Merge branch 'pci/msi' into next
* pci/msi:
PCI/MSI: Update MSI/MSI-X bits in PCIEBUS-HOWTO
PCI/MSI: Document pci_alloc_irq_vectors(), deprecate pci_enable_msi()
PCI/MSI: Return -ENOSPC if pci_enable_msi_range() can't get enough vectors
PCI/portdrv: Use pci_irq_alloc_vectors()
PCI/MSI: Check that we have a legacy interrupt line before using it
PCI/MSI: Remove pci_msi_domain_{alloc,free}_irqs()
PCI/MSI: Remove unused pci_msi_create_default_irq_domain()
PCI/MSI: Return failure when msix_setup_entries() fails
PCI/MSI: Remove pci_enable_msi_{exact,range}()
amd-xgbe: Update PCI support to use new IRQ functions
[media] cobalt: use pci_irq_allocate_vectors()
PCI/MSI: Fix msi_capability_init() kernel-doc warnings
diff --git a/Documentation/PCI/pci-error-recovery.txt b/Documentation/PCI/pci-error-recovery.txt
index ac26869..da3b217 100644
--- a/Documentation/PCI/pci-error-recovery.txt
+++ b/Documentation/PCI/pci-error-recovery.txt
@@ -78,7 +78,6 @@
{
int (*error_detected)(struct pci_dev *dev, enum pci_channel_state);
int (*mmio_enabled)(struct pci_dev *dev);
- int (*link_reset)(struct pci_dev *dev);
int (*slot_reset)(struct pci_dev *dev);
void (*resume)(struct pci_dev *dev);
};
@@ -104,8 +103,7 @@
is not implemented, the corresponding feature is considered unsupported.
For example, if mmio_enabled() and resume() aren't there, then it
is assumed that the driver is not doing any direct recovery and requires
-a slot reset. If link_reset() is not implemented, the card is assumed to
-not care about link resets. Typically a driver will want to know about
+a slot reset. Typically a driver will want to know about
a slot_reset().
The actual steps taken by a platform to recover from a PCI error
@@ -232,25 +230,9 @@
STEP 3: Link Reset
------------------
-The platform resets the link, and then calls the link_reset() callback
-on all affected device drivers. This is a PCI-Express specific state
+The platform resets the link. This is a PCI-Express specific step
and is done whenever a non-fatal error has been detected that can be
-"solved" by resetting the link. This call informs the driver of the
-reset and the driver should check to see if the device appears to be
-in working condition.
-
-The driver is not supposed to restart normal driver I/O operations
-at this point. It should limit itself to "probing" the device to
-check its recoverability status. If all is right, then the platform
-will call resume() once all drivers have ack'd link_reset().
-
- Result codes:
- (identical to STEP 3 (MMIO Enabled)
-
-The platform then proceeds to either STEP 4 (Slot Reset) or STEP 5
-(Resume Operations).
-
->>> The current powerpc implementation does not implement this callback.
+"solved" by resetting the link.
STEP 4: Slot Reset
------------------
diff --git a/drivers/acpi/pci_mcfg.c b/drivers/acpi/pci_mcfg.c
index a6a4cea..2944353 100644
--- a/drivers/acpi/pci_mcfg.c
+++ b/drivers/acpi/pci_mcfg.c
@@ -195,11 +195,10 @@ int pci_mcfg_lookup(struct acpi_pci_root *root, struct resource *cfgres,
goto skip_lookup;
/*
- * We expect exact match, unless MCFG entry end bus covers more than
- * specified by caller.
+ * We expect the range in bus_res in the coverage of MCFG bus range.
*/
list_for_each_entry(e, &pci_mcfg_list, list) {
- if (e->segment == seg && e->bus_start == bus_res->start &&
+ if (e->segment == seg && e->bus_start <= bus_res->start &&
e->bus_end >= bus_res->end) {
root->mcfg_addr = e->addr;
}
diff --git a/drivers/infiniband/hw/hfi1/pcie.c b/drivers/infiniband/hw/hfi1/pcie.c
index 4ac8f33..ebd941f 100644
--- a/drivers/infiniband/hw/hfi1/pcie.c
+++ b/drivers/infiniband/hw/hfi1/pcie.c
@@ -598,15 +598,6 @@ pci_slot_reset(struct pci_dev *pdev)
return PCI_ERS_RESULT_CAN_RECOVER;
}
-static pci_ers_result_t
-pci_link_reset(struct pci_dev *pdev)
-{
- struct hfi1_devdata *dd = pci_get_drvdata(pdev);
-
- dd_dev_info(dd, "HFI1 link_reset function called, ignored\n");
- return PCI_ERS_RESULT_CAN_RECOVER;
-}
-
static void
pci_resume(struct pci_dev *pdev)
{
@@ -625,7 +616,6 @@ pci_resume(struct pci_dev *pdev)
const struct pci_error_handlers hfi1_pci_err_handler = {
.error_detected = pci_error_detected,
.mmio_enabled = pci_mmio_enabled,
- .link_reset = pci_link_reset,
.slot_reset = pci_slot_reset,
.resume = pci_resume,
};
