openkylin
/
linux
mirror of https://gitee.com/openkylin/linux.git
9
0
Fork 0
Code Issues Projects Releases Wiki Activity

habanalabs: add 'needs reset' state in driver 

The new state indicates that device should be reset in order
to re-gain funcionality.
This unique state can occur if reset_on_lockup is disabled
and an actual lockup has occurred.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
This commit is contained in:
Ofir Bitton 2020-10-05 14:40:10 +03:00 committed by Oded Gabbay
parent f2d032ee13
commit 66a76401c5
14 changed files with 101 additions and 55 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
drivers/misc/habanalabs/common/command_buffer.c
View File

@ -379,13 +379,14 @@ int hl_cb_ioctl(struct hl_fpriv *hpriv, void *data)
{
union hl_cb_args *args = data;
struct hl_device *hdev = hpriv->hdev;
enum hl_device_status status;
u64 handle = 0;
int rc;
if (hl_device_disabled_or_in_reset(hdev)) {
if (!hl_device_operational(hdev, &status)) {
dev_warn_ratelimited(hdev->dev,
"Device is %s. Can't execute CB IOCTL\n",
atomic_read(&hdev->in_reset) ? "in_reset" : "disabled");
hdev->status[status]);
return -EBUSY;
}

7
drivers/misc/habanalabs/common/command_submission.c
View File

@ -427,6 +427,8 @@ static void cs_timedout(struct work_struct *work)
if (hdev->reset_on_lockup)
hl_device_reset(hdev, false, false);
else
hdev->needs_reset = true;
}
static int allocate_cs(struct hl_device *hdev, struct hl_ctx *ctx,
@ -689,12 +691,13 @@ static int hl_cs_sanity_checks(struct hl_fpriv *hpriv, union hl_cs_args *args)
struct hl_device *hdev = hpriv->hdev;
struct hl_ctx *ctx = hpriv->ctx;
u32 cs_type_flags, num_chunks;
enum hl_device_status status;
enum hl_cs_type cs_type;
if (hl_device_disabled_or_in_reset(hdev)) {
if (!hl_device_operational(hdev, &status)) {
dev_warn_ratelimited(hdev->dev,
"Device is %s. Can't submit new CS\n",
atomic_read(&hdev->in_reset) ? "in_reset" : "disabled");
hdev->status[status]);
return -EBUSY;
}

6
drivers/misc/habanalabs/common/debugfs.c
View File

@ -24,7 +24,7 @@ static int hl_debugfs_i2c_read(struct hl_device *hdev, u8 i2c_bus, u8 i2c_addr,
struct cpucp_packet pkt;
int rc;
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
return -EBUSY;
memset(&pkt, 0, sizeof(pkt));
@ -50,7 +50,7 @@ static int hl_debugfs_i2c_write(struct hl_device *hdev, u8 i2c_bus, u8 i2c_addr,
struct cpucp_packet pkt;
int rc;
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
return -EBUSY;
memset(&pkt, 0, sizeof(pkt));
@ -76,7 +76,7 @@ static void hl_debugfs_led_set(struct hl_device *hdev, u8 led, u8 state)
struct cpucp_packet pkt;
int rc;
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
return;
memset(&pkt, 0, sizeof(pkt));

