]> asedeno.scripts.mit.edu Git - linux.git/commitdiff
drm/amdgpu: add navi10 ih ip block (v3)
authorHawking Zhang <Hawking.Zhang@amd.com>
Sun, 3 Mar 2019 05:02:40 +0000 (13:02 +0800)
committerAlex Deucher <alexander.deucher@amd.com>
Fri, 21 Jun 2019 02:35:30 +0000 (21:35 -0500)
IH is the interrupt handler block.

v1: add initial ih support (Ray)
v2: add dummy prescreen iv function for navi10 (Hawking)
v3: squash in additional updates (Alex)

Signed-off-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drivers/gpu/drm/amd/amdgpu/Makefile
drivers/gpu/drm/amd/amdgpu/navi10_ih.c [new file with mode: 0644]
drivers/gpu/drm/amd/amdgpu/navi10_ih.h [new file with mode: 0644]

index 8e98e242bb7ff9958bce40c6b66611db298806a5..6e1ac38c99e0817c796d0e73586ce36c448e9dcd 100644 (file)
@@ -85,7 +85,8 @@ amdgpu-y += \
        iceland_ih.o \
        tonga_ih.o \
        cz_ih.o \
-       vega10_ih.o
+       vega10_ih.o \
+       navi10_ih.o
 
 # add PSP block
 amdgpu-y += \
diff --git a/drivers/gpu/drm/amd/amdgpu/navi10_ih.c b/drivers/gpu/drm/amd/amdgpu/navi10_ih.c
new file mode 100644 (file)
index 0000000..e963746
--- /dev/null
@@ -0,0 +1,486 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+
+#include "oss/osssys_5_0_0_offset.h"
+#include "oss/osssys_5_0_0_sh_mask.h"
+
+#include "soc15_common.h"
+#include "navi10_ih.h"
+
+
+static void navi10_ih_set_interrupt_funcs(struct amdgpu_device *adev);
+
+/**
+ * navi10_ih_enable_interrupts - Enable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Enable the interrupt ring buffer (NAVI10).
+ */
+static void navi10_ih_enable_interrupts(struct amdgpu_device *adev)
+{
+       u32 ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL);
+
+       ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 1);
+       ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, ENABLE_INTR, 1);
+       WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL, ih_rb_cntl);
+       adev->irq.ih.enabled = true;
+}
+
+/**
+ * navi10_ih_disable_interrupts - Disable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable the interrupt ring buffer (NAVI10).
+ */
+static void navi10_ih_disable_interrupts(struct amdgpu_device *adev)
+{
+       u32 ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL);
+
+       ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 0);
+       ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, ENABLE_INTR, 0);
+       WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL, ih_rb_cntl);
+       /* set rptr, wptr to 0 */
+       WREG32_SOC15(OSSSYS, 0, mmIH_RB_RPTR, 0);
+       WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR, 0);
+       adev->irq.ih.enabled = false;
+       adev->irq.ih.rptr = 0;
+}
+
+static uint32_t navi10_ih_rb_cntl(struct amdgpu_ih_ring *ih, uint32_t ih_rb_cntl)
+{
+       int rb_bufsz = order_base_2(ih->ring_size / 4);
+
+       ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+                                  MC_SPACE, ih->use_bus_addr ? 1 : 4);
+       ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+                                  WPTR_OVERFLOW_CLEAR, 1);
+       ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+                                  WPTR_OVERFLOW_ENABLE, 1);
+       ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_SIZE, rb_bufsz);
+       /* Ring Buffer write pointer writeback. If enabled, IH_RB_WPTR register
+        * value is written to memory
+        */
+       ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+                                  WPTR_WRITEBACK_ENABLE, 1);
+       ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_SNOOP, 1);
+       ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_RO, 0);
+       ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_VMID, 0);
+
+       return ih_rb_cntl;
+}
+
+/**
+ * navi10_ih_irq_init - init and enable the interrupt ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate a ring buffer for the interrupt controller,
+ * enable the RLC, disable interrupts, enable the IH
+ * ring buffer and enable it (NAVI).
+ * Called at device load and reume.
+ * Returns 0 for success, errors for failure.