diff --git a/drivers/infiniband/hw/qib/qib_pcie.c b/drivers/infiniband/hw/qib/qib_pcie.c
index 6abe1c6..c379b83 100644
--- a/drivers/infiniband/hw/qib/qib_pcie.c
+++ b/drivers/infiniband/hw/qib/qib_pcie.c
@@ -682,13 +682,6 @@ qib_pci_slot_reset(struct pci_dev *pdev)
return PCI_ERS_RESULT_CAN_RECOVER;
}
-static pci_ers_result_t
-qib_pci_link_reset(struct pci_dev *pdev)
-{
- qib_devinfo(pdev, "QIB link_reset function called, ignored\n");
- return PCI_ERS_RESULT_CAN_RECOVER;
-}
-
static void
qib_pci_resume(struct pci_dev *pdev)
{
@@ -707,7 +700,6 @@ qib_pci_resume(struct pci_dev *pdev)
const struct pci_error_handlers qib_pci_err_handler = {
.error_detected = qib_pci_error_detected,
.mmio_enabled = qib_pci_mmio_enabled,
- .link_reset = qib_pci_link_reset,
.slot_reset = qib_pci_slot_reset,
.resume = qib_pci_resume,
};
diff --git a/drivers/media/pci/ngene/ngene-cards.c b/drivers/media/pci/ngene/ngene-cards.c
index 423e8c8..8438c1c 100644
--- a/drivers/media/pci/ngene/ngene-cards.c
+++ b/drivers/media/pci/ngene/ngene-cards.c
@@ -781,12 +781,6 @@ static pci_ers_result_t ngene_error_detected(struct pci_dev *dev,
return PCI_ERS_RESULT_CAN_RECOVER;
}
-static pci_ers_result_t ngene_link_reset(struct pci_dev *dev)
-{
- printk(KERN_INFO DEVICE_NAME ": link reset\n");
- return 0;
-}
-
static pci_ers_result_t ngene_slot_reset(struct pci_dev *dev)
{
printk(KERN_INFO DEVICE_NAME ": slot reset\n");
@@ -800,7 +794,6 @@ static void ngene_resume(struct pci_dev *dev)
static const struct pci_error_handlers ngene_errors = {
.error_detected = ngene_error_detected,
- .link_reset = ngene_link_reset,
.slot_reset = ngene_slot_reset,
.resume = ngene_resume,
};
diff --git a/drivers/misc/genwqe/card_base.c b/drivers/misc/genwqe/card_base.c
index 6c1f49a..4fd21e8 100644
--- a/drivers/misc/genwqe/card_base.c
+++ b/drivers/misc/genwqe/card_base.c
@@ -1336,7 +1336,6 @@ static int genwqe_sriov_configure(struct pci_dev *dev, int numvfs)
static struct pci_error_handlers genwqe_err_handler = {
.error_detected = genwqe_err_error_detected,
.mmio_enabled = genwqe_err_result_none,
- .link_reset = genwqe_err_result_none,
.slot_reset = genwqe_err_slot_reset,
.resume = genwqe_err_resume,
};
diff --git a/drivers/pci/access.c b/drivers/pci/access.c
index db23954..68bd720 100644
--- a/drivers/pci/access.c
+++ b/drivers/pci/access.c
@@ -684,8 +684,9 @@ void pci_cfg_access_unlock(struct pci_dev *dev)
WARN_ON(!dev->block_cfg_access);
dev->block_cfg_access = 0;
- wake_up_all(&pci_cfg_wait);
raw_spin_unlock_irqrestore(&pci_lock, flags);
+
+ wake_up_all(&pci_cfg_wait);
}
EXPORT_SYMBOL_GPL(pci_cfg_access_unlock);
diff --git a/drivers/pci/hotplug/acpiphp_ibm.c b/drivers/pci/hotplug/acpiphp_ibm.c
index 68d105a..984c7e8 100644
--- a/drivers/pci/hotplug/acpiphp_ibm.c
+++ b/drivers/pci/hotplug/acpiphp_ibm.c
@@ -107,7 +107,7 @@ static void __exit ibm_acpiphp_exit(void);
static acpi_handle ibm_acpi_handle;
static struct notification ibm_note;
-static struct bin_attribute ibm_apci_table_attr = {
+static struct bin_attribute ibm_apci_table_attr __ro_after_init = {
.attr = {
.name = "apci_table",
.mode = S_IRUGO,
diff --git a/drivers/pci/hotplug/rpadlpar_core.c b/drivers/pci/hotplug/rpadlpar_core.c
index c614ff7..3f93a4e 100644
--- a/drivers/pci/hotplug/rpadlpar_core.c
+++ b/drivers/pci/hotplug/rpadlpar_core.c
@@ -463,7 +463,6 @@ static inline int is_dlpar_capable(void)
int __init rpadlpar_io_init(void)
{
- int rc = 0;
if (!is_dlpar_capable()) {
printk(KERN_WARNING "%s: partition not DLPAR capable\n",
@@ -471,8 +470,7 @@ int __init rpadlpar_io_init(void)
return -EPERM;
}
- rc = dlpar_sysfs_init();
- return rc;
+ return dlpar_sysfs_init();
}
void rpadlpar_io_exit(void)
diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
index 1ccce1c..3e0516e 100644
--- a/drivers/pci/pci-driver.c
+++ b/drivers/pci/pci-driver.c
@@ -381,8 +381,6 @@ static int __pci_device_probe(struct pci_driver *drv, struct pci_dev *pci_dev)
id = pci_match_device(drv, pci_dev);
if (id)
error = pci_call_probe(drv, pci_dev, id);
- if (error >= 0)
- error = 0;
}
return error;
}
diff --git a/drivers/pci/pcie/Kconfig b/drivers/pci/pcie/Kconfig
index 7ce7763..ac53edb 100644
--- a/drivers/pci/pcie/Kconfig
+++ b/drivers/pci/pcie/Kconfig
@@ -71,6 +71,14 @@
Enable PCI Express ASPM L0s and L1 where possible, even if the
BIOS did not.