33
drivers/misc/habanalabs/common/device.c
View File

@ -15,14 +15,6 @@
#define HL_PLDM_PENDING_RESET_PER_SEC (HL_PENDING_RESET_PER_SEC * 10)
bool hl_device_disabled_or_in_reset(struct hl_device *hdev)
{
if ((hdev->disabled) || (atomic_read(&hdev->in_reset)))
return true;
else
return false;
}
enum hl_device_status hl_device_status(struct hl_device *hdev)
{
enum hl_device_status status;
@ -31,12 +23,34 @@ enum hl_device_status hl_device_status(struct hl_device *hdev)
status = HL_DEVICE_STATUS_MALFUNCTION;
else if (atomic_read(&hdev->in_reset))
status = HL_DEVICE_STATUS_IN_RESET;
else if (hdev->needs_reset)
status = HL_DEVICE_STATUS_NEEDS_RESET;
else
status = HL_DEVICE_STATUS_OPERATIONAL;
return status;
}
bool hl_device_operational(struct hl_device *hdev,
enum hl_device_status *status)
{
enum hl_device_status current_status;
current_status = hl_device_status(hdev);
if (status)
*status = current_status;
switch (current_status) {
case HL_DEVICE_STATUS_IN_RESET:
case HL_DEVICE_STATUS_MALFUNCTION:
case HL_DEVICE_STATUS_NEEDS_RESET:
return false;
case HL_DEVICE_STATUS_OPERATIONAL:
default:
return true;
}
}
static void hpriv_release(struct kref *ref)
{
struct hl_fpriv *hpriv;
@ -411,7 +425,7 @@ static void hl_device_heartbeat(struct work_struct *work)
struct hl_device *hdev = container_of(work, struct hl_device,
work_heartbeat.work);
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
goto reschedule;
if (!hdev->asic_funcs->send_heartbeat(hdev))
@ -1091,6 +1105,7 @@ int hl_device_reset(struct hl_device *hdev, bool hard_reset,
}
atomic_set(&hdev->in_reset, 0);
hdev->needs_reset = false;
if (hard_reset)
hdev->hard_reset_cnt++;

14
drivers/misc/habanalabs/common/habanalabs.h
View File

@ -1432,6 +1432,10 @@ struct hl_dbg_device_entry {
* DEVICES
*/
#define HL_STR_MAX 32
#define HL_DEV_STS_MAX (HL_DEVICE_STATUS_NEEDS_RESET + 1)
/* Theoretical limit only. A single host can only contain up to 4 or 8 PCIe
* x16 cards. In extreme cases, there are hosts that can accommodate 16 cards.
*/
@ -1706,6 +1710,7 @@ struct hl_mmu_funcs {
* @hwmon_dev: H/W monitor device.
* @pm_mng_profile: current power management profile.
* @hl_chip_info: ASIC's sensors information.
* @device_status_description: device status description.
* @hl_debugfs: device's debugfs manager.
* @cb_pool: list of preallocated CBs.
* @cb_pool_lock: protects the CB pool.
@ -1774,6 +1779,8 @@ struct hl_mmu_funcs {
* @supports_coresight: is CoreSight supported.
* @supports_soft_reset: is soft reset supported.
* @supports_cb_mapping: is mapping a CB to the device's MMU supported.
* @needs_reset: true if reset_on_lockup is false and device should be reset
* due to lockup.
*/
struct hl_device {
struct pci_dev *pdev;
@ -1786,7 +1793,8 @@ struct hl_device {
struct device *dev_ctrl;
struct delayed_work work_freq;
struct delayed_work work_heartbeat;
char asic_name[32];
char asic_name[HL_STR_MAX];
char status[HL_DEV_STS_MAX][HL_STR_MAX];
enum hl_asic_type asic_type;
struct hl_cq *completion_queue;
struct workqueue_struct **cq_wq;
@ -1876,6 +1884,7 @@ struct hl_device {
u8 supports_coresight;
u8 supports_soft_reset;
u8 supports_cb_mapping;
u8 needs_reset;
/* Parameters for bring-up */
u64 nic_ports_mask;
@ -1978,7 +1987,8 @@ static inline bool hl_mem_area_crosses_range(u64 address, u32 size,
int hl_device_open(struct inode *inode, struct file *filp);
int hl_device_open_ctrl(struct inode *inode, struct file *filp);
bool hl_device_disabled_or_in_reset(struct hl_device *hdev);
bool hl_device_operational(struct hl_device *hdev,
enum hl_device_status *status);
enum hl_device_status hl_device_status(struct hl_device *hdev);
int hl_device_set_debug_mode(struct hl_device *hdev, bool enable);
int create_hdev(struct hl_device **dev, struct pci_dev *pdev,

