habanalabs: add uapi to retrieve engines status

Currently, to get engines status, user needed to read debugfs file
with root permissions.

This new uapi allows user apace apps retrieve status, so for example,
in case of failure, status can be retrieved immediately by the
application itself which runs without root permissions.

Signed-off-by: Dani Liberman <dliberman@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

diff --git a/drivers/misc/habanalabs/common/debugfs.c b/drivers/misc/habanalabs/common/debugfs.c
index 90c91c1b2c1020c0e35332ce8196f56d60f7d493..c297248748d31f23611dee194d6a7e8762736767 100644 (file)
--- a/drivers/misc/habanalabs/common/debugfs.c
+++ b/drivers/misc/habanalabs/common/debugfs.c
@@ -17,7 +17,6 @@
 #define MMU_ASID_BUF_SIZE      10
 #define MMU_KBUF_SIZE          (MMU_ADDR_BUF_SIZE + MMU_ASID_BUF_SIZE)
 #define I2C_MAX_TRANSACTION_LEN        8
-#define ENGINES_DATA_MAX_SIZE  SZ_16K
 
 static struct dentry *hl_debug_root;
 
@@ -626,7 +625,7 @@ static int engines_show(struct seq_file *s, void *data)
        }
 
        eng_data.actual_size = 0;
-       eng_data.allocated_buf_size = ENGINES_DATA_MAX_SIZE;
+       eng_data.allocated_buf_size = HL_ENGINES_DATA_MAX_SIZE;
        eng_data.buf = vmalloc(eng_data.allocated_buf_size);
        if (!eng_data.buf)
                return -ENOMEM;

diff --git a/drivers/misc/habanalabs/common/habanalabs_ioctl.c b/drivers/misc/habanalabs/common/habanalabs_ioctl.c
index 6a30bd98ab5e771d171dba642b69b3d6130bec72..ec55c66fedd6b99f186a0ea2ca0cd90cf84a2e53 100644 (file)
--- a/drivers/misc/habanalabs/common/habanalabs_ioctl.c
+++ b/drivers/misc/habanalabs/common/habanalabs_ioctl.c
@@ -14,6 +14,7 @@
 #include <linux/fs.h>
 #include <linux/uaccess.h>
 #include <linux/slab.h>
+#include <linux/vmalloc.h>
 
 static u32 hl_debug_struct_size[HL_DEBUG_OP_TIMESTAMP + 1] = {
        [HL_DEBUG_OP_ETR] = sizeof(struct hl_debug_params_etr),
@@ -697,6 +698,42 @@ static int eventfd_unregister(struct hl_fpriv *hpriv, struct hl_info_args *args)
        return 0;
 }
 
+static int engine_status_info(struct hl_fpriv *hpriv, struct hl_info_args *args)
+{
+       void __user *out = (void __user *) (uintptr_t) args->return_pointer;
+       u32 status_buf_size = args->return_size;
+       struct hl_device *hdev = hpriv->hdev;
+       struct engines_data eng_data;
+       int rc;
+
+       if ((status_buf_size < SZ_1K) || (status_buf_size > HL_ENGINES_DATA_MAX_SIZE) || (!out))
+               return -EINVAL;
+
+       eng_data.actual_size = 0;
+       eng_data.allocated_buf_size = status_buf_size;
+       eng_data.buf = vmalloc(status_buf_size);
+       if (!eng_data.buf)
+               return -ENOMEM;
+
+       hdev->asic_funcs->is_device_idle(hdev, NULL, 0, &eng_data);
+
+       if (eng_data.actual_size > eng_data.allocated_buf_size) {
+               dev_err(hdev->dev,
+                       "Engines data size (%d Bytes) is bigger than allocated size (%u Bytes)\n",
+                       eng_data.actual_size, status_buf_size);
+               vfree(eng_data.buf);
+               return -ENOMEM;
+       }
+
+       args->user_buffer_actual_size = eng_data.actual_size;
+       rc = copy_to_user(out, eng_data.buf, min_t(size_t, status_buf_size, eng_data.actual_size)) ?
+                               -EFAULT : 0;
+
+       vfree(eng_data.buf);
+
+       return rc;
+}
+
 static int _hl_info_ioctl(struct hl_fpriv *hpriv, void *data,
                                struct device *dev)
 {
@@ -812,6 +849,9 @@ static int _hl_info_ioctl(struct hl_fpriv *hpriv, void *data,
        case HL_INFO_UNREGISTER_EVENTFD:
                return eventfd_unregister(hpriv, args);
 
+       case HL_INFO_ENGINE_STATUS:
+               return engine_status_info(hpriv, args);
+
        default:
                dev_err(dev, "Invalid request %d\n", args->op);
                rc = -EINVAL;

diff --git a/include/uapi/misc/habanalabs.h b/include/uapi/misc/habanalabs.h
index be06b1307c4464c280fb72fcf47be4a2286309dc..83ca6f40f4ba00fe4b5bb11b846db7f80b12069c 100644 (file)
--- a/include/uapi/misc/habanalabs.h
+++ b/include/uapi/misc/habanalabs.h
@@ -787,10 +787,14 @@ enum hl_server_type {
 #define HL_INFO_UNREGISTER_EVENTFD             29
 #define HL_INFO_GET_EVENTS                     30
 #define HL_INFO_UNDEFINED_OPCODE_EVENT         31
+#define HL_INFO_ENGINE_STATUS                  32
 
 #define HL_INFO_VERSION_MAX_LEN                        128
 #define HL_INFO_CARD_NAME_MAX_LEN              16
 
+/* Maximum buffer size for retrieving engines status */
+#define HL_ENGINES_DATA_MAX_SIZE       SZ_1M
+
 /**
  * struct hl_info_hw_ip_info - hardware information on various IPs in the ASIC
  * @sram_base_address: The first SRAM physical base address that is free to be
@@ -1130,6 +1134,10 @@ enum gaudi_dcores {
  *             resolution. Currently not in use.
  * @pll_index: Index as defined in hl_<asic type>_pll_index enumeration.
  * @eventfd: event file descriptor for event notifications.
+ * @user_buffer_actual_size: Actual data size which was copied to user allocated buffer by the
+ *                           driver. It is possible for the user to allocate buffer larger than
+ *                           needed, hence updating this variable so user will know the exact amount
+ *                           of bytes copied by the kernel to the buffer.
  * @pad: Padding to 64 bit.
  */
 struct hl_info_args {
@@ -1143,6 +1151,7 @@ struct hl_info_args {
                __u32 period_ms;
                __u32 pll_index;
                __u32 eventfd;
+               __u32 user_buffer_actual_size;
        };
 
        __u32 pad;