+config PCIEASPM_POWER_SUPERSAVE
+ bool "Power Supersave"
+ depends on PCIEASPM
+ help
+ Same as PCIEASPM_POWERSAVE, except it also enables L1 substates where
+ possible. This would result in higher power savings while staying in L1
+ where the components support it.
+
config PCIEASPM_PERFORMANCE
bool "Performance"
depends on PCIEASPM
diff --git a/drivers/pci/pcie/aspm.c b/drivers/pci/pcie/aspm.c
index 17ac1dc..a9bcd56 100644
--- a/drivers/pci/pcie/aspm.c
+++ b/drivers/pci/pcie/aspm.c
@@ -30,8 +30,29 @@
#define ASPM_STATE_L0S_UP (1) /* Upstream direction L0s state */
#define ASPM_STATE_L0S_DW (2) /* Downstream direction L0s state */
#define ASPM_STATE_L1 (4) /* L1 state */
+#define ASPM_STATE_L1_1 (8) /* ASPM L1.1 state */
+#define ASPM_STATE_L1_2 (0x10) /* ASPM L1.2 state */
+#define ASPM_STATE_L1_1_PCIPM (0x20) /* PCI PM L1.1 state */
+#define ASPM_STATE_L1_2_PCIPM (0x40) /* PCI PM L1.2 state */
+#define ASPM_STATE_L1_SS_PCIPM (ASPM_STATE_L1_1_PCIPM | ASPM_STATE_L1_2_PCIPM)
+#define ASPM_STATE_L1_2_MASK (ASPM_STATE_L1_2 | ASPM_STATE_L1_2_PCIPM)
+#define ASPM_STATE_L1SS (ASPM_STATE_L1_1 | ASPM_STATE_L1_1_PCIPM |\
+ ASPM_STATE_L1_2_MASK)
#define ASPM_STATE_L0S (ASPM_STATE_L0S_UP | ASPM_STATE_L0S_DW)
-#define ASPM_STATE_ALL (ASPM_STATE_L0S | ASPM_STATE_L1)
+#define ASPM_STATE_ALL (ASPM_STATE_L0S | ASPM_STATE_L1 | \
+ ASPM_STATE_L1SS)
+
+/*
+ * When L1 substates are enabled, the LTR L1.2 threshold is a timing parameter
+ * that decides whether L1.1 or L1.2 is entered (Refer PCIe spec for details).
+ * Not sure is there is a way to "calculate" this on the fly, but maybe we
+ * could turn it into a parameter in future. This value has been taken from
+ * the following files from Intel's coreboot (which is the only code I found
+ * to have used this):
+ * https://www.coreboot.org/pipermail/coreboot-gerrit/2015-March/021134.html
+ * https://review.coreboot.org/#/c/8832/
+ */
+#define LTR_L1_2_THRESHOLD_BITS ((1 << 21) | (1 << 23) | (1 << 30))
struct aspm_latency {
u32 l0s; /* L0s latency (nsec) */
@@ -40,6 +61,7 @@ struct aspm_latency {
struct pcie_link_state {
struct pci_dev *pdev; /* Upstream component of the Link */
+ struct pci_dev *downstream; /* Downstream component, function 0 */
struct pcie_link_state *root; /* pointer to the root port link */
struct pcie_link_state *parent; /* pointer to the parent Link state */
struct list_head sibling; /* node in link_list */
@@ -47,11 +69,11 @@ struct pcie_link_state {
struct list_head link; /* node in parent's children list */
/* ASPM state */
- u32 aspm_support:3; /* Supported ASPM state */
- u32 aspm_enabled:3; /* Enabled ASPM state */
- u32 aspm_capable:3; /* Capable ASPM state with latency */
- u32 aspm_default:3; /* Default ASPM state by BIOS */
- u32 aspm_disable:3; /* Disabled ASPM state */
+ u32 aspm_support:7; /* Supported ASPM state */
+ u32 aspm_enabled:7; /* Enabled ASPM state */
+ u32 aspm_capable:7; /* Capable ASPM state with latency */
+ u32 aspm_default:7; /* Default ASPM state by BIOS */
+ u32 aspm_disable:7; /* Disabled ASPM state */
/* Clock PM state */
u32 clkpm_capable:1; /* Clock PM capable? */
@@ -66,6 +88,14 @@ struct pcie_link_state {
* has one slot under it, so at most there are 8 functions.