17
drivers/misc/habanalabs/common/habanalabs_drv.c
View File

@ -92,6 +92,7 @@ static enum hl_asic_type get_asic_type(u16 device)
*/
int hl_device_open(struct inode *inode, struct file *filp)
{
enum hl_device_status status;
struct hl_device *hdev;
struct hl_fpriv *hpriv;
int rc;
@ -124,10 +125,10 @@ int hl_device_open(struct inode *inode, struct file *filp)
mutex_lock(&hdev->fpriv_list_lock);
if (hl_device_disabled_or_in_reset(hdev)) {
if (!hl_device_operational(hdev, &status)) {
dev_err_ratelimited(hdev->dev,
"Can't open %s because it is disabled or in reset\n",
dev_name(hdev->dev));
"Can't open %s because it is %s\n",
dev_name(hdev->dev), hdev->status[status]);
rc = -EPERM;
goto out_err;
}
@ -204,7 +205,7 @@ int hl_device_open_ctrl(struct inode *inode, struct file *filp)
mutex_lock(&hdev->fpriv_list_lock);
if (hl_device_disabled_or_in_reset(hdev)) {
if (!hl_device_operational(hdev, NULL)) {
dev_err_ratelimited(hdev->dev_ctrl,
"Can't open %s because it is disabled or in reset\n",
dev_name(hdev->dev_ctrl));
@ -287,6 +288,14 @@ int create_hdev(struct hl_device **dev, struct pci_dev *pdev,
hdev->asic_type = asic_type;
}
/* Assign status description string */
strncpy(hdev->status[HL_DEVICE_STATUS_MALFUNCTION],
"disabled", HL_STR_MAX);
strncpy(hdev->status[HL_DEVICE_STATUS_IN_RESET],
"in reset", HL_STR_MAX);
strncpy(hdev->status[HL_DEVICE_STATUS_NEEDS_RESET],
"needs reset", HL_STR_MAX);
hdev->major = hl_major;
hdev->reset_on_lockup = reset_on_lockup;
hdev->memory_scrub = memory_scrub;

12
drivers/misc/habanalabs/common/habanalabs_ioctl.c
View File

@ -406,8 +406,10 @@ static int total_energy_consumption_info(struct hl_fpriv *hpriv,
static int _hl_info_ioctl(struct hl_fpriv *hpriv, void *data,
struct device *dev)
{
enum hl_device_status status;
struct hl_info_args *args = data;
struct hl_device *hdev = hpriv->hdev;
int rc;
/*
@ -428,10 +430,10 @@ static int _hl_info_ioctl(struct hl_fpriv *hpriv, void *data,
break;
}
if (hl_device_disabled_or_in_reset(hdev)) {
if (!hl_device_operational(hdev, &status)) {
dev_warn_ratelimited(dev,
"Device is %s. Can't execute INFO IOCTL\n",
atomic_read(&hdev->in_reset) ? "in_reset" : "disabled");
hdev->status[status]);
return -EBUSY;
}
@ -501,12 +503,14 @@ static int hl_debug_ioctl(struct hl_fpriv *hpriv, void *data)
{
struct hl_debug_args *args = data;
struct hl_device *hdev = hpriv->hdev;
enum hl_device_status status;
int rc = 0;
if (hl_device_disabled_or_in_reset(hdev)) {
if (!hl_device_operational(hdev, &status)) {
dev_warn_ratelimited(hdev->dev,
"Device is %s. Can't execute DEBUG IOCTL\n",
atomic_read(&hdev->in_reset) ? "in_reset" : "disabled");
hdev->status[status]);
return -EBUSY;
}

6
drivers/misc/habanalabs/common/hw_queue.c
View File

@ -515,6 +515,7 @@ static void init_signal_wait_cs(struct hl_cs *cs)
*/
int hl_hw_queue_schedule_cs(struct hl_cs *cs)
{
enum hl_device_status status;
struct hl_cs_counters_atomic *cntr;
struct hl_ctx *ctx = cs->ctx;
struct hl_device *hdev = ctx->hdev;
@ -527,11 +528,10 @@ int hl_hw_queue_schedule_cs(struct hl_cs *cs)
hdev->asic_funcs->hw_queues_lock(hdev);
if (hl_device_disabled_or_in_reset(hdev)) {
if (!hl_device_operational(hdev, &status)) {
atomic64_inc(&ctx->cs_counters.device_in_reset_drop_cnt);
atomic64_inc(&cntr->device_in_reset_drop_cnt);
dev_err(hdev->dev,
"device is disabled or in reset, CS rejected!\n");
"device is %s, CS rejected!\n", hdev->status[status]);
rc = -EPERM;
goto out;
}

