misc: tegra-profiler: add task comm events

Add task COMM events:
- Add support for quadd_event_comm() callback.
- Send task COMM events to userspace.

Bug 2514095
Jira DTSP-2432

Change-Id: Id5b92f544497b69098e3b16bf82baf0f6abb77e1
Signed-off-by: Igor Nabirushkin <inabirushkin@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2027490
(cherry picked from commit a0c0d3bf9eeabbdad2a43b768235e1cc846b04ef)
Reviewed-on: https://git-master.nvidia.com/r/2093405
GVS: Gerrit_Virtual_Submit
Reviewed-by: Roman Rybalko <rrybalko@nvidia.com>
Reviewed-by: Bibek Basu <bbasu@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>

8 files changed, 278 insertions, 148 deletions

diff --git a/drivers/misc/tegra-profiler/hrt.c b/drivers/misc/tegra-profiler/hrt.c
index 35c4f01bc..642cfffdb 100644
--- a/drivers/misc/tegra-profiler/hrt.c
+++ b/drivers/misc/tegra-profiler/hrt.c
@@ -1,7 +1,7 @@
 /*
  * drivers/misc/tegra-profiler/hrt.c
  *
- * Copyright (c) 2015-2018, NVIDIA CORPORATION.  All rights reserved.
+ * Copyright (c) 2015-2019, NVIDIA CORPORATION.  All rights reserved.
  *
  * This program is free software; you can redistribute it and/or modify it
  * under the terms and conditions of the GNU General Public License,
@@ -339,6 +339,36 @@ put_sched_sample(struct task_struct *task, int is_sched_in)
         quadd_put_sample_this_cpu(&record, vec, vec_idx);
 }
 
+static void put_comm_sample(struct task_struct *task, bool exec)
+{
+        char name[TASK_COMM_LEN];
+        struct quadd_iovec vec;
+        struct quadd_record_data record;
+        struct quadd_comm_data *s = &record.comm;
+
+        s->time = quadd_get_time();
+        s->flags = 0;
+
+        s->pid = (u32)task_pid_nr(task);
+        s->tgid = (u32)task_tgid_nr(task);
+
+        memset(name, 0, sizeof(name));
+        get_task_comm(name, task);
+        name[sizeof(name) - 1] = '\0';
+
+        record.record_type = QUADD_RECORD_TYPE_COMM;
+
+        if (exec)
+                s->flags |= QUADD_COMM_FLAG_EXEC;
+
+        vec.base = name;
+        vec.len = ALIGN(strlen(name) + 1, sizeof(u64));
+
+        s->length = (u16)vec.len;
+
+        quadd_put_sample(&record, &vec, 1);
+}
+
 static int get_sample_data(struct quadd_sample_data *s,
                            struct pt_regs *regs,
                            struct task_struct *task)
@@ -629,48 +659,69 @@ static void init_hrtimer(struct quadd_cpu_context *cpu_ctx)
         cpu_ctx->hrtimer.function = hrtimer_handler;
 }
 
-static inline int
-is_profile_process(struct task_struct *task, int is_trace)
+static inline bool
+__is_profile_process(struct task_struct *task, bool is_trace, bool is_sample)
 {
         pid_t root_pid = hrt.root_pid;
         struct quadd_ctx *ctx = hrt.quadd_ctx;
 
         if (root_pid > 0 && task_tgid_nr(task) == root_pid)
-                return 1;
+                return true;
 
         if ((is_trace && quadd_mode_is_trace_tree(ctx)) ||
-            (!is_trace && quadd_mode_is_sample_tree(ctx)))
+            (is_sample && quadd_mode_is_sample_tree(ctx)))
                 return pid_list_search(task_tgid_nr(task));
 
-        return 0;
+        return false;
 }
 
-static inline int
+static inline bool
 validate_task(struct task_struct *task)
 {
         return task && !is_idle_task(task);
 }
 
-static inline int
+static inline bool
 is_sample_process(struct task_struct *task)
 {
         struct quadd_ctx *ctx = hrt.quadd_ctx;
 
         if (!validate_task(task) || !quadd_mode_is_sampling(ctx))
-                return 0;
+                return false;
 