*/
struct aspm_latency acceptable[8];
+
+ /* L1 PM Substate info */
+ struct {
+ u32 up_cap_ptr; /* L1SS cap ptr in upstream dev */
+ u32 dw_cap_ptr; /* L1SS cap ptr in downstream dev */
+ u32 ctl1; /* value to be programmed in ctl1 */
+ u32 ctl2; /* value to be programmed in ctl2 */
+ } l1ss;
};
static int aspm_disabled, aspm_force;
@@ -76,11 +106,14 @@ static LIST_HEAD(link_list);
#define POLICY_DEFAULT 0 /* BIOS default setting */
#define POLICY_PERFORMANCE 1 /* high performance */
#define POLICY_POWERSAVE 2 /* high power saving */
+#define POLICY_POWER_SUPERSAVE 3 /* possibly even more power saving */
#ifdef CONFIG_PCIEASPM_PERFORMANCE
static int aspm_policy = POLICY_PERFORMANCE;
#elif defined CONFIG_PCIEASPM_POWERSAVE
static int aspm_policy = POLICY_POWERSAVE;
+#elif defined CONFIG_PCIEASPM_POWER_SUPERSAVE
+static int aspm_policy = POLICY_POWER_SUPERSAVE;
#else
static int aspm_policy;
#endif
@@ -88,7 +121,8 @@ static int aspm_policy;
static const char *policy_str[] = {
[POLICY_DEFAULT] = "default",
[POLICY_PERFORMANCE] = "performance",
- [POLICY_POWERSAVE] = "powersave"
+ [POLICY_POWERSAVE] = "powersave",
+ [POLICY_POWER_SUPERSAVE] = "powersupersave"
};
#define LINK_RETRAIN_TIMEOUT HZ
@@ -101,6 +135,9 @@ static int policy_to_aspm_state(struct pcie_link_state *link)
return 0;
case POLICY_POWERSAVE:
/* Enable ASPM L0s/L1 */
+ return (ASPM_STATE_L0S | ASPM_STATE_L1);
+ case POLICY_POWER_SUPERSAVE:
+ /* Enable Everything */
return ASPM_STATE_ALL;
case POLICY_DEFAULT:
return link->aspm_default;
@@ -115,7 +152,8 @@ static int policy_to_clkpm_state(struct pcie_link_state *link)
/* Disable ASPM and Clock PM */
return 0;
case POLICY_POWERSAVE:
- /* Disable Clock PM */
+ case POLICY_POWER_SUPERSAVE:
+ /* Enable Clock PM */
return 1;
case POLICY_DEFAULT:
return link->clkpm_default;
@@ -278,11 +316,33 @@ static u32 calc_l1_acceptable(u32 encoding)
return (1000 << encoding);
}
+/* Convert L1SS T_pwr encoding to usec */
+static u32 calc_l1ss_pwron(struct pci_dev *pdev, u32 scale, u32 val)
+{
+ switch (scale) {
+ case 0:
+ return val * 2;
+ case 1:
+ return val * 10;
+ case 2:
+ return val * 100;
+ }
+ dev_err(&pdev->dev, "%s: Invalid T_PwrOn scale: %u\n",
+ __func__, scale);
+ return 0;
+}
+
struct aspm_register_info {
u32 support:2;
u32 enabled:2;
u32 latency_encoding_l0s;
u32 latency_encoding_l1;
+
+ /* L1 substates */
+ u32 l1ss_cap_ptr;
+ u32 l1ss_cap;
+ u32 l1ss_ctl1;
+ u32 l1ss_ctl2;
};
static void pcie_get_aspm_reg(struct pci_dev *pdev,
@@ -297,6 +357,22 @@ static void pcie_get_aspm_reg(struct pci_dev *pdev,
info->latency_encoding_l1 = (reg32 & PCI_EXP_LNKCAP_L1EL) >> 15;
pcie_capability_read_word(pdev, PCI_EXP_LNKCTL, ®16);
info->enabled = reg16 & PCI_EXP_LNKCTL_ASPMC;
+
+ /* Read L1 PM substate capabilities */
+ info->l1ss_cap = info->l1ss_ctl1 = info->l1ss_ctl2 = 0;
+ info->l1ss_cap_ptr = pci_find_ext_capability(pdev, PCI_EXT_CAP_ID_L1SS);
+ if (!info->l1ss_cap_ptr)
+ return;
+ pci_read_config_dword(pdev, info->l1ss_cap_ptr + PCI_L1SS_CAP,
+ &info->l1ss_cap);
+ if (!(info->l1ss_cap & PCI_L1SS_CAP_L1_PM_SS)) {
+ info->l1ss_cap = 0;
+ return;
+ }
+ pci_read_config_dword(pdev, info->l1ss_cap_ptr + PCI_L1SS_CTL1,
+ &info->l1ss_ctl1);
+ pci_read_config_dword(pdev, info->l1ss_cap_ptr + PCI_L1SS_CTL2,
+ &info->l1ss_ctl2);
}
static void pcie_aspm_check_latency(struct pci_dev *endpoint)
@@ -327,6 +403,14 @@ static void pcie_aspm_check_latency(struct pci_dev *endpoint)
* Check L1 latency.
* Every switch on the path to root complex need 1
* more microsecond for L1. Spec doesn't mention L0s.
+ *
+ * The exit latencies for L1 substates are not advertised
+ * by a device. Since the spec also doesn't mention a way
+ * to determine max latencies introduced by enabling L1
+ * substates on the components, it is not clear how to do
+ * a L1 substate exit latency check. We assume that the
+ * L1 exit latencies advertised by a device include L1
+ * substate latencies (and hence do not do any check).
*/
latency = max_t(u32, link->latency_up.l1, link->latency_dw.l1);
if ((link->aspm_capable & ASPM_STATE_L1) &&
@@ -338,6 +422,60 @@ static void pcie_aspm_check_latency(struct pci_dev *endpoint)
}
}
+/*
+ * The L1 PM substate capability is only implemented in function 0 in a
+ * multi function device.
