Skip to contentSkip to conversation

PCI: Add ACS quirk for Broadcom BCM5760X NIC

PATCH New Sasha Levin 1 participant 7 messages Jul 28, 2024

Subsystems (4)
lore
Share & tools
lore

Sasha Levin

author maintainer
#0 lore raw

PATCH 1/7 PCI: Add ACS quirk for Broadcom BCM5760X NIC

From: Ajit Khaparde <[email protected]>

[ Upstream commit 524e057b2d66b61f9b63b6db30467ab7b0bb4796 ]

The Broadcom BCM5760X NIC may be a multi-function device.

While it does not advertise an ACS capability, peer-to-peer transactions are not possible between the individual functions. So it is ok to treat them as fully isolated.

Add an ACS quirk for this device so the functions can be in independent IOMMU groups and attached individually to userspace applications using VFIO.

---
 drivers/pci/quirks.c | 4 ++++
 1 file changed, 4 insertions(+)
drivers/pci/quirks.c +4 -0
index 3bc7058404156..aef73bc36ee98 100644
@@ -4986,6 +4986,10 @@ static const struct pci_dev_acs_enabled {
 	{ PCI_VENDOR_ID_BROADCOM, 0x1750, pci_quirk_mf_endpoint_acs },
 	{ PCI_VENDOR_ID_BROADCOM, 0x1751, pci_quirk_mf_endpoint_acs },
 	{ PCI_VENDOR_ID_BROADCOM, 0x1752, pci_quirk_mf_endpoint_acs },
+	{ PCI_VENDOR_ID_BROADCOM, 0x1760, pci_quirk_mf_endpoint_acs },
+	{ PCI_VENDOR_ID_BROADCOM, 0x1761, pci_quirk_mf_endpoint_acs },
+	{ PCI_VENDOR_ID_BROADCOM, 0x1762, pci_quirk_mf_endpoint_acs },
+	{ PCI_VENDOR_ID_BROADCOM, 0x1763, pci_quirk_mf_endpoint_acs },
 	{ PCI_VENDOR_ID_BROADCOM, 0xD714, pci_quirk_brcm_acs },
 	/* Amazon Annapurna Labs */
 	{ PCI_VENDOR_ID_AMAZON_ANNAPURNA_LABS, 0x0031, pci_quirk_al_acs },
Signature · 1 line
2.43.0
author maintainer
#1 lore raw

PATCH 2/7 usb: dwc2: gadget: Don't write invalid mapped sg entries into dma_desc with iommu enabled

From: Peng Hongchi <[email protected]>

[ Upstream commit 1134289b6b93d73721340b66c310fd985385e8fa ]

When using dma_map_sg() to map the scatterlist with iommu enabled, the entries in the scatterlist can be mergerd into less but longer entries in the function __finalise_sg(). So that the number of valid mapped entries is actually smaller than ureq->num_reqs,and there are still some invalid entries in the scatterlist with dma_addr=0xffffffff and len=0. Writing these invalid sg entries into the dma_desc can cause a data transmission error.

The function dma_map_sg() returns the number of valid map entries and the return value is assigned to usb_request::num_mapped_sgs in function usb_gadget_map_request_by_dev(). So that just write valid mapped entries into dma_desc according to the usb_request::num_mapped_sgs, and set the IOC bit if it's the last valid mapped entry.

This patch poses no risk to no-iommu situation, cause ureq->num_mapped_sgs equals ureq->num_sgs while using dma_direct_map_sg() to map the scatterlist whith iommu disabled.