-        return (quadd_mode_is_sample_all(ctx) || is_profile_process(task, 0));
+        if (quadd_mode_is_sample_all(ctx))
+                return true;
+
+        return __is_profile_process(task, false, true);
 }
 
-static inline int
+static inline bool
 is_trace_process(struct task_struct *task)
 {
         struct quadd_ctx *ctx = hrt.quadd_ctx;
 
         if (!validate_task(task) || !quadd_mode_is_tracing(ctx))
-                return 0;
+                return false;
+
+        if (quadd_mode_is_trace_all(ctx))
+                return true;
+
+        return __is_profile_process(task, true, false);
+}
+
+static inline bool
+is_profile_process(struct task_struct *task)
+{
+        struct quadd_ctx *ctx = hrt.quadd_ctx;
+
+        if (!validate_task(task) ||
+            (!quadd_mode_is_tracing(ctx) && !quadd_mode_is_sampling(ctx)))
+                return false;
 
-        return (quadd_mode_is_trace_all(ctx) || is_profile_process(task, 1));
+        if (quadd_mode_is_process_all(ctx))
+                return true;
+
+        return __is_profile_process(task, true, true);
 }
 
 static void
@@ -790,23 +841,23 @@ void __quadd_event_mmap(struct vm_area_struct *vma)
         quadd_process_mmap(vma, current);
 }
 
-int quadd_is_inherited(struct task_struct *task)
+bool quadd_is_inherited(struct task_struct *task)
 {
         struct task_struct *p;
 
         if (unlikely(hrt.root_pid == 0))
-                return 0;
+                return false;
 
         for (p = task; p != &init_task;) {
                 if (task_pid_nr(p) == hrt.root_pid)
-                        return 1;
+                        return true;
 
                 rcu_read_lock();
                 p = rcu_dereference(p->real_parent);
                 rcu_read_unlock();
         }
 
-        return 0;
+        return false;
 }
 
 void __quadd_event_fork(struct task_struct *task)
@@ -839,6 +890,17 @@ void __quadd_event_exit(struct task_struct *task)
         read_unlock(&tasklist_lock);
 }
 
+void __quadd_event_comm(struct task_struct *task, bool exec)
+{
+        if (likely(!atomic_read(&hrt.active)))
+                return;
+
+        if (!is_profile_process(task))
+                return;
+
+        put_comm_sample(task, exec);
+}
+
 static void reset_cpu_ctx(void)
 {
         int cpu_id;
@@ -857,6 +919,51 @@ static void reset_cpu_ctx(void)
         }
 }
 
+static void get_initial_samples(struct quadd_ctx *ctx)
+{
+        struct task_struct *p, *t;
+
+        if (quadd_mode_is_sampling(ctx))
+                quadd_get_mmaps(ctx);
+
+        if (quadd_mode_is_process_all(ctx)) {
+                bool is_tree = quadd_mode_is_process_tree(ctx);
+
+                read_lock(&tasklist_lock);
+                for_each_process(p) {
+                        for_each_thread(p, t)
+                                put_comm_sample(t, false);
+
+                        if (is_tree && quadd_is_inherited(p))
+                                pid_list_add(task_pid_nr(p));
+                }
+                read_unlock(&tasklist_lock);
+        } else if (quadd_mode_is_process_tree(ctx)) {
+                read_lock(&tasklist_lock);
+                for_each_process(p) {
+                        if (quadd_is_inherited(p)) {
+                                pid_list_add(task_pid_nr(p));
+                                for_each_thread(p, t)
+                                        put_comm_sample(t, false);
+                        }
+                }
+                read_unlock(&tasklist_lock);
+        } else {
+                pid_t root_pid = hrt.root_pid;
+
+                if (root_pid > 0) {
+                        read_lock(&tasklist_lock);
+                        p = get_pid_task(find_vpid(root_pid), PIDTYPE_PID);
+                        if (p) {
+                                for_each_thread(p, t)
+                                        put_comm_sample(t, false);
+                                put_task_struct(p);
+                        }
+                        read_unlock(&tasklist_lock);
+                }
+        }
+}
+
 int quadd_hrt_start(void)
 {
         int cpuid;
@@ -920,25 +1027,13 @@ int quadd_hrt_start(void)
          */
         smp_wmb();
 