+ */
+static struct pci_dev *pci_function_0(struct pci_bus *linkbus)
+{
+ struct pci_dev *child;
+
+ list_for_each_entry(child, &linkbus->devices, bus_list)
+ if (PCI_FUNC(child->devfn) == 0)
+ return child;
+ return NULL;
+}
+
+/* Calculate L1.2 PM substate timing parameters */
+static void aspm_calc_l1ss_info(struct pcie_link_state *link,
+ struct aspm_register_info *upreg,
+ struct aspm_register_info *dwreg)
+{
+ u32 val1, val2, scale1, scale2;
+
+ link->l1ss.up_cap_ptr = upreg->l1ss_cap_ptr;
+ link->l1ss.dw_cap_ptr = dwreg->l1ss_cap_ptr;
+ link->l1ss.ctl1 = link->l1ss.ctl2 = 0;
+
+ if (!(link->aspm_support & ASPM_STATE_L1_2_MASK))
+ return;
+
+ /* Choose the greater of the two T_cmn_mode_rstr_time */
+ val1 = (upreg->l1ss_cap >> 8) & 0xFF;
+ val2 = (upreg->l1ss_cap >> 8) & 0xFF;
+ if (val1 > val2)
+ link->l1ss.ctl1 |= val1 << 8;
+ else
+ link->l1ss.ctl1 |= val2 << 8;
+ /*
+ * We currently use LTR L1.2 threshold to be fixed constant picked from
+ * Intel's coreboot.
+ */
+ link->l1ss.ctl1 |= LTR_L1_2_THRESHOLD_BITS;
+
+ /* Choose the greater of the two T_pwr_on */
+ val1 = (upreg->l1ss_cap >> 19) & 0x1F;
+ scale1 = (upreg->l1ss_cap >> 16) & 0x03;
+ val2 = (dwreg->l1ss_cap >> 19) & 0x1F;
+ scale2 = (dwreg->l1ss_cap >> 16) & 0x03;
+
+ if (calc_l1ss_pwron(link->pdev, scale1, val1) >
+ calc_l1ss_pwron(link->downstream, scale2, val2))
+ link->l1ss.ctl2 |= scale1 | (val1 << 3);
+ else
+ link->l1ss.ctl2 |= scale2 | (val2 << 3);
+}
+
static void pcie_aspm_cap_init(struct pcie_link_state *link, int blacklist)
{
struct pci_dev *child, *parent = link->pdev;
@@ -353,8 +491,9 @@ static void pcie_aspm_cap_init(struct pcie_link_state *link, int blacklist)
/* Get upstream/downstream components' register state */
pcie_get_aspm_reg(parent, &upreg);
- child = list_entry(linkbus->devices.next, struct pci_dev, bus_list);
+ child = pci_function_0(linkbus);
pcie_get_aspm_reg(child, &dwreg);
+ link->downstream = child;
/*
* If ASPM not supported, don't mess with the clocks and link,
@@ -397,6 +536,28 @@ static void pcie_aspm_cap_init(struct pcie_link_state *link, int blacklist)
link->latency_up.l1 = calc_l1_latency(upreg.latency_encoding_l1);
link->latency_dw.l1 = calc_l1_latency(dwreg.latency_encoding_l1);
+ /* Setup L1 substate */
+ if (upreg.l1ss_cap & dwreg.l1ss_cap & PCI_L1SS_CAP_ASPM_L1_1)
+ link->aspm_support |= ASPM_STATE_L1_1;
+ if (upreg.l1ss_cap & dwreg.l1ss_cap & PCI_L1SS_CAP_ASPM_L1_2)
+ link->aspm_support |= ASPM_STATE_L1_2;
+ if (upreg.l1ss_cap & dwreg.l1ss_cap & PCI_L1SS_CAP_PCIPM_L1_1)
+ link->aspm_support |= ASPM_STATE_L1_1_PCIPM;
+ if (upreg.l1ss_cap & dwreg.l1ss_cap & PCI_L1SS_CAP_PCIPM_L1_2)
+ link->aspm_support |= ASPM_STATE_L1_2_PCIPM;
+
+ if (upreg.l1ss_ctl1 & dwreg.l1ss_ctl1 & PCI_L1SS_CTL1_ASPM_L1_1)
+ link->aspm_enabled |= ASPM_STATE_L1_1;
+ if (upreg.l1ss_ctl1 & dwreg.l1ss_ctl1 & PCI_L1SS_CTL1_ASPM_L1_2)
+ link->aspm_enabled |= ASPM_STATE_L1_2;
+ if (upreg.l1ss_ctl1 & dwreg.l1ss_ctl1 & PCI_L1SS_CTL1_PCIPM_L1_1)
+ link->aspm_enabled |= ASPM_STATE_L1_1_PCIPM;
+ if (upreg.l1ss_ctl1 & dwreg.l1ss_ctl1 & PCI_L1SS_CTL1_PCIPM_L1_2)
+ link->aspm_enabled |= ASPM_STATE_L1_2_PCIPM;
+
+ if (link->aspm_support & ASPM_STATE_L1SS)
+ aspm_calc_l1ss_info(link, &upreg, &dwreg);
+
/* Save default state */
link->aspm_default = link->aspm_enabled;
@@ -435,6 +596,92 @@ static void pcie_aspm_cap_init(struct pcie_link_state *link, int blacklist)
}
}
+static void pci_clear_and_set_dword(struct pci_dev *pdev, int pos,
+ u32 clear, u32 set)
+{
+ u32 val;
+
+ pci_read_config_dword(pdev, pos, &val);
+ val &= ~clear;
+ val |= set;
+ pci_write_config_dword(pdev, pos, val);
+}
+
+/* Configure the ASPM L1 substates */
+static void pcie_config_aspm_l1ss(struct pcie_link_state *link, u32 state)
+{
+ u32 val, enable_req;
+ struct pci_dev *child = link->downstream, *parent = link->pdev;
+ u32 up_cap_ptr = link->l1ss.up_cap_ptr;
+ u32 dw_cap_ptr = link->l1ss.dw_cap_ptr;
+
+ enable_req = (link->aspm_enabled ^ state) & state;
+
+ /*
+ * Here are the rules specified in the PCIe spec for enabling L1SS:
+ * - When enabling L1.x, enable bit at parent first, then at child
+ * - When disabling L1.x, disable bit at child first, then at parent
+ * - When enabling ASPM L1.x, need to disable L1
+ * (at child followed by parent).