---
 drivers/usb/dwc2/gadget.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/usb/dwc2/gadget.c +2 -2
index b341bbc9f1dad..abc2271799e0a 100644
@@ -885,10 +885,10 @@ static void dwc2_gadget_config_nonisoc_xfer_ddma(struct dwc2_hsotg_ep *hs_ep,
 	}
 
 	/* DMA sg buffer */
-	for_each_sg(ureq->sg, sg, ureq->num_sgs, i) {
+	for_each_sg(ureq->sg, sg, ureq->num_mapped_sgs, i) {
 		dwc2_gadget_fill_nonisoc_xfer_ddma_one(hs_ep, &desc,
 			sg_dma_address(sg) + sg->offset, sg_dma_len(sg),
-			sg_is_last(sg));
+			(i == (ureq->num_mapped_sgs - 1)));
 		desc_count += hs_ep->desc_count;
 	}
 
Signature · 1 line
2.43.0
author maintainer
#2 lore raw

PATCH 3/7 usb: typec: ucsi: Fix null pointer dereference in trace

In reply to Sasha Levin

From: Abhishek Pandit-Subedi <[email protected]>

[ Upstream commit 99516f76db48e1a9d54cdfed63c1babcee4e71a5 ]

ucsi_register_altmode checks IS_ERR for the alt pointer and treats NULL as valid. When CONFIG_TYPEC_DP_ALTMODE is not enabled, ucsi_register_displayport returns NULL which causes a NULL pointer dereference in trace. Rather than return NULL, call typec_port_register_altmode to register DisplayPort alternate mode as a non-controllable mode when CONFIG_TYPEC_DP_ALTMODE is not enabled.

---
 drivers/usb/typec/ucsi/ucsi.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/usb/typec/ucsi/ucsi.h +1 -1
index de87d0b8319d0..179ad343f42f0 100644
@@ -446,7 +446,7 @@ ucsi_register_displayport(struct ucsi_connector *con,
 			  bool override, int offset,
 			  struct typec_altmode_desc *desc)
 {
-	return NULL;
+	return typec_port_register_altmode(con->port, desc);
 }
 
 static inline void
Signature · 1 line
2.43.0
author maintainer
#3 lore raw

PATCH 4/7 PCI/AER: Disable AER service on suspend

In reply to Sasha Levin

From: Kai-Heng Feng <[email protected]>

[ Upstream commit 5afc2f763edc5daae4722ee46fea4e627d01fa90 ]

If the link is powered off during suspend, electrical noise may cause errors that are logged via AER. If the AER interrupt is enabled and shares an IRQ with PME, that causes a spurious wakeup during suspend.

Disable the AER interrupt during suspend to prevent this. Clear error status before re-enabling IRQ interrupts during resume so we don't get an interrupt for errors that occurred during the suspend/resume process.

---
 drivers/pci/pcie/aer.c | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)
drivers/pci/pcie/aer.c +18 -0
index 6b5c9f7916fac..ee38ca162a7c8 100644
@@ -1407,6 +1407,22 @@ static int aer_probe(struct pcie_device *dev)
 	return 0;
 }
 
+static int aer_suspend(struct pcie_device *dev)
+{
+	struct aer_rpc *rpc = get_service_data(dev);
+
+	aer_disable_rootport(rpc);
+	return 0;
+}
+
+static int aer_resume(struct pcie_device *dev)
+{
+	struct aer_rpc *rpc = get_service_data(dev);
+
+	aer_enable_rootport(rpc);
+	return 0;
+}
+
 /**
  * aer_root_reset - reset link on Root Port
  * @dev: pointer to Root Port's pci_dev data structure
@@ -1447,6 +1463,8 @@ static struct pcie_port_service_driver aerdriver = {
 	.service	= PCIE_PORT_SERVICE_AER,
 
 	.probe		= aer_probe,
+	.suspend	= aer_suspend,
+	.resume		= aer_resume,
 	.remove		= aer_remove,
 	.reset_link	= aer_root_reset,
 };
Signature · 1 line
2.43.0
author maintainer
#4 lore raw

PATCH 5/7 PCI: keystone: Add workaround for Errata #i2037 (AM65x SR 1.0)

In reply to Sasha Levin

From: Kishon Vijay Abraham I <[email protected]>

[ Upstream commit 86f271f22bbb6391410a07e08d6ca3757fda01fa ]

Errata #i2037 in AM65x/DRA80xM Processors Silicon Revision 1.0 (SPRZ452D_July 2018_Revised December 2019 [1]) mentions when an inbound PCIe TLP spans more than two internal AXI 128-byte bursts, the bus may corrupt the packet payload and the corrupt data may cause associated applications or the processor to hang.