-        if (quadd_mode_is_sampling(ctx)) {
-                if (extra & QUADD_PARAM_EXTRA_GET_MMAP)
-                        quadd_get_mmaps(ctx);
+        get_initial_samples(ctx);
 
+        if (quadd_mode_is_sampling(ctx)) {
                 if (ctx->pl310)
                         ctx->pl310->start();
         }
 
-        if (quadd_mode_is_process_tree(ctx)) {
-                struct task_struct *p;
-
-                read_lock(&tasklist_lock);
-                for_each_process(p) {
-                        if (quadd_is_inherited(p))
-                                pid_list_add(task_pid_nr(p));
-                }
-                read_unlock(&tasklist_lock);
-        }
-
         quadd_ma_start(&hrt);
 
         /* Enable the sampling only after quadd_get_mmaps() */
diff --git a/drivers/misc/tegra-profiler/hrt.h b/drivers/misc/tegra-profiler/hrt.h
index e545b73e2..267426b71 100644
--- a/drivers/misc/tegra-profiler/hrt.h
+++ b/drivers/misc/tegra-profiler/hrt.h
@@ -1,7 +1,7 @@
 /*
  * drivers/misc/tegra-profiler/hrt.h
  *
- * Copyright (c) 2015-2018, NVIDIA CORPORATION.  All rights reserved.
+ * Copyright (c) 2015-2019, NVIDIA CORPORATION.  All rights reserved.
  *
  * This program is free software; you can redistribute it and/or modify it
  * under the terms and conditions of the GNU General Public License,
@@ -19,6 +19,7 @@
 
 #ifdef __KERNEL__
 
+#include <linux/types.h>
 #include <linux/hrtimer.h>
 #include <linux/limits.h>
 
@@ -113,7 +114,7 @@ quadd_put_sample(struct quadd_record_data *data,
 
 void quadd_hrt_get_state(struct quadd_module_state *state);
 u64 quadd_get_time(void);
-int quadd_is_inherited(struct task_struct *task);
+bool quadd_is_inherited(struct task_struct *task);
 
 #endif  /* __KERNEL__ */
 
diff --git a/drivers/misc/tegra-profiler/main.c b/drivers/misc/tegra-profiler/main.c
index 39ede51d8..23e43efd2 100644
--- a/drivers/misc/tegra-profiler/main.c
+++ b/drivers/misc/tegra-profiler/main.c
@@ -1,7 +1,7 @@
 /*
  * drivers/misc/tegra-profiler/main.c
  *
- * Copyright (c) 2013-2018, NVIDIA CORPORATION.  All rights reserved.
+ * Copyright (c) 2013-2019, NVIDIA CORPORATION.  All rights reserved.
  *
  * This program is free software; you can redistribute it and/or modify it
  * under the terms and conditions of the GNU General Public License,
@@ -267,6 +267,11 @@ set_parameters(struct quadd_parameters *p)
         ctx.mode_is_trace_tree =
                 extra & QUADD_PARAM_EXTRA_TRACE_TREE ? 1 : 0;
 
+        if (ctx.mode_is_sample_all)
+                ctx.mode_is_sample_tree = 0;
+        if (ctx.mode_is_trace_all)
+                ctx.mode_is_trace_tree = 0;
+
         pr_info("flags: s/t/sa/ta/st/tt: %u/%u/%u/%u/%u/%u\n",
                 ctx.mode_is_sampling,
                 ctx.mode_is_tracing,
@@ -295,10 +300,12 @@ set_parameters(struct quadd_parameters *p)
                 }
 
                 /* Currently only first process */
-                if (p->nr_pids != 1)
+                if (p->nr_pids != 1 || p->pids[0] == 0)
                         return -EINVAL;
 
+                rcu_read_lock();
                 task = get_pid_task(find_vpid(p->pids[0]), PIDTYPE_PID);
+                rcu_read_unlock();
                 if (!task) {
                         pr_err("error: process not found: %u\n", p->pids[0]);
                         return -ESRCH;
diff --git a/drivers/misc/tegra-profiler/mmap.c b/drivers/misc/tegra-profiler/mmap.c
index 228ddd75b..3adaa5bef 100644
--- a/drivers/misc/tegra-profiler/mmap.c
+++ b/drivers/misc/tegra-profiler/mmap.c
@@ -214,7 +214,7 @@ __get_process_vmas(struct task_struct *task,
                 process_mmap(vma, task, buf, buf_size);
 }
 