4
drivers/misc/habanalabs/common/hwmon.c
View File

@ -114,7 +114,7 @@ static int hl_read(struct device *dev, enum hwmon_sensor_types type,
struct hl_device *hdev = dev_get_drvdata(dev);
int rc;
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
return -ENODEV;
switch (type) {
@ -192,7 +192,7 @@ static int hl_write(struct device *dev, enum hwmon_sensor_types type,
{
struct hl_device *hdev = dev_get_drvdata(dev);
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
return -ENODEV;
switch (type) {

5
drivers/misc/habanalabs/common/memory.c
View File

@ -1237,6 +1237,7 @@ static int mem_ioctl_no_mmu(struct hl_fpriv *hpriv, union hl_mem_args *args)
int hl_mem_ioctl(struct hl_fpriv *hpriv, void *data)
{
enum hl_device_status status;
union hl_mem_args *args = data;
struct hl_device *hdev = hpriv->hdev;
struct hl_ctx *ctx = hpriv->ctx;
@ -1244,10 +1245,10 @@ int hl_mem_ioctl(struct hl_fpriv *hpriv, void *data)
u32 handle = 0;
int rc;
if (hl_device_disabled_or_in_reset(hdev)) {
if (!hl_device_operational(hdev, &status)) {
dev_warn_ratelimited(hdev->dev,
"Device is %s. Can't execute MEMORY IOCTL\n",
atomic_read(&hdev->in_reset) ? "in_reset" : "disabled");
hdev->status[status]);
return -EBUSY;
}

8
drivers/misc/habanalabs/common/sysfs.c
View File

@ -276,6 +276,8 @@ static ssize_t status_show(struct device *dev, struct device_attribute *attr,
str = "In reset";
else if (hdev->disabled)
str = "Malfunction";
else if (hdev->needs_reset)
str = "Needs Reset";
else
str = "Operational";
@ -304,7 +306,7 @@ static ssize_t max_power_show(struct device *dev, struct device_attribute *attr,
struct hl_device *hdev = dev_get_drvdata(dev);
long val;
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
return -ENODEV;
val = hl_get_max_power(hdev);
@ -319,7 +321,7 @@ static ssize_t max_power_store(struct device *dev,
unsigned long value;
int rc;
if (hl_device_disabled_or_in_reset(hdev)) {
if (!hl_device_operational(hdev, NULL)) {
count = -ENODEV;
goto out;
}
@ -347,7 +349,7 @@ static ssize_t eeprom_read_handler(struct file *filp, struct kobject *kobj,
char *data;
int rc;
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
return -ENODEV;
if (!max_size)

8
drivers/misc/habanalabs/gaudi/gaudi_hwmgr.c
View File

@ -20,7 +20,7 @@ int gaudi_get_clk_rate(struct hl_device *hdev, u32 *cur_clk, u32 *max_clk)
{
long value;
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
return -ENODEV;
value = hl_get_frequency(hdev, MME_PLL, false);
@ -54,7 +54,7 @@ static ssize_t clk_max_freq_mhz_show(struct device *dev,
struct gaudi_device *gaudi = hdev->asic_specific;
long value;
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
return -ENODEV;
value = hl_get_frequency(hdev, MME_PLL, false);
@ -72,7 +72,7 @@ static ssize_t clk_max_freq_mhz_store(struct device *dev,
int rc;
u64 value;
if (hl_device_disabled_or_in_reset(hdev)) {
if (!hl_device_operational(hdev, NULL)) {
count = -ENODEV;
goto fail;
}
@ -97,7 +97,7 @@ static ssize_t clk_cur_freq_mhz_show(struct device *dev,
struct hl_device *hdev = dev_get_drvdata(dev);
long value;
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
return -ENODEV;
value = hl_get_frequency(hdev, MME_PLL, true);