The workaround for Errata #i2037 is to limit the maximum read request size and maximum payload size to 128 bytes. Add workaround for Errata #i2037 here.

The errata and workaround is applicable only to AM65x SR 1.0 and later versions of the silicon will have this fixed.

[1] -> https://www.ti.com/lit/er/sprz452i/sprz452i.pdf

---
 drivers/pci/controller/dwc/pci-keystone.c | 44 ++++++++++++++++++++++-
 1 file changed, 43 insertions(+), 1 deletion(-)
drivers/pci/controller/dwc/pci-keystone.c +43 -1
index b18ddb2b9ef8b..a16fe2a558c7a 100644
@@ -35,6 +35,11 @@
 #define PCIE_DEVICEID_SHIFT	16
 
 /* Application registers */
+#define PID				0x000
+#define RTL				GENMASK(15, 11)
+#define RTL_SHIFT			11
+#define AM6_PCI_PG1_RTL_VER		0x15
+
 #define CMD_STATUS			0x004
 #define LTSSM_EN_VAL		        BIT(0)
 #define OB_XLAT_EN_VAL		        BIT(1)
@@ -107,6 +112,8 @@
 
 #define to_keystone_pcie(x)		dev_get_drvdata((x)->dev)
 
+#define PCI_DEVICE_ID_TI_AM654X		0xb00c
+
 struct ks_pcie_of_data {
 	enum dw_pcie_device_mode mode;
 	const struct dw_pcie_host_ops *host_ops;
@@ -534,7 +541,11 @@ static int ks_pcie_start_link(struct dw_pcie *pci)
 static void ks_pcie_quirk(struct pci_dev *dev)
 {
 	struct pci_bus *bus = dev->bus;
+	struct keystone_pcie *ks_pcie;
+	struct device *bridge_dev;
 	struct pci_dev *bridge;
+	u32 val;
+
 	static const struct pci_device_id rc_pci_devids[] = {
 		{ PCI_DEVICE(PCI_VENDOR_ID_TI, PCIE_RC_K2HK),
 		 .class = PCI_CLASS_BRIDGE_PCI << 8, .class_mask = ~0, },
@@ -546,6 +557,11 @@ static void ks_pcie_quirk(struct pci_dev *dev)
 		 .class = PCI_CLASS_BRIDGE_PCI << 8, .class_mask = ~0, },
 		{ 0, },
 	};
+	static const struct pci_device_id am6_pci_devids[] = {
+		{ PCI_DEVICE(PCI_VENDOR_ID_TI, PCI_DEVICE_ID_TI_AM654X),
+		 .class = PCI_CLASS_BRIDGE_PCI << 8, .class_mask = ~0, },
+		{ 0, },
+	};
 
