From: Ofir Bitton <obit...@habana.ai>
In order for reserving VA ranges for kernel memory, we need
to allow the VM module to be initiated with kernel context.
Signed-off-by: Ofir Bitton <obit...@habana.ai>
Reviewed-by: Oded Gabbay <ogab...@kernel.org>
Signed-off-by: Oded Gabbay <ogab...@kernel.org>
---
drivers/misc/habanalabs/common/context.c | 16 +++++++++++++---
drivers/misc/habanalabs/common/device.c | 15 ++++++++++-----
drivers/misc/habanalabs/common/memory.c | 3 ++-
drivers/misc/habanalabs/gaudi/gaudi.c | 10 ++++------
4 files changed, 29 insertions(+), 15 deletions(-)
diff --git a/drivers/misc/habanalabs/common/context.c
b/drivers/misc/habanalabs/common/context.c
index f65e6559149b..3d86b83f4ca6 100644
--- a/drivers/misc/habanalabs/common/context.c
+++ b/drivers/misc/habanalabs/common/context.c
@@ -56,6 +56,8 @@ static void hl_ctx_fini(struct hl_ctx *ctx)
idle_mask);
} else {
dev_dbg(hdev->dev, "closing kernel context\n");
+ hdev->asic_funcs->ctx_fini(ctx);
+ hl_vm_ctx_fini(ctx);
hl_mmu_ctx_fini(ctx);
}
}
@@ -151,11 +153,18 @@ int hl_ctx_init(struct hl_device *hdev, struct hl_ctx
*ctx, bool is_kernel_ctx)
if (is_kernel_ctx) {
ctx->asid = HL_KERNEL_ASID_ID; /* Kernel driver gets ASID 0 */
- rc = hl_mmu_ctx_init(ctx);
+ rc = hl_vm_ctx_init(ctx);
if (rc) {
- dev_err(hdev->dev, "Failed to init mmu ctx module\n");
+ dev_err(hdev->dev, "Failed to init mem ctx module\n");
+ rc = -ENOMEM;
goto err_free_cs_pending;
}
+
+ rc = hdev->asic_funcs->ctx_init(ctx);
+ if (rc) {
+ dev_err(hdev->dev, "ctx_init failed\n");
+ goto err_vm_ctx_fini;
+ }
} else {
ctx->asid = hl_asid_alloc(hdev);
if (!ctx->asid) {
@@ -194,7 +203,8 @@ int hl_ctx_init(struct hl_device *hdev, struct hl_ctx *ctx,
bool is_kernel_ctx)
err_vm_ctx_fini:
hl_vm_ctx_fini(ctx);
err_asid_free:
- hl_asid_free(hdev, ctx->asid);
+ if (ctx->asid != HL_KERNEL_ASID_ID)
+ hl_asid_free(hdev, ctx->asid);
err_free_cs_pending:
kfree(ctx->cs_pending);
diff --git a/drivers/misc/habanalabs/common/device.c
b/drivers/misc/habanalabs/common/device.c
index 59212706a836..671d1eb41146 100644
--- a/drivers/misc/habanalabs/common/device.c
+++ b/drivers/misc/habanalabs/common/device.c
@@ -1310,11 +1310,16 @@ int hl_device_init(struct hl_device *hdev, struct class
*hclass)
hdev->compute_ctx = NULL;
+ hl_debugfs_add_device(hdev);
+
+ /* debugfs nodes are created in hl_ctx_init so it must be called after
+ * hl_debugfs_add_device.
+ */
rc = hl_ctx_init(hdev, hdev->kernel_ctx, true);
if (rc) {
dev_err(hdev->dev, "failed to initialize kernel context\n");
kfree(hdev->kernel_ctx);
- goto mmu_fini;
+ goto remove_device_from_debugfs;
}
rc = hl_cb_pool_init(hdev);
@@ -1323,8 +1328,6 @@ int hl_device_init(struct hl_device *hdev, struct class
*hclass)
goto release_ctx;
}
- hl_debugfs_add_device(hdev);
-
/*
* From this point, in case of an error, add char devices and create
* sysfs nodes as part of the error flow, to allow debugging.
@@ -1413,6 +1416,8 @@ int hl_device_init(struct hl_device *hdev, struct class
*hclass)
if (hl_ctx_put(hdev->kernel_ctx) != 1)
dev_err(hdev->dev,
"kernel ctx is still alive on initialization
failure\n");
+remove_device_from_debugfs:
+ hl_debugfs_remove_device(hdev);
mmu_fini:
hl_mmu_fini(hdev);
eq_fini:
@@ -1502,8 +1507,6 @@ void hl_device_fini(struct hl_device *hdev)
device_late_fini(hdev);
- hl_debugfs_remove_device(hdev);
-
/*
* Halt the engines and disable interrupts so we won't get any more
* completions from H/W and we won't have any accesses from the
@@ -1535,6 +1538,8 @@ void hl_device_fini(struct hl_device *hdev)
if ((hdev->kernel_ctx) && (hl_ctx_put(hdev->kernel_ctx) != 1))
dev_err(hdev->dev, "kernel ctx is still alive\n");
+ hl_debugfs_remove_device(hdev);
+
hl_vm_fini(hdev);
hl_mmu_fini(hdev);
diff --git a/drivers/misc/habanalabs/common/memory.c
b/drivers/misc/habanalabs/common/memory.c
index d4a94a1c3eef..c16c829bfbf2 100644
--- a/drivers/misc/habanalabs/common/memory.c
+++ b/drivers/misc/habanalabs/common/memory.c
@@ -1923,7 +1923,8 @@ void hl_vm_ctx_fini(struct hl_ctx *ctx)
* because the user notifies us on allocations. If the user is no more,
* all DRAM is available
*/
- if (!ctx->hdev->asic_prop.dram_supports_virtual_memory)
+ if (ctx->asid != HL_KERNEL_ASID_ID &&
+ !ctx->hdev->asic_prop.dram_supports_virtual_memory)
atomic64_set(&ctx->hdev->dram_used_mem, 0);
}
diff --git a/drivers/misc/habanalabs/gaudi/gaudi.c
b/drivers/misc/habanalabs/gaudi/gaudi.c
index 4776dfed9098..658d180d3d4d 100644
--- a/drivers/misc/habanalabs/gaudi/gaudi.c
+++ b/drivers/misc/habanalabs/gaudi/gaudi.c
@@ -7849,18 +7849,16 @@ static void gaudi_internal_cb_pool_fini(struct
hl_device *hdev,
static int gaudi_ctx_init(struct hl_ctx *ctx)
{
+ if (ctx->asid == HL_KERNEL_ASID_ID)
+ return 0;
+
gaudi_mmu_prepare(ctx->hdev, ctx->asid);
return gaudi_internal_cb_pool_init(ctx->hdev, ctx);
}
static void gaudi_ctx_fini(struct hl_ctx *ctx)
{
- struct hl_device *hdev = ctx->hdev;
-
- /* Gaudi will NEVER support more then a single compute context.
- * Therefore, don't clear anything unless it is the compute context
- */
- if (hdev->compute_ctx != ctx)
+ if (ctx->asid == HL_KERNEL_ASID_ID)
return;
gaudi_internal_cb_pool_fini(ctx->hdev, ctx);
--
2.17.1