+ * - The ASPM/PCIPM L1.2 must be disabled while programming timing
+ * parameters
+ *
+ * To keep it simple, disable all L1SS bits first, and later enable
+ * what is needed.
+ */
+
+ /* Disable all L1 substates */
+ pci_clear_and_set_dword(child, dw_cap_ptr + PCI_L1SS_CTL1,
+ PCI_L1SS_CTL1_L1SS_MASK, 0);
+ pci_clear_and_set_dword(parent, up_cap_ptr + PCI_L1SS_CTL1,
+ PCI_L1SS_CTL1_L1SS_MASK, 0);
+ /*
+ * If needed, disable L1, and it gets enabled later
+ * in pcie_config_aspm_link().
+ */
+ if (enable_req & (ASPM_STATE_L1_1 | ASPM_STATE_L1_2)) {
+ pcie_capability_clear_and_set_word(child, PCI_EXP_LNKCTL,
+ PCI_EXP_LNKCTL_ASPM_L1, 0);
+ pcie_capability_clear_and_set_word(parent, PCI_EXP_LNKCTL,
+ PCI_EXP_LNKCTL_ASPM_L1, 0);
+ }
+
+ if (enable_req & ASPM_STATE_L1_2_MASK) {
+
+ /* Program T_pwr_on in both ports */
+ pci_write_config_dword(parent, up_cap_ptr + PCI_L1SS_CTL2,
+ link->l1ss.ctl2);
+ pci_write_config_dword(child, dw_cap_ptr + PCI_L1SS_CTL2,
+ link->l1ss.ctl2);
+
+ /* Program T_cmn_mode in parent */
+ pci_clear_and_set_dword(parent, up_cap_ptr + PCI_L1SS_CTL1,
+ 0xFF00, link->l1ss.ctl1);
+
+ /* Program LTR L1.2 threshold in both ports */
+ pci_clear_and_set_dword(parent, dw_cap_ptr + PCI_L1SS_CTL1,
+ 0xE3FF0000, link->l1ss.ctl1);
+ pci_clear_and_set_dword(child, dw_cap_ptr + PCI_L1SS_CTL1,
+ 0xE3FF0000, link->l1ss.ctl1);
+ }
+
+ val = 0;
+ if (state & ASPM_STATE_L1_1)
+ val |= PCI_L1SS_CTL1_ASPM_L1_1;
+ if (state & ASPM_STATE_L1_2)
+ val |= PCI_L1SS_CTL1_ASPM_L1_2;
+ if (state & ASPM_STATE_L1_1_PCIPM)
+ val |= PCI_L1SS_CTL1_PCIPM_L1_1;
+ if (state & ASPM_STATE_L1_2_PCIPM)
+ val |= PCI_L1SS_CTL1_PCIPM_L1_2;
+
+ /* Enable what we need to enable */
+ pci_clear_and_set_dword(parent, up_cap_ptr + PCI_L1SS_CTL1,
+ PCI_L1SS_CAP_L1_PM_SS, val);
+ pci_clear_and_set_dword(child, dw_cap_ptr + PCI_L1SS_CTL1,
+ PCI_L1SS_CAP_L1_PM_SS, val);
+}
+
static void pcie_config_aspm_dev(struct pci_dev *pdev, u32 val)
{
pcie_capability_clear_and_set_word(pdev, PCI_EXP_LNKCTL,
@@ -444,11 +691,23 @@ static void pcie_config_aspm_dev(struct pci_dev *pdev, u32 val)
static void pcie_config_aspm_link(struct pcie_link_state *link, u32 state)
{
u32 upstream = 0, dwstream = 0;
- struct pci_dev *child, *parent = link->pdev;
+ struct pci_dev *child = link->downstream, *parent = link->pdev;
struct pci_bus *linkbus = parent->subordinate;
- /* Nothing to do if the link is already in the requested state */
+ /* Enable only the states that were not explicitly disabled */
state &= (link->aspm_capable & ~link->aspm_disable);
+
+ /* Can't enable any substates if L1 is not enabled */
+ if (!(state & ASPM_STATE_L1))
+ state &= ~ASPM_STATE_L1SS;
+
+ /* Spec says both ports must be in D0 before enabling PCI PM substates*/
+ if (parent->current_state != PCI_D0 || child->current_state != PCI_D0) {
+ state &= ~ASPM_STATE_L1_SS_PCIPM;
+ state |= (link->aspm_enabled & ASPM_STATE_L1_SS_PCIPM);
+ }
+
+ /* Nothing to do if the link is already in the requested state */
if (link->aspm_enabled == state)
return;
/* Convert ASPM state to upstream/downstream ASPM register state */
@@ -460,6 +719,10 @@ static void pcie_config_aspm_link(struct pcie_link_state *link, u32 state)
upstream |= PCI_EXP_LNKCTL_ASPM_L1;
dwstream |= PCI_EXP_LNKCTL_ASPM_L1;
}
+
+ if (link->aspm_capable & ASPM_STATE_L1SS)
+ pcie_config_aspm_l1ss(link, state);
+
/*
* Spec 2.0 suggests all functions should be configured the
* same setting for ASPM. Enabling ASPM L1 should be done in
@@ -612,7 +875,8 @@ void pcie_aspm_init_link_state(struct pci_dev *pdev)
* the BIOS's expectation, we'll do so once pci_enable_device() is
* called.