 	if (pci_is_root_bus(bus))
 		bridge = dev;
@@ -567,10 +583,36 @@ static void ks_pcie_quirk(struct pci_dev *dev)
 	 */
 	if (pci_match_id(rc_pci_devids, bridge)) {
 		if (pcie_get_readrq(dev) > 256) {
-			dev_info(&dev->dev, "limiting MRRS to 256\n");
+			dev_info(&dev->dev, "limiting MRRS to 256 bytes\n");
 			pcie_set_readrq(dev, 256);
 		}
 	}
+
+	/*
+	 * Memory transactions fail with PCI controller in AM654 PG1.0
+	 * when MRRS is set to more than 128 bytes. Force the MRRS to
+	 * 128 bytes in all downstream devices.
+	 */
+	if (pci_match_id(am6_pci_devids, bridge)) {
+		bridge_dev = pci_get_host_bridge_device(dev);
+		if (!bridge_dev && !bridge_dev->parent)
+			return;
+
+		ks_pcie = dev_get_drvdata(bridge_dev->parent);
+		if (!ks_pcie)
+			return;
+
+		val = ks_pcie_app_readl(ks_pcie, PID);
+		val &= RTL;
+		val >>= RTL_SHIFT;
+		if (val != AM6_PCI_PG1_RTL_VER)
+			return;
+
+		if (pcie_get_readrq(dev) > 128) {
+			dev_info(&dev->dev, "limiting MRRS to 128 bytes\n");
+			pcie_set_readrq(dev, 128);
+		}
+	}
 }
 DECLARE_PCI_FIXUP_ENABLE(PCI_ANY_ID, PCI_ANY_ID, ks_pcie_quirk);
 
Signature · 1 line
2.43.0
author maintainer
#5 lore raw

PATCH 6/7 PCI: Add missing bridge lock to pci_bus_lock()

In reply to Sasha Levin

From: Dan Williams <[email protected]>

[ Upstream commit a4e772898f8bf2e7e1cf661a12c60a5612c4afab ]

One of the true positives that the cfg_access_lock lockdep effort identified is this sequence:

  WARNING: CPU: 14 PID: 1 at drivers/pci/pci.c:4886 pci_bridge_secondary_bus_reset+0x5d/0x70
  RIP: 0010:pci_bridge_secondary_bus_reset+0x5d/0x70
  Call Trace:
   <TASK>
   ? __warn+0x8c/0x190
   ? pci_bridge_secondary_bus_reset+0x5d/0x70
   ? report_bug+0x1f8/0x200
   ? handle_bug+0x3c/0x70
   ? exc_invalid_op+0x18/0x70
   ? asm_exc_invalid_op+0x1a/0x20
   ? pci_bridge_secondary_bus_reset+0x5d/0x70
   pci_reset_bus+0x1d8/0x270
   vmd_probe+0x778/0xa10
   pci_device_probe+0x95/0x120

Where pci_reset_bus() users are triggering unlocked secondary bus resets. Ironically pci_bus_reset(), several calls down from pci_reset_bus(), uses pci_bus_lock() before issuing the reset which locks everything *but* the bridge itself.

For the same motivation as adding:

  bridge = pci_upstream_bridge(dev);
  if (bridge)
    pci_dev_lock(bridge);

to pci_reset_function() for the "bus" and "cxl_bus" reset cases, add pci_dev_lock() for @bus->self to pci_bus_lock().

[bhelgaas: squash in recursive locking deadlock fix from Keith Busch:
https://lore.kernel.org/r/[email protected]]

---
 drivers/pci/pci.c | 35 +++++++++++++++++++++--------------
 1 file changed, 21 insertions(+), 14 deletions(-)
drivers/pci/pci.c +21 -14
index deafd229ef8b4..41050a35631fa 100644
@@ -5200,10 +5200,12 @@ static void pci_bus_lock(struct pci_bus *bus)
 {
 	struct pci_dev *dev;
 
+	pci_dev_lock(bus->self);
 	list_for_each_entry(dev, &bus->devices, bus_list) {
-		pci_dev_lock(dev);
 		if (dev->subordinate)
 			pci_bus_lock(dev->subordinate);
+		else
+			pci_dev_lock(dev);
 	}
 }
 
@@ -5215,8 +5217,10 @@ static void pci_bus_unlock(struct pci_bus *bus)
 	list_for_each_entry(dev, &bus->devices, bus_list) {
 		if (dev->subordinate)
 			pci_bus_unlock(dev->subordinate);
-		pci_dev_unlock(dev);
+		else
+			pci_dev_unlock(dev);
 	}
+	pci_dev_unlock(bus->self);
 }
 