-static void get_all_processes(int is_root_pid)
+static void get_all_processes(bool is_root_pid)
 {
         char *buf;
         struct task_struct *p;
@@ -255,7 +255,6 @@ __continue:
 
 void quadd_get_mmaps(struct quadd_ctx *ctx)
 {
-        struct pid *pid;
         struct task_struct *task;
         struct quadd_parameters *param = &ctx->param;
 
@@ -263,22 +262,21 @@ void quadd_get_mmaps(struct quadd_ctx *ctx)
                 return;
 
         if (quadd_mode_is_sample_all(ctx)) {
-                get_all_processes(0);
+                get_all_processes(false);
                 return;
         }
 
-        pid = find_vpid(param->pids[0]);
-
-        task = get_pid_task(pid, PIDTYPE_PID);
-        if (!task)
-                return;
-
-        if (quadd_mode_is_sample_tree(ctx))
-                get_all_processes(1);
-        else
-                get_process_vmas(task);
+        rcu_read_lock();
+        task = get_pid_task(find_vpid(param->pids[0]), PIDTYPE_PID);
+        rcu_read_unlock();
+        if (task) {
+                if (quadd_mode_is_sample_tree(ctx))
+                        get_all_processes(true);
+                else
+                        get_process_vmas(task);
 
-        put_task_struct(task);
+                put_task_struct(task);
+        }
 }
 
 void quadd_get_task_mmaps(struct quadd_ctx *ctx, struct task_struct *task)
diff --git a/drivers/misc/tegra-profiler/quadd.h b/drivers/misc/tegra-profiler/quadd.h
index ba166e085..eb91a0737 100644
--- a/drivers/misc/tegra-profiler/quadd.h
+++ b/drivers/misc/tegra-profiler/quadd.h
@@ -1,7 +1,7 @@
 /*
  * drivers/misc/tegra-profiler/quadd.h
  *
- * Copyright (c) 2014-2018, NVIDIA CORPORATION.  All rights reserved.
+ * Copyright (c) 2014-2019, NVIDIA CORPORATION.  All rights reserved.
  *
  * This program is free software; you can redistribute it and/or modify it
  * under the terms and conditions of the GNU General Public License,
@@ -17,6 +17,7 @@
 #ifndef __QUADD_H
 #define __QUADD_H
 
+#include <linux/types.h>
 #include <linux/list.h>
 #include <linux/spinlock.h>
 
@@ -96,39 +97,44 @@ struct quadd_ctx {
         raw_spinlock_t mmaps_lock;
 };
 
-static inline int quadd_mode_is_sampling(struct quadd_ctx *ctx)
+static inline bool quadd_mode_is_sampling(struct quadd_ctx *ctx)
 {
-        return ctx->mode_is_sampling;
+        return ctx->mode_is_sampling != 0;
 }
 
-static inline int quadd_mode_is_tracing(struct quadd_ctx *ctx)
+static inline bool quadd_mode_is_tracing(struct quadd_ctx *ctx)
 {
-        return ctx->mode_is_tracing;
+        return ctx->mode_is_tracing != 0;
 }
 
-static inline int quadd_mode_is_sample_all(struct quadd_ctx *ctx)
+static inline bool quadd_mode_is_sample_all(struct quadd_ctx *ctx)
 {
-        return ctx->mode_is_sample_all;
+        return ctx->mode_is_sample_all != 0;
 }
 
-static inline int quadd_mode_is_trace_all(struct quadd_ctx *ctx)
+static inline bool quadd_mode_is_trace_all(struct quadd_ctx *ctx)
 {
-        return ctx->mode_is_trace_all;
+        return ctx->mode_is_trace_all != 0;
 }
 
-static inline int quadd_mode_is_sample_tree(struct quadd_ctx *ctx)
+static inline bool quadd_mode_is_sample_tree(struct quadd_ctx *ctx)
 {
-        return ctx->mode_is_sample_tree;
+        return ctx->mode_is_sample_tree != 0;
 }
 
