habanalabs: perform context switch flow only if needed

Except Goya, none of our ASICs require context switch flow, hence we
enable this flow only where it is needed.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

diff --git a/drivers/misc/habanalabs/common/command_submission.c b/drivers/misc/habanalabs/common/command_submission.c
index cf411851567876368d58c63d1d93001ace4ce35c..746b688d34cf378eac4b2c6ad88305ac8f444a24 100644 (file)
--- a/drivers/misc/habanalabs/common/command_submission.c
+++ b/drivers/misc/habanalabs/common/command_submission.c
@@ -1590,13 +1590,14 @@ static int hl_cs_ctx_switch(struct hl_fpriv *hpriv, union hl_cs_args *args,
        struct hl_device *hdev = hpriv->hdev;
        struct hl_ctx *ctx = hpriv->ctx;
        bool need_soft_reset = false;
-       int rc = 0, do_ctx_switch;
+       int rc = 0, do_ctx_switch = 0;
        void __user *chunks;
        u32 num_chunks, tmp;
        u16 sob_count;
        int ret;
 
-       do_ctx_switch = atomic_cmpxchg(&ctx->thread_ctx_switch_token, 1, 0);
+       if (hdev->supports_ctx_switch)
+               do_ctx_switch = atomic_cmpxchg(&ctx->thread_ctx_switch_token, 1, 0);
 
        if (do_ctx_switch || (args->in.cs_flags & HL_CS_FLAGS_FORCE_RESTORE)) {
                mutex_lock(&hpriv->restore_phase_mutex);
@@ -1667,9 +1668,10 @@ wait_again:
                        }
                }
 
-               ctx->thread_ctx_switch_wait_token = 1;
+               if (hdev->supports_ctx_switch)
+                       ctx->thread_ctx_switch_wait_token = 1;
 
-       } else if (!ctx->thread_ctx_switch_wait_token) {
+       } else if (hdev->supports_ctx_switch && !ctx->thread_ctx_switch_wait_token) {
                rc = hl_poll_timeout_memory(hdev,
                        &ctx->thread_ctx_switch_wait_token, tmp, (tmp == 1),
                        100, jiffies_to_usecs(hdev->timeout_jiffies), false);

diff --git a/drivers/misc/habanalabs/common/habanalabs.h b/drivers/misc/habanalabs/common/habanalabs.h
index d7fd4f57abf368540e82cd1874430d818ca31610..33c6476b60a96a8d637089cc8f55e027cc0a84c9 100644 (file)
--- a/drivers/misc/habanalabs/common/habanalabs.h
+++ b/drivers/misc/habanalabs/common/habanalabs.h
@@ -3199,6 +3199,7 @@ struct hl_reset_info {
  *                         Used only for testing.
  * @heartbeat: Controls if we want to enable the heartbeat mechanism vs. the f/w, which verifies
  *             that the f/w is always alive. Used only for testing.
+ * @supports_ctx_switch: true if a ctx switch is required upon first submission.
  */
 struct hl_device {
        struct pci_dev                  *pdev;
@@ -3335,6 +3336,7 @@ struct hl_device {
        u8                              compute_ctx_in_release;
        u8                              supports_mmu_prefetch;
        u8                              reset_upon_device_release;
+       u8                              supports_ctx_switch;
 
        /* Parameters for bring-up */
        u64                             nic_ports_mask;

diff --git a/drivers/misc/habanalabs/goya/goya.c b/drivers/misc/habanalabs/goya/goya.c
index 87465a28af0d5aa80913aace998e402bf793fe27..d8fb91d257b9e20d9c1f8d58d30d8f6e59aba5dd 100644 (file)
--- a/drivers/misc/habanalabs/goya/goya.c
+++ b/drivers/misc/habanalabs/goya/goya.c
@@ -1025,6 +1025,7 @@ static int goya_sw_init(struct hl_device *hdev)
        hdev->asic_prop.supports_compute_reset = true;
        hdev->asic_prop.allow_inference_soft_reset = true;
        hdev->supports_wait_for_multi_cs = false;
+       hdev->supports_ctx_switch = true;
 
        hdev->asic_funcs->set_pci_memory_regions(hdev);