28
drivers/misc/habanalabs/goya/goya_hwmgr.c
View File

@ -36,7 +36,7 @@ int goya_get_clk_rate(struct hl_device *hdev, u32 *cur_clk, u32 *max_clk)
{
long value;
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
return -ENODEV;
value = hl_get_frequency(hdev, MME_PLL, false);
@ -69,7 +69,7 @@ static ssize_t mme_clk_show(struct device *dev, struct device_attribute *attr,
struct hl_device *hdev = dev_get_drvdata(dev);
long value;
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
return -ENODEV;
value = hl_get_frequency(hdev, MME_PLL, false);
@ -88,7 +88,7 @@ static ssize_t mme_clk_store(struct device *dev, struct device_attribute *attr,
int rc;
long value;
if (hl_device_disabled_or_in_reset(hdev)) {
if (!hl_device_operational(hdev, NULL)) {
count = -ENODEV;
goto fail;
}
@ -118,7 +118,7 @@ static ssize_t tpc_clk_show(struct device *dev, struct device_attribute *attr,
struct hl_device *hdev = dev_get_drvdata(dev);
long value;
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
return -ENODEV;
value = hl_get_frequency(hdev, TPC_PLL, false);
@ -137,7 +137,7 @@ static ssize_t tpc_clk_store(struct device *dev, struct device_attribute *attr,
int rc;
long value;
if (hl_device_disabled_or_in_reset(hdev)) {
if (!hl_device_operational(hdev, NULL)) {
count = -ENODEV;
goto fail;
}
@ -167,7 +167,7 @@ static ssize_t ic_clk_show(struct device *dev, struct device_attribute *attr,
struct hl_device *hdev = dev_get_drvdata(dev);
long value;
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
return -ENODEV;
value = hl_get_frequency(hdev, IC_PLL, false);
@ -186,7 +186,7 @@ static ssize_t ic_clk_store(struct device *dev, struct device_attribute *attr,
int rc;
long value;
if (hl_device_disabled_or_in_reset(hdev)) {
if (!hl_device_operational(hdev, NULL)) {
count = -ENODEV;
goto fail;
}
@ -216,7 +216,7 @@ static ssize_t mme_clk_curr_show(struct device *dev,
struct hl_device *hdev = dev_get_drvdata(dev);
long value;
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
return -ENODEV;
value = hl_get_frequency(hdev, MME_PLL, true);
@ -233,7 +233,7 @@ static ssize_t tpc_clk_curr_show(struct device *dev,
struct hl_device *hdev = dev_get_drvdata(dev);
long value;
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
return -ENODEV;
value = hl_get_frequency(hdev, TPC_PLL, true);
@ -250,7 +250,7 @@ static ssize_t ic_clk_curr_show(struct device *dev,
struct hl_device *hdev = dev_get_drvdata(dev);
long value;
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
return -ENODEV;
value = hl_get_frequency(hdev, IC_PLL, true);
@ -266,7 +266,7 @@ static ssize_t pm_mng_profile_show(struct device *dev,
{
struct hl_device *hdev = dev_get_drvdata(dev);
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
return -ENODEV;
return sprintf(buf, "%s\n",
@ -280,7 +280,7 @@ static ssize_t pm_mng_profile_store(struct device *dev,
{
struct hl_device *hdev = dev_get_drvdata(dev);
if (hl_device_disabled_or_in_reset(hdev)) {
if (!hl_device_operational(hdev, NULL)) {
count = -ENODEV;
goto out;
}
@ -335,7 +335,7 @@ static ssize_t high_pll_show(struct device *dev, struct device_attribute *attr,
{
struct hl_device *hdev = dev_get_drvdata(dev);
if (hl_device_disabled_or_in_reset(hdev))
if (!hl_device_operational(hdev, NULL))
return -ENODEV;
return sprintf(buf, "%u\n", hdev->high_pll);
@ -348,7 +348,7 @@ static ssize_t high_pll_store(struct device *dev, struct device_attribute *attr,
long value;
int rc;
if (hl_device_disabled_or_in_reset(hdev)) {
if (!hl_device_operational(hdev, NULL)) {
count = -ENODEV;
goto out;
}

3
include/uapi/misc/habanalabs.h
View File

@ -242,7 +242,8 @@ enum gaudi_engine_id {
enum hl_device_status {
HL_DEVICE_STATUS_OPERATIONAL,
HL_DEVICE_STATUS_IN_RESET,
HL_DEVICE_STATUS_MALFUNCTION
HL_DEVICE_STATUS_MALFUNCTION,
HL_DEVICE_STATUS_NEEDS_RESET
};
/* Opcode for management ioctl