*/
- if (aspm_policy != POLICY_POWERSAVE) {
+ if (aspm_policy != POLICY_POWERSAVE &&
+ aspm_policy != POLICY_POWER_SUPERSAVE) {
pcie_config_aspm_path(link);
pcie_set_clkpm(link, policy_to_clkpm_state(link));
}
@@ -712,7 +976,8 @@ void pcie_aspm_powersave_config_link(struct pci_dev *pdev)
if (aspm_disabled || !link)
return;
- if (aspm_policy != POLICY_POWERSAVE)
+ if (aspm_policy != POLICY_POWERSAVE &&
+ aspm_policy != POLICY_POWER_SUPERSAVE)
return;
down_read(&pci_bus_sem);
diff --git a/drivers/pci/pcie/pcie-dpc.c b/drivers/pci/pcie/pcie-dpc.c
index 9811b14..d4d70ef 100644
--- a/drivers/pci/pcie/pcie-dpc.c
+++ b/drivers/pci/pcie/pcie-dpc.c
@@ -19,8 +19,28 @@ struct dpc_dev {
struct pcie_device *dev;
struct work_struct work;
int cap_pos;
+ bool rp;
};
+static int dpc_wait_rp_inactive(struct dpc_dev *dpc)
+{
+ unsigned long timeout = jiffies + HZ;
+ struct pci_dev *pdev = dpc->dev->port;
+ u16 status;
+
+ pci_read_config_word(pdev, dpc->cap_pos + PCI_EXP_DPC_STATUS, &status);
+ while (status & PCI_EXP_DPC_RP_BUSY &&
+ !time_after(jiffies, timeout)) {
+ msleep(10);
+ pci_read_config_word(pdev, dpc->cap_pos + PCI_EXP_DPC_STATUS, &status);
+ }
+ if (status & PCI_EXP_DPC_RP_BUSY) {
+ dev_warn(&pdev->dev, "DPC root port still busy\n");
+ return -EBUSY;
+ }
+ return 0;
+}
+
static void dpc_wait_link_inactive(struct pci_dev *pdev)
{
unsigned long timeout = jiffies + HZ;
@@ -33,7 +53,7 @@ static void dpc_wait_link_inactive(struct pci_dev *pdev)
pcie_capability_read_word(pdev, PCI_EXP_LNKSTA, &lnk_status);
}
if (lnk_status & PCI_EXP_LNKSTA_DLLLA)
- dev_warn(&pdev->dev, "Link state not disabled for DPC event");
+ dev_warn(&pdev->dev, "Link state not disabled for DPC event\n");
}
static void interrupt_event_handler(struct work_struct *work)
@@ -52,6 +72,8 @@ static void interrupt_event_handler(struct work_struct *work)
pci_unlock_rescan_remove();
dpc_wait_link_inactive(pdev);
+ if (dpc->rp && dpc_wait_rp_inactive(dpc))
+ return;
pci_write_config_word(pdev, dpc->cap_pos + PCI_EXP_DPC_STATUS,
PCI_EXP_DPC_STATUS_TRIGGER | PCI_EXP_DPC_STATUS_INTERRUPT);
}
@@ -73,11 +95,15 @@ static irqreturn_t dpc_irq(int irq, void *context)
if (status & PCI_EXP_DPC_STATUS_TRIGGER) {
u16 reason = (status >> 1) & 0x3;
+ u16 ext_reason = (status >> 5) & 0x3;
- dev_warn(&dpc->dev->device, "DPC %s triggered, remove downstream devices\n",
+ dev_warn(&dpc->dev->device, "DPC %s detected, remove downstream devices\n",
(reason == 0) ? "unmasked uncorrectable error" :
(reason == 1) ? "ERR_NONFATAL" :
- (reason == 2) ? "ERR_FATAL" : "extended error");
+ (reason == 2) ? "ERR_FATAL" :
+ (ext_reason == 0) ? "RP PIO error" :
+ (ext_reason == 1) ? "software trigger" :
+ "reserved error");
schedule_work(&dpc->work);
}
return IRQ_HANDLED;
@@ -111,6 +137,8 @@ static int dpc_probe(struct pcie_device *dev)
pci_read_config_word(pdev, dpc->cap_pos + PCI_EXP_DPC_CAP, &cap);
pci_read_config_word(pdev, dpc->cap_pos + PCI_EXP_DPC_CTL, &ctl);
+ dpc->rp = (cap & PCI_EXP_DPC_CAP_RP_EXT);
+
ctl |= PCI_EXP_DPC_CTL_EN_NONFATAL | PCI_EXP_DPC_CTL_INT_EN;
pci_write_config_word(pdev, dpc->cap_pos + PCI_EXP_DPC_CTL, ctl);
diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c
index e164b5c..3abc942 100644
--- a/drivers/pci/probe.c
+++ b/drivers/pci/probe.c
@@ -1554,8 +1554,16 @@ static void program_hpp_type0(struct pci_dev *dev, struct hpp_type0 *hpp)
static void program_hpp_type1(struct pci_dev *dev, struct hpp_type1 *hpp)
{
- if (hpp)
- dev_warn(&dev->dev, "PCI-X settings not supported\n");
+ int pos;
+
+ if (!hpp)
+ return;
+
+ pos = pci_find_capability(dev, PCI_CAP_ID_PCIX);
+ if (!pos)
+ return;
+
+ dev_warn(&dev->dev, "PCI-X settings not supported\n");
}
static bool pcie_root_rcb_set(struct pci_dev *dev)
@@ -1581,6 +1589,9 @@ static void program_hpp_type2(struct pci_dev *dev, struct hpp_type2 *hpp)
if (!hpp)
return;
+ if (!pci_is_pcie(dev))
+ return;
+
if (hpp->revision > 1) {
dev_warn(&dev->dev, "PCIe settings rev %d not supported\n",
hpp->revision);
@@ -1650,12 +1661,30 @@ static void program_hpp_type2(struct pci_dev *dev, struct hpp_type2 *hpp)
*/
}
+static void pci_configure_extended_tags(struct pci_dev *dev)
+{
+ u32 dev_cap;
+ int ret;
+
+ if (!pci_is_pcie(dev))
+ return;
+
+ ret = pcie_capability_read_dword(dev, PCI_EXP_DEVCAP, &dev_cap);
+ if (ret)
+ return;
+
+ if (dev_cap & PCI_EXP_DEVCAP_EXT_TAG)
+ pcie_capability_set_word(dev, PCI_EXP_DEVCTL,
+ PCI_EXP_DEVCTL_EXT_TAG);
+}
+
static void pci_configure_device(struct pci_dev *dev)
{
struct hotplug_params hpp;
int ret;
pci_configure_mps(dev);
+ pci_configure_extended_tags(dev);
memset(&hpp, 0, sizeof(hpp));
ret = pci_get_hp_params(dev, &hpp);
diff --git a/include/linux/pci.h b/include/linux/pci.h
index 2159376..6732d32 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -678,9 +678,6 @@ struct pci_error_handlers {
/* MMIO has been re-enabled, but not DMA */
pci_ers_result_t (*mmio_enabled)(struct pci_dev *dev);
- /* PCI Express link has been reset */
- pci_ers_result_t (*link_reset)(struct pci_dev *dev);
-
/* PCI slot has been reset */
pci_ers_result_t (*slot_reset)(struct pci_dev *dev);
diff --git a/include/uapi/linux/pci_regs.h b/include/uapi/linux/pci_regs.h
index 174d114..634c9c4 100644
--- a/include/uapi/linux/pci_regs.h
+++ b/include/uapi/linux/pci_regs.h
@@ -682,6 +682,7 @@
#define PCI_EXT_CAP_ID_PMUX 0x1A /* Protocol Multiplexing */
#define PCI_EXT_CAP_ID_PASID 0x1B /* Process Address Space ID */
#define PCI_EXT_CAP_ID_DPC 0x1D /* Downstream Port Containment */
+#define PCI_EXT_CAP_ID_L1SS 0x1E /* L1 PM Substates */
#define PCI_EXT_CAP_ID_PTM 0x1F /* Precision Time Measurement */
#define PCI_EXT_CAP_ID_MAX PCI_EXT_CAP_ID_PTM
@@ -973,6 +974,7 @@
#define PCI_EXP_DPC_STATUS 8 /* DPC Status */
#define PCI_EXP_DPC_STATUS_TRIGGER 0x01 /* Trigger Status */
#define PCI_EXP_DPC_STATUS_INTERRUPT 0x08 /* Interrupt Status */
+#define PCI_EXP_DPC_RP_BUSY 0x10 /* Root Port Busy */
#define PCI_EXP_DPC_SOURCE_ID 10 /* DPC Source Identifier */
@@ -985,4 +987,19 @@
#define PCI_PTM_CTRL_ENABLE 0x00000001 /* PTM enable */
#define PCI_PTM_CTRL_ROOT 0x00000002 /* Root select */
+/* L1 PM Substates */
+#define PCI_L1SS_CAP 4 /* capability register */
+#define PCI_L1SS_CAP_PCIPM_L1_2 1 /* PCI PM L1.2 Support */
+#define PCI_L1SS_CAP_PCIPM_L1_1 2 /* PCI PM L1.1 Support */
+#define PCI_L1SS_CAP_ASPM_L1_2 4 /* ASPM L1.2 Support */
+#define PCI_L1SS_CAP_ASPM_L1_1 8 /* ASPM L1.1 Support */
+#define PCI_L1SS_CAP_L1_PM_SS 16 /* L1 PM Substates Support */
+#define PCI_L1SS_CTL1 8 /* Control Register 1 */
+#define PCI_L1SS_CTL1_PCIPM_L1_2 1 /* PCI PM L1.2 Enable */
+#define PCI_L1SS_CTL1_PCIPM_L1_1 2 /* PCI PM L1.1 Support */
+#define PCI_L1SS_CTL1_ASPM_L1_2 4 /* ASPM L1.2 Support */
+#define PCI_L1SS_CTL1_ASPM_L1_1 8 /* ASPM L1.1 Support */
+#define PCI_L1SS_CTL1_L1SS_MASK 0x0000000F
+#define PCI_L1SS_CTL2 0xC /* Control Register 2 */
+
#endif /* LINUX_PCI_REGS_H */