 /* Return 1 on successful lock, 0 on contention */
@@ -5224,15 +5228,15 @@ static int pci_bus_trylock(struct pci_bus *bus)
 {
 	struct pci_dev *dev;
 
+	if (!pci_dev_trylock(bus->self))
+		return 0;
+
 	list_for_each_entry(dev, &bus->devices, bus_list) {
-		if (!pci_dev_trylock(dev))
-			goto unlock;
 		if (dev->subordinate) {
-			if (!pci_bus_trylock(dev->subordinate)) {
-				pci_dev_unlock(dev);
+			if (!pci_bus_trylock(dev->subordinate))
 				goto unlock;
-			}
-		}
+		} else if (!pci_dev_trylock(dev))
+			goto unlock;
 	}
 	return 1;
 
@@ -5240,8 +5244,10 @@ static int pci_bus_trylock(struct pci_bus *bus)
 	list_for_each_entry_continue_reverse(dev, &bus->devices, bus_list) {
 		if (dev->subordinate)
 			pci_bus_unlock(dev->subordinate);
-		pci_dev_unlock(dev);
+		else
+			pci_dev_unlock(dev);
 	}
+	pci_dev_unlock(bus->self);
 	return 0;
 }
 
@@ -5273,9 +5279,10 @@ static void pci_slot_lock(struct pci_slot *slot)
 	list_for_each_entry(dev, &slot->bus->devices, bus_list) {
 		if (!dev->slot || dev->slot != slot)
 			continue;
-		pci_dev_lock(dev);
 		if (dev->subordinate)
 			pci_bus_lock(dev->subordinate);
+		else
+			pci_dev_lock(dev);
 	}
 }
 
@@ -5301,14 +5308,13 @@ static int pci_slot_trylock(struct pci_slot *slot)
 	list_for_each_entry(dev, &slot->bus->devices, bus_list) {
 		if (!dev->slot || dev->slot != slot)
 			continue;
-		if (!pci_dev_trylock(dev))
-			goto unlock;
 		if (dev->subordinate) {
 			if (!pci_bus_trylock(dev->subordinate)) {
 				pci_dev_unlock(dev);
 				goto unlock;
 			}
-		}
+		} else if (!pci_dev_trylock(dev))
+			goto unlock;
 	}
 	return 1;
 
@@ -5319,7 +5325,8 @@ static int pci_slot_trylock(struct pci_slot *slot)
 			continue;
 		if (dev->subordinate)
 			pci_bus_unlock(dev->subordinate);
-		pci_dev_unlock(dev);
+		else
+			pci_dev_unlock(dev);
 	}
 	return 0;
 }
Signature · 1 line
2.43.0
author maintainer
#6 lore raw

PATCH 7/7 ALSA: usb: Fix UBSAN warning in parse_audio_unit()

In reply to Sasha Levin

From: Takashi Iwai <[email protected]>

[ Upstream commit 2f38cf730caedaeacdefb7ff35b0a3c1168117f9 ]

A malformed USB descriptor may pass the lengthy mixer description with a lot of channels, and this may overflow the 32bit integer shift size, as caught by syzbot UBSAN test. Although this won't cause any real trouble, it's better to address.

This patch introduces a sanity check of the number of channels to bail out the parsing when too many channels are found.

---
 sound/usb/mixer.c | 7 +++++++
 1 file changed, 7 insertions(+)
sound/usb/mixer.c +7 -0
index 67eb1293fa155..1374a4e093b3f 100644
@@ -1906,6 +1906,13 @@ static int parse_audio_feature_unit(struct mixer_build *state, int unitid,
 		bmaControls = ftr->bmaControls;
 	}
 
+	if (channels > 32) {
+		usb_audio_info(state->chip,
+			       "usbmixer: too many channels (%d) in unit %d\n",
+			       channels, unitid);
+		return -EINVAL;
+	}
+
 	/* parse the source unit */
 	err = parse_audio_unit(state, hdr->bSourceID);
 	if (err < 0)
Signature · 1 line
2.43.0