+ */
+static int navi10_ih_irq_init(struct amdgpu_device *adev)
+{
+       struct amdgpu_ih_ring *ih = &adev->irq.ih;
+       int ret = 0;
+       u32 ih_rb_cntl, ih_doorbell_rtpr, ih_chicken;
+       u32 tmp;
+
+       /* disable irqs */
+       navi10_ih_disable_interrupts(adev);
+
+       adev->nbio_funcs->ih_control(adev);
+
+       /* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
+       WREG32_SOC15(OSSSYS, 0, mmIH_RB_BASE, ih->gpu_addr >> 8);
+       WREG32_SOC15(OSSSYS, 0, mmIH_RB_BASE_HI, (ih->gpu_addr >> 40) & 0xff);
+
+       ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL);
+       ih_rb_cntl = navi10_ih_rb_cntl(ih, ih_rb_cntl);
+       ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RPTR_REARM,
+                                  !!adev->irq.msi_enabled);
+
+       if (unlikely(adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT)) {
+               if (ih->use_bus_addr) {
+                       ih_chicken = RREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN);
+                       ih_chicken = REG_SET_FIELD(ih_chicken,
+                                       IH_CHICKEN, MC_SPACE_GPA_ENABLE, 1);
+                       WREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN, ih_chicken);
+               }
+       }
+
+       WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL, ih_rb_cntl);
+
+       /* set the writeback address whether it's enabled or not */
+       WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR_ADDR_LO,
+                    lower_32_bits(ih->wptr_addr));
+       WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR_ADDR_HI,
+                    upper_32_bits(ih->wptr_addr) & 0xFFFF);
+
+       /* set rptr, wptr to 0 */
+       WREG32_SOC15(OSSSYS, 0, mmIH_RB_RPTR, 0);
+       WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR, 0);
+
+       ih_doorbell_rtpr = RREG32_SOC15(OSSSYS, 0, mmIH_DOORBELL_RPTR);
+       if (ih->use_doorbell) {
+               ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+                                                IH_DOORBELL_RPTR, OFFSET,
+                                                ih->doorbell_index);
+               ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+                                                IH_DOORBELL_RPTR, ENABLE, 1);
+       } else {
+               ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+                                                IH_DOORBELL_RPTR, ENABLE, 0);
+       }
+       WREG32_SOC15(OSSSYS, 0, mmIH_DOORBELL_RPTR, ih_doorbell_rtpr);
+
+       adev->nbio_funcs->ih_doorbell_range(adev, ih->use_doorbell,
+                                           ih->doorbell_index);
+
+       tmp = RREG32_SOC15(OSSSYS, 0, mmIH_STORM_CLIENT_LIST_CNTL);
+       tmp = REG_SET_FIELD(tmp, IH_STORM_CLIENT_LIST_CNTL,
+                           CLIENT18_IS_STORM_CLIENT, 1);
+       WREG32_SOC15(OSSSYS, 0, mmIH_STORM_CLIENT_LIST_CNTL, tmp);
+
+       tmp = RREG32_SOC15(OSSSYS, 0, mmIH_INT_FLOOD_CNTL);
+       tmp = REG_SET_FIELD(tmp, IH_INT_FLOOD_CNTL, FLOOD_CNTL_ENABLE, 1);
+       WREG32_SOC15(OSSSYS, 0, mmIH_INT_FLOOD_CNTL, tmp);
+
+       pci_set_master(adev->pdev);
+
+       /* enable interrupts */
+       navi10_ih_enable_interrupts(adev);
+
+       return ret;
+}
+
+/**
+ * navi10_ih_irq_disable - disable interrupts
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable interrupts on the hw (NAVI10).
+ */
+static void navi10_ih_irq_disable(struct amdgpu_device *adev)
+{
+       navi10_ih_disable_interrupts(adev);
+
+       /* Wait and acknowledge irq */
+       mdelay(1);
+}
+
+/**
+ * navi10_ih_get_wptr - get the IH ring buffer wptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Get the IH ring buffer wptr from either the register
+ * or the writeback memory buffer (NAVI10).  Also check for
+ * ring buffer overflow and deal with it.
+ * Returns the value of the wptr.
+ */
+static u32 navi10_ih_get_wptr(struct amdgpu_device *adev,
+                             struct amdgpu_ih_ring *ih)
+{
+       u32 wptr, reg, tmp;
+
+       wptr = le32_to_cpu(*ih->wptr_cpu);
+
+       if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+               goto out;
+
+       reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR);
+       wptr = RREG32_NO_KIQ(reg);
+       if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+               goto out;
+       wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+
+       /* When a ring buffer overflow happen start parsing interrupt
+        * from the last not overwritten vector (wptr + 32). Hopefully
+        * this should allow us to catch up.
+        */
+       tmp = (wptr + 32) & ih->ptr_mask;
+       dev_warn(adev->dev, "IH ring buffer overflow "
+                "(0x%08X, 0x%08X, 0x%08X)\n",
+                wptr, ih->rptr, tmp);
+       ih->rptr = tmp;
+
+       reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL);
+       tmp = RREG32_NO_KIQ(reg);
+       tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+       WREG32_NO_KIQ(reg, tmp);
+out:
+       return (wptr & ih->ptr_mask);
+}
+
+/**
+ * navi10_ih_decode_iv - decode an interrupt vector
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Decodes the interrupt vector at the current rptr
+ * position and also advance the position.
+ */
+static void navi10_ih_decode_iv(struct amdgpu_device *adev,
+                               struct amdgpu_ih_ring *ih,
+                               struct amdgpu_iv_entry *entry)
+{
+       /* wptr/rptr are in bytes! */
+       u32 ring_index = ih->rptr >> 2;
+       uint32_t dw[8];
+
+       dw[0] = le32_to_cpu(ih->ring[ring_index + 0]);
+       dw[1] = le32_to_cpu(ih->ring[ring_index + 1]);
+       dw[2] = le32_to_cpu(ih->ring[ring_index + 2]);
+       dw[3] = le32_to_cpu(ih->ring[ring_index + 3]);
+       dw[4] = le32_to_cpu(ih->ring[ring_index + 4]);
+       dw[5] = le32_to_cpu(ih->ring[ring_index + 5]);
+       dw[6] = le32_to_cpu(ih->ring[ring_index + 6]);
+       dw[7] = le32_to_cpu(ih->ring[ring_index + 7]);
+
+       entry->client_id = dw[0] & 0xff;
+       entry->src_id = (dw[0] >> 8) & 0xff;
+       entry->ring_id = (dw[0] >> 16) & 0xff;
+       entry->vmid = (dw[0] >> 24) & 0xf;
+       entry->vmid_src = (dw[0] >> 31);
+       entry->timestamp = dw[1] | ((u64)(dw[2] & 0xffff) << 32);
+       entry->timestamp_src = dw[2] >> 31;
+       entry->pasid = dw[3] & 0xffff;
+       entry->pasid_src = dw[3] >> 31;
+       entry->src_data[0] = dw[4];
+       entry->src_data[1] = dw[5];
+       entry->src_data[2] = dw[6];
+       entry->src_data[3] = dw[7];
+
+       /* wptr/rptr are in bytes! */
+       ih->rptr += 32;
+}
+
+/**
+ * navi10_ih_set_rptr - set the IH ring buffer rptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set the IH ring buffer rptr.
+ */
+static void navi10_ih_set_rptr(struct amdgpu_device *adev,
+                              struct amdgpu_ih_ring *ih)
+{
+       if (ih->use_doorbell) {
+               /* XXX check if swapping is necessary on BE */
+               *ih->rptr_cpu = ih->rptr;
+               WDOORBELL32(ih->doorbell_index, ih->rptr);
+       } else
+               WREG32_SOC15(OSSSYS, 0, mmIH_RB_RPTR, ih->rptr);
+}
+
+static int navi10_ih_early_init(void *handle)
+{
+       struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+       navi10_ih_set_interrupt_funcs(adev);
+       return 0;
+}
+
+static int navi10_ih_sw_init(void *handle)
+{
+       int r;
+       struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+       bool use_bus_addr;
+
+       /* use gpu virtual address for ih ring
+        * until ih_checken is programmed to allow
+        * use bus address for ih ring by psp bl */
+       use_bus_addr =
+               (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) ? false : true;
+       r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 256 * 1024, use_bus_addr);
+       if (r)
+               return r;
+
+       adev->irq.ih.use_doorbell = true;
+       adev->irq.ih.doorbell_index = adev->doorbell_index.ih << 1;
+
+       r = amdgpu_irq_init(adev);
+
+       return r;
+}
+
+static int navi10_ih_sw_fini(void *handle)
+{
+       struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+       amdgpu_irq_fini(adev);
+       amdgpu_ih_ring_fini(adev, &adev->irq.ih);
+
+       return 0;
+}
+
+static int navi10_ih_hw_init(void *handle)
+{
+       int r;
+       struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+       r = navi10_ih_irq_init(adev);
+       if (r)
+               return r;
+
+       return 0;
+}
+
+static int navi10_ih_hw_fini(void *handle)
+{
+       struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+       navi10_ih_irq_disable(adev);
+
+       return 0;
+}
+
+static int navi10_ih_suspend(void *handle)
+{
+       struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+       return navi10_ih_hw_fini(adev);
+}
+
+static int navi10_ih_resume(void *handle)
+{
+       struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+       return navi10_ih_hw_init(adev);
+}
+
+static bool navi10_ih_is_idle(void *handle)
+{
+       /* todo */
+       return true;
+}
+
+static int navi10_ih_wait_for_idle(void *handle)
+{
+       /* todo */
+       return -ETIMEDOUT;
+}
+
+static int navi10_ih_soft_reset(void *handle)
+{
+       /* todo */
+       return 0;
+}
+
+static void navi10_ih_update_clockgating_state(struct amdgpu_device *adev,
+                                              bool enable)
+{
+       uint32_t data, def, field_val;
+
+       if (adev->cg_flags & AMD_CG_SUPPORT_IH_CG) {
+               def = data = RREG32_SOC15(OSSSYS, 0, mmIH_CLK_CTRL);
+               field_val = enable ? 0 : 1;
+               data = REG_SET_FIELD(data, IH_CLK_CTRL,
+                                    DBUS_MUX_CLK_SOFT_OVERRIDE, field_val);
+               data = REG_SET_FIELD(data, IH_CLK_CTRL,
+                                    OSSSYS_SHARE_CLK_SOFT_OVERRIDE, field_val);
+               data = REG_SET_FIELD(data, IH_CLK_CTRL,
+                                    LIMIT_SMN_CLK_SOFT_OVERRIDE, field_val);
+               data = REG_SET_FIELD(data, IH_CLK_CTRL,
+                                    DYN_CLK_SOFT_OVERRIDE, field_val);
+               data = REG_SET_FIELD(data, IH_CLK_CTRL,
+                                    REG_CLK_SOFT_OVERRIDE, field_val);
+               if (def != data)
+                       WREG32_SOC15(OSSSYS, 0, mmIH_CLK_CTRL, data);
+       }
+
+       return;
+}
+
+static int navi10_ih_set_clockgating_state(void *handle,
+                                          enum amd_clockgating_state state)
+{
+       struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+       navi10_ih_update_clockgating_state(adev,
+                               state == AMD_CG_STATE_GATE ? true : false);
+       return 0;
+}
+
+static int navi10_ih_set_powergating_state(void *handle,
+                                          enum amd_powergating_state state)
+{
+       return 0;
+}
+
+static void navi10_ih_get_clockgating_state(void *handle, u32 *flags)
+{
+       struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+       if (!RREG32_SOC15(OSSSYS, 0, mmIH_CLK_CTRL))
+               *flags |= AMD_CG_SUPPORT_IH_CG;
+
+       return;
+}
+
+static const struct amd_ip_funcs navi10_ih_ip_funcs = {
+       .name = "navi10_ih",
+       .early_init = navi10_ih_early_init,
+       .late_init = NULL,
+       .sw_init = navi10_ih_sw_init,
+       .sw_fini = navi10_ih_sw_fini,
+       .hw_init = navi10_ih_hw_init,
+       .hw_fini = navi10_ih_hw_fini,
+       .suspend = navi10_ih_suspend,
+       .resume = navi10_ih_resume,
+       .is_idle = navi10_ih_is_idle,
+       .wait_for_idle = navi10_ih_wait_for_idle,
+       .soft_reset = navi10_ih_soft_reset,
+       .set_clockgating_state = navi10_ih_set_clockgating_state,
+       .set_powergating_state = navi10_ih_set_powergating_state,
+       .get_clockgating_state = navi10_ih_get_clockgating_state,
+};
+
+static const struct amdgpu_ih_funcs navi10_ih_funcs = {
+       .get_wptr = navi10_ih_get_wptr,
+       .decode_iv = navi10_ih_decode_iv,
+       .set_rptr = navi10_ih_set_rptr
+};
+
+static void navi10_ih_set_interrupt_funcs(struct amdgpu_device *adev)
+{
+       if (adev->irq.ih_funcs == NULL)
+               adev->irq.ih_funcs = &navi10_ih_funcs;
+}
+
+const struct amdgpu_ip_block_version navi10_ih_ip_block =
+{
+       .type = AMD_IP_BLOCK_TYPE_IH,
+       .major = 5,
+       .minor = 0,
+       .rev = 0,
+       .funcs = &navi10_ih_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/navi10_ih.h b/drivers/gpu/drm/amd/amdgpu/navi10_ih.h
new file mode 100644 (file)
index 0000000..140fbda
--- /dev/null
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __NAVI10_IH_H__
+#define __NAVI10_IH_H__
+
+extern const struct amdgpu_ip_block_version navi10_ih_ip_block;
+
+#endif