid)
-		return;
-
-	zdev = get_zdev_by_fid(fid);
-	if (!zdev)
-		return;
-
-	zdev->fh = entry->fh;
-}
-
 /*
- * Refresh the function handle of the function matching @fid
+ * Get the current function handle of the function matching @fid
  */
-static int clp_refresh_fh(u32 fid)
+int clp_refresh_fh(u32 fid, u32 *fh)
 {
 	struct clp_req_rsp_list_pci *rrb;
+	struct clp_fh_list_entry entry;
 	int rc;
 
 	rrb = clp_alloc_block(GFP_NOWAIT);
 	if (!rrb)
 		return -ENOMEM;
 
-	rc = clp_list_pci(rrb, &fid, __clp_refresh_fh);
+	rc = clp_find_pci(rrb, fid, &entry);
+	if (!rc)
+		*fh = entry.fh;
 
 	clp_free_block(rrb);
 	return rc;
 }
 
-struct clp_state_data {
-	u32 fid;
-	enum zpci_state state;
-};
-
-static void __clp_get_state(struct clp_fh_list_entry *entry, void *data)
-{
-	struct clp_state_data *sd = data;
-
-	if (entry->fid != sd->fid)
-		return;
-
-	sd->state = entry->config_state;
-}
-
 int clp_get_state(u32 fid, enum zpci_state *state)
 {
 	struct clp_req_rsp_list_pci *rrb;
-	struct clp_state_data sd = {fid, ZPCI_FN_STATE_RESERVED};
+	struct clp_fh_list_entry entry;
 	int rc;
 
+	*state = ZPCI_FN_STATE_RESERVED;
 	rrb = clp_alloc_block(GFP_ATOMIC);
 	if (!rrb)
 		return -ENOMEM;
 
-	rc = clp_list_pci(rrb, &sd, __clp_get_state);
+	rc = clp_find_pci(rrb, fid, &entry);
 	if (!rc)
-		*state = sd.state;
+		*state = entry.config_state;
 
 	clp_free_block(rrb);
 	return rc;
diff --git a/arch/s390/pci/pci_dma.c b/arch/s390/pci/pci_dma.c
index be48e5b5bfcf..93223bd110c3 100644
--- a/arch/s390/pci/pci_dma.c
+++ b/arch/s390/pci/pci_dma.c
@@ -591,10 +591,11 @@ int zpci_dma_init_device(struct zpci_dev *zdev)
 		}
 
 	}
-	rc = zpci_register_ioat(zdev, 0, zdev->start_dma, zdev->end_dma,
-				(u64) zdev->dma_table);
-	if (rc)
+	if (zpci_register_ioat(zdev, 0, zdev->start_dma, zdev->end_dma,
+			       (u64)zdev->dma_table)) {
+		rc = -EIO;
 		goto free_bitmap;
+	}
 
 	return 0;
 free_bitmap:
@@ -609,17 +610,25 @@ out:
 	return rc;
 }
 
-void zpci_dma_exit_device(struct zpci_dev *zdev)
+int zpci_dma_exit_device(struct zpci_dev *zdev)
 {
+	int cc = 0;
+
 	/*
 	 * At this point, if the device is part of an IOMMU domain, this would
 	 * be a strong hint towards a bug in the IOMMU API (common) code and/or
 	 * simultaneous access via IOMMU and DMA API. So let's issue a warning.
 	 */
 	WARN_ON(zdev->s390_domain);
-
-	if (zpci_unregister_ioat(zdev, 0))
-		return;
+	if (zdev_enabled(zdev))
+		cc = zpci_unregister_ioat(zdev, 0);
+	/*
+	 * cc == 3 indicates the function is gone already. This can happen
+	 * if the function was deconfigured/disabled suddenly and we have not
+	 * received a new handle yet.
+	 */
+	if (cc && cc != 3)
+		return -EIO;
 
 	dma_cleanup_tables(zdev->dma_table);
 	zdev->dma_table = NULL;
@@ -627,8 +636,8 @@ void zpci_dma_exit_device(struct zpci_dev *zdev)
 	zdev->iommu_bitmap = NULL;
 	vfree(zdev->lazy_bitmap);
 	zdev->lazy_bitmap = NULL;
-
 	zdev->next_bit = 0;
+	return 0;
 }
 
 static int __init dma_alloc_cpu_table_caches(void)
diff --git a/arch/s390/pci/pci_event.c b/arch/s390/pci/pci_event.c
index cd447b96b4b1..c856f80cb21b 100644
--- a/arch/s390/pci/pci_event.c
+++ b/arch/s390/pci/pci_event.c
@@ -84,7 +84,10 @@ static void zpci_event_hard_deconfigured(struct zpci_dev *zdev, u32 fh)
 	/* Even though the device is already gone we still
 	 * need to free zPCI resources as part of the disable.
 	 */
-	zpci_disable_device(zdev);
+	if (zdev->dma_table)
+		zpci_dma_exit_device(zdev);
+	if (zdev_enabled(zdev))
+		zpci_disable_device(zdev);
 	zdev->state = ZPCI_FN_STATE_STANDBY;
 }
 
diff --git a/arch/s390/pci/pci_irq.c b/arch/s390/pci/pci_irq.c
index 9c7de9089939..3823e159bf74 100644
--- a/arch/s390/pci/pci_irq.c
+++ b/arch/s390/pci/pci_irq.c
@@ -365,10 +365,6 @@ void arch_teardown_msi_irqs(struct pci_dev *pdev)
 	for_each_pci_msi_entry(msi, pdev) {
 		if (!msi->irq)
 			continue;
-		if (msi->msi_attrib.is_msix)
-			__pci_msix_desc_mask_irq(msi, 1);
-		else
-			__pci_msi_desc_mask_irq(msi, 1, 1);
 		irq_set_msi_desc(msi->irq, NULL);
 		irq_free_desc(msi->irq);
 		msi->msg.address_lo = 0;
diff --git a/arch/s390/pci/pci_sysfs.c b/arch/s390/pci/pci_sysfs.c
index 6e2450c2b9c1..335c281811c7 100644
--- a/arch/s390/pci/pci_sysfs.c
+++ b/arch/s390/pci/pci_sysfs.c
@@ -82,13 +82,26 @@ static ssize_t recover_store(struct device *dev, struct device_attribute *attr,
 	pci_lock_rescan_remove();
 	if (pci_dev_is_added(pdev)) {
 		pci_stop_and_remove_bus_device(pdev);
-		ret = zpci_disable_device(zdev);
-		if (ret)
-			goto out;
+		if (zdev->dma_table) {
+			ret = zpci_dma_exit_device(zdev);
+			if (ret)
+				goto out;
+		}
+
+		if (zdev_enabled(zdev)) {
+			ret = zpci_disable_device(zdev);
+			if (ret)
+				goto out;
+		}
 
 		ret = zpci_enable_device(zdev);
 		if (ret)
 			goto out;
+		ret = zpci_dma_init_device(zdev);
+		if (ret) {
+			zpci_disable_device(zdev);
+			goto out;
+		}
 		pci_rescan_bus(zdev->zbus->bus);
 	}
 out: