gpu: nvgpu: add refcounting for MMU debug mode

GPC MMU debug mode should be set if at least one channel
in the TSG has requested it. Add refcounting for MMU debug
mode, to make sure debug mode is disabled only when no
channel in the TSG is using it.

Bug 2515097
Bug 2713590

Change-Id: Ic5530f93523a9ec2cd3bfebc97adf7b7000531e0
Signed-off-by: Thomas Fleury <tfleury@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2123017
(cherry picked from commit a1248d87fe6e20aab3e5f2e0764f9fe8d80d0552)
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvgpu/+/2208769
Reviewed-by: Kajetan Dutka <kdutka@nvidia.com>
Reviewed-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: Winnie Hsu <whsu@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: Kajetan Dutka <kdutka@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
GVS: Gerrit_Virtual_Submit

5 files changed, 64 insertions, 5 deletions

diff --git a/drivers/gpu/nvgpu/common/fifo/tsg.c b/drivers/gpu/nvgpu/common/fifo/tsg.c
index 6d6c322e..5883667f 100644
--- a/drivers/gpu/nvgpu/common/fifo/tsg.c
+++ b/drivers/gpu/nvgpu/common/fifo/tsg.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2014-2019, NVIDIA CORPORATION.  All rights reserved.
+ * Copyright (c) 2014-2020, NVIDIA CORPORATION.  All rights reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -445,3 +445,49 @@ void gk20a_tsg_update_sm_error_state_locked(struct tsg_gk20a *tsg,
         tsg_sm_error_states->hww_warp_esr_report_mask =
                         sm_error_state->hww_warp_esr_report_mask;
 }
+
+int nvgpu_tsg_set_mmu_debug_mode(struct tsg_gk20a *tsg,
+                struct channel_gk20a *ch, bool enable)
+{
+        struct gk20a *g;
+        int err = 0;
+        u32 tsg_refcnt;
+
+        if ((ch == NULL) || (tsg == NULL)) {
+                return -EINVAL;
+        }
+        g = ch->g;
+
+        if (g->ops.gr.set_mmu_debug_mode == NULL) {
+                return -ENOSYS;
+        }
+
+        if (enable) {
+                if (ch->mmu_debug_mode_enabled) {
+                        /* already enabled for this channel */
+                        return 0;
+                }
+                tsg_refcnt = tsg->mmu_debug_mode_refcnt + 1U;
+        } else {
+                if (!ch->mmu_debug_mode_enabled) {
+                        /* already disabled for this channel */
+                        return 0;
+                }
+                tsg_refcnt = tsg->mmu_debug_mode_refcnt - 1U;
+        }
+
+        /*
+         * enable GPC MMU debug mode if it was requested for at
+         * least one channel in the TSG
+         */
+        err = g->ops.gr.set_mmu_debug_mode(g, ch, tsg_refcnt > 0U);
+        if (err != 0) {
+                nvgpu_err(g, "set mmu debug mode failed, err=%d", err);
+                return err;
+        }
+
+        ch->mmu_debug_mode_enabled = enable;
+        tsg->mmu_debug_mode_refcnt = tsg_refcnt;
+
+        return err;
+}
diff --git a/drivers/gpu/nvgpu/gm20b/gr_gm20b.c b/drivers/gpu/nvgpu/gm20b/gr_gm20b.c
index d00181af..dacef784 100644
--- a/drivers/gpu/nvgpu/gm20b/gr_gm20b.c
+++ b/drivers/gpu/nvgpu/gm20b/gr_gm20b.c
@@ -1468,10 +1468,15 @@ int gm20b_gr_set_mmu_debug_mode(struct gk20a *g,
                         gr_gpcs_pri_mmu_debug_ctrl_debug_disabled_f(),
         };
         int err;
+        struct tsg_gk20a *tsg = tsg_gk20a_from_ch(ch);
+
+        if (tsg == NULL) {
+                return enable ? -EINVAL : 0;
+        }
 
         err = gr_gk20a_exec_ctx_ops(ch, &ctx_ops, 1, 1, 0, NULL);
         if (err != 0) {
-                nvgpu_err(g, "Failed to access register");
+                nvgpu_err(g, "update MMU debug mode failed");
         }
         return err;
 }
diff --git a/drivers/gpu/nvgpu/include/nvgpu/channel.h b/drivers/gpu/nvgpu/include/nvgpu/channel.h
index 7ca60fe3..e98af09b 100644
--- a/drivers/gpu/nvgpu/include/nvgpu/channel.h
+++ b/drivers/gpu/nvgpu/include/nvgpu/channel.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2018-2019, NVIDIA CORPORATION.  All rights reserved.
+ * Copyright (c) 2018-2020, NVIDIA CORPORATION.  All rights reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -328,6 +328,7 @@ struct channel_gk20a {
         bool has_os_fence_framework_support;
 
         bool is_privileged_channel;
+        bool mmu_debug_mode_enabled;
 };
 
 static inline struct channel_gk20a *
diff --git a/drivers/gpu/nvgpu/include/nvgpu/tsg.h b/drivers/gpu/nvgpu/include/nvgpu/tsg.h
index 212fb77c..b43a6b98 100644
--- a/drivers/gpu/nvgpu/include/nvgpu/tsg.h
+++ b/drivers/gpu/nvgpu/include/nvgpu/tsg.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2014-2018, NVIDIA CORPORATION.  All rights reserved.
+ * Copyright (c) 2014-2020, NVIDIA CORPORATION.  All rights reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -76,6 +76,9 @@ struct tsg_gk20a {
         bool tpc_num_initialized;
         bool in_use;
 
+        /* MMU debug mode enabled if mmu_debug_mode_refcnt > 0 */
+        u32  mmu_debug_mode_refcnt;
+
         struct nvgpu_tsg_sm_error_state *sm_error_states;
 
 #define NVGPU_SM_EXCEPTION_TYPE_MASK_NONE               (0x0U)
@@ -124,4 +127,7 @@ gk20a_event_id_data_from_event_id_node(struct nvgpu_list_node *node)
                 ((uintptr_t)node - offsetof(struct gk20a_event_id_data, event_id_node));
 };
 
+int nvgpu_tsg_set_mmu_debug_mode(struct tsg_gk20a *tsg,
+                struct channel_gk20a *ch, bool enable);
+
 #endif /* TSG_GK20A_H */
diff --git a/drivers/gpu/nvgpu/os/linux/ioctl_dbg.c b/drivers/gpu/nvgpu/os/linux/ioctl_dbg.c
index f7a65f2b..408bbc64 100644
--- a/drivers/gpu/nvgpu/os/linux/ioctl_dbg.c
+++ b/drivers/gpu/nvgpu/os/linux/ioctl_dbg.c
@@ -1107,10 +1107,11 @@ static int nvgpu_dbg_gpu_ioctl_set_mmu_debug_mode(
         ch = nvgpu_dbg_gpu_get_session_channel(dbg_s);
         if (!ch) {
                 nvgpu_err(g, "no bound channel for mmu debug mode");
+                err = -EINVAL;
                 goto clean_up;
         }
 
-        err = g->ops.gr.set_mmu_debug_mode(g, ch, enable);
+        err = nvgpu_tsg_set_mmu_debug_mode(tsg_gk20a_from_ch(ch), ch, enable);
         if (err) {
                 nvgpu_err(g, "set mmu debug mode failed, err=%d", err);
         }