-static inline int quadd_mode_is_trace_tree(struct quadd_ctx *ctx)
+static inline bool quadd_mode_is_trace_tree(struct quadd_ctx *ctx)
 {
-        return ctx->mode_is_trace_tree;
+        return ctx->mode_is_trace_tree != 0;
 }
 
-static inline int quadd_mode_is_process_tree(struct quadd_ctx *ctx)
+static inline bool quadd_mode_is_process_tree(struct quadd_ctx *ctx)
 {
-        return (ctx->mode_is_sample_tree || ctx->mode_is_trace_tree);
+        return (ctx->mode_is_sample_tree != 0 || ctx->mode_is_trace_tree != 0);
+}
+
+static inline bool quadd_mode_is_process_all(struct quadd_ctx *ctx)
+{
+        return (ctx->mode_is_sample_all != 0 || ctx->mode_is_trace_all != 0);
 }
 
 void quadd_get_state(struct quadd_module_state *state);
diff --git a/drivers/misc/tegra-profiler/version.h b/drivers/misc/tegra-profiler/version.h
index c0e648d87..b0b75a7ee 100644
--- a/drivers/misc/tegra-profiler/version.h
+++ b/drivers/misc/tegra-profiler/version.h
@@ -17,7 +17,7 @@
 #ifndef __QUADD_VERSION_H
 #define __QUADD_VERSION_H
 
-#define QUADD_MODULE_VERSION            "1.132"
+#define QUADD_MODULE_VERSION            "1.133"
 #define QUADD_MODULE_BRANCH             "Dev"
 
 #endif  /* __QUADD_VERSION_H */
diff --git a/include/linux/tegra_profiler.h b/include/linux/tegra_profiler.h
index c55f25b57..edca37b25 100644
--- a/include/linux/tegra_profiler.h
+++ b/include/linux/tegra_profiler.h
@@ -31,6 +31,7 @@ extern void __quadd_task_sched_out(struct task_struct *prev,
 extern void __quadd_event_mmap(struct vm_area_struct *vma);
 extern void __quadd_event_fork(struct task_struct *task);
 extern void __quadd_event_exit(struct task_struct *task);
+extern void __quadd_event_comm(struct task_struct *task, bool exec);
 
 static inline void quadd_task_sched_in(struct task_struct *prev,
                                        struct task_struct *task)
@@ -59,6 +60,11 @@ static inline void quadd_event_exit(struct task_struct *task)
         __quadd_event_exit(task);
 }
 
+static inline void quadd_event_comm(struct task_struct *task, bool exec)
+{
+        __quadd_event_comm(task, exec);
+}
+
 #else   /* CONFIG_TEGRA_PROFILER */
 
 static inline void quadd_task_sched_in(struct task_struct *prev,
@@ -83,6 +89,10 @@ static inline void quadd_event_exit(struct task_struct *task)
 {
 }
 
+static inline void quadd_event_comm(struct task_struct *task, bool exec)
+{
+}
+
 #endif  /* CONFIG_TEGRA_PROFILER */
 
 #endif  /* __TEGRA_PROFILER_H */
diff --git a/include/uapi/linux/tegra_profiler.h b/include/uapi/linux/tegra_profiler.h
index 0e0c36f15..b05e46724 100644
--- a/include/uapi/linux/tegra_profiler.h
+++ b/include/uapi/linux/tegra_profiler.h
@@ -20,71 +20,72 @@
 #include <linux/ioctl.h>
 #include <linux/types.h>
 
-#define QUADD_SAMPLES_VERSION   46
-#define QUADD_IO_VERSION        26
-
-#define QUADD_IO_VERSION_DYNAMIC_RB             5
-#define QUADD_IO_VERSION_RB_MAX_FILL_COUNT      6
-#define QUADD_IO_VERSION_MOD_STATE_STATUS_FIELD 7
-#define QUADD_IO_VERSION_BT_KERNEL_CTX          8
-#define QUADD_IO_VERSION_GET_MMAP               9
-#define QUADD_IO_VERSION_BT_UNWIND_TABLES       10
-#define QUADD_IO_VERSION_UNWIND_MIXED           11
-#define QUADD_IO_VERSION_EXTABLES_MMAP          12
-#define QUADD_IO_VERSION_ARCH_TIMER_OPT         13
-#define QUADD_IO_VERSION_DATA_MMAP              14
-#define QUADD_IO_VERSION_BT_LOWER_BOUND         15
-#define QUADD_IO_VERSION_STACK_OFFSET           16
-#define QUADD_IO_VERSION_SECTIONS_INFO          17
-#define QUADD_IO_VERSION_UNW_METHODS_OPT        18
-#define QUADD_IO_VERSION_PER_CPU_SETUP          19
-#define QUADD_IO_VERSION_TRACE_ALL_TASKS        20
-#define QUADD_IO_VERSION_CB_POWER_OF_2          21
-#define QUADD_IO_VERSION_RAW_EVENTS             22
-#define QUADD_IO_VERSION_SAMPLING_MODE          23
-#define QUADD_IO_VERSION_FORCE_ARCH_TIMER       24
-#define QUADD_IO_VERSION_SAMPLE_ALL_TASKS       25
-#define QUADD_IO_VERSION_EXTABLES_PID           26
-
-#define QUADD_SAMPLE_VERSION_THUMB_MODE_FLAG            17
-#define QUADD_SAMPLE_VERSION_GROUP_SAMPLES              18
-#define QUADD_SAMPLE_VERSION_THREAD_STATE_FLD           19
-#define QUADD_SAMPLE_VERSION_BT_UNWIND_TABLES           22
-#define QUADD_SAMPLE_VERSION_SUPPORT_IP64               23
-#define QUADD_SAMPLE_VERSION_SPECIAL_MMAP               24
-#define QUADD_SAMPLE_VERSION_UNWIND_MIXED               25
-#define QUADD_SAMPLE_VERSION_UNW_ENTRY_TYPE             26
-#define QUADD_SAMPLE_VERSION_USE_ARCH_TIMER             27
-#define QUADD_SAMPLE_VERSION_SCHED_SAMPLES              28
-#define QUADD_SAMPLE_VERSION_HDR_UNW_METHOD             29
-#define QUADD_SAMPLE_VERSION_HDR_ARCH_TIMER             30
-#define QUADD_SAMPLE_VERSION_STACK_OFFSET               31
-#define QUADD_SAMPLE_VERSION_SCHED_TASK_STATE           32
-#define QUADD_SAMPLE_VERSION_URCS                       33
-#define QUADD_SAMPLE_VERSION_HOTPLUG                    34
-#define QUADD_SAMPLE_VERSION_PER_CPU_SETUP              35
-#define QUADD_SAMPLE_VERSION_REPORT_TGID                36
-#define QUADD_SAMPLE_VERSION_MMAP_TS                    37
-#define QUADD_SAMPLE_VERSION_RAW_EVENTS                 38
-#define QUADD_SAMPLE_VERSION_OVERHEAD_INFO              39
-#define QUADD_SAMPLE_VERSION_REPORT_VPID                40
-#define QUADD_SAMPLE_VERSION_SCHED_REPORT_VPID          41
-#define QUADD_SAMPLE_VERSION_SAMPLING_MODE              42
-#define QUADD_SAMPLE_VERSION_SAMPLE_ALL_TASKS           43
-#define QUADD_SAMPLE_VERSION_KTHREAD_TSK_FLAG           44
-#define QUADD_SAMPLE_VERSION_MMAP_CPUID                 45
-#define QUADD_SAMPLE_VERSION_PCLK_SEND_CHANGES          46
-
-#define QUADD_MMAP_HEADER_VERSION       1
-
-#define QUADD_MAX_COUNTERS      32
-#define QUADD_MAX_PROCESS       64
-
-#define QUADD_DEVICE_NAME       "quadd"
-#define QUADD_AUTH_DEVICE_NAME  "quadd_auth"
-
-#define QUADD_MOD_DEVICE_NAME           "quadd_mod"
-#define QUADD_MOD_AUTH_DEVICE_NAME      "quadd_mod_auth"
+#define QUADD_SAMPLES_VERSION   47
+#define QUADD_IO_VERSION        26
+
+#define QUADD_IO_VERSION_DYNAMIC_RB             5
+#define QUADD_IO_VERSION_RB_MAX_FILL_COUNT      6
+#define QUADD_IO_VERSION_MOD_STATE_STATUS_FIELD 7
+#define QUADD_IO_VERSION_BT_KERNEL_CTX          8
+#define QUADD_IO_VERSION_GET_MMAP               9
+#define QUADD_IO_VERSION_BT_UNWIND_TABLES       10
+#define QUADD_IO_VERSION_UNWIND_MIXED           11
+#define QUADD_IO_VERSION_EXTABLES_MMAP          12
+#define QUADD_IO_VERSION_ARCH_TIMER_OPT         13
+#define QUADD_IO_VERSION_DATA_MMAP              14
+#define QUADD_IO_VERSION_BT_LOWER_BOUND         15
+#define QUADD_IO_VERSION_STACK_OFFSET           16
+#define QUADD_IO_VERSION_SECTIONS_INFO          17
+#define QUADD_IO_VERSION_UNW_METHODS_OPT        18
+#define QUADD_IO_VERSION_PER_CPU_SETUP          19
+#define QUADD_IO_VERSION_TRACE_ALL_TASKS        20
+#define QUADD_IO_VERSION_CB_POWER_OF_2          21
+#define QUADD_IO_VERSION_RAW_EVENTS             22
+#define QUADD_IO_VERSION_SAMPLING_MODE          23
+#define QUADD_IO_VERSION_FORCE_ARCH_TIMER       24
+#define QUADD_IO_VERSION_SAMPLE_ALL_TASKS       25
+#define QUADD_IO_VERSION_EXTABLES_PID           26
+
+#define QUADD_SAMPLE_VERSION_THUMB_MODE_FLAG    17
+#define QUADD_SAMPLE_VERSION_GROUP_SAMPLES      18
+#define QUADD_SAMPLE_VERSION_THREAD_STATE_FLD   19
+#define QUADD_SAMPLE_VERSION_BT_UNWIND_TABLES   22
+#define QUADD_SAMPLE_VERSION_SUPPORT_IP64       23
+#define QUADD_SAMPLE_VERSION_SPECIAL_MMAP       24
+#define QUADD_SAMPLE_VERSION_UNWIND_MIXED       25
+#define QUADD_SAMPLE_VERSION_UNW_ENTRY_TYPE     26
+#define QUADD_SAMPLE_VERSION_USE_ARCH_TIMER     27
+#define QUADD_SAMPLE_VERSION_SCHED_SAMPLES      28
+#define QUADD_SAMPLE_VERSION_HDR_UNW_METHOD     29
+#define QUADD_SAMPLE_VERSION_HDR_ARCH_TIMER     30
+#define QUADD_SAMPLE_VERSION_STACK_OFFSET       31
+#define QUADD_SAMPLE_VERSION_SCHED_TASK_STATE   32
+#define QUADD_SAMPLE_VERSION_URCS               33
+#define QUADD_SAMPLE_VERSION_HOTPLUG            34
+#define QUADD_SAMPLE_VERSION_PER_CPU_SETUP      35
+#define QUADD_SAMPLE_VERSION_REPORT_TGID        36
+#define QUADD_SAMPLE_VERSION_MMAP_TS            37
+#define QUADD_SAMPLE_VERSION_RAW_EVENTS         38
+#define QUADD_SAMPLE_VERSION_OVERHEAD_INFO      39
+#define QUADD_SAMPLE_VERSION_REPORT_VPID        40
+#define QUADD_SAMPLE_VERSION_SCHED_REPORT_VPID  41
+#define QUADD_SAMPLE_VERSION_SAMPLING_MODE      42
+#define QUADD_SAMPLE_VERSION_SAMPLE_ALL_TASKS   43
+#define QUADD_SAMPLE_VERSION_KTHREAD_TSK_FLAG   44
+#define QUADD_SAMPLE_VERSION_MMAP_CPUID         45
+#define QUADD_SAMPLE_VERSION_PCLK_SEND_CHANGES  46
+#define QUADD_SAMPLE_VERSION_COMM_SAMPLES       47
+
+#define QUADD_MMAP_HEADER_VERSION       1
+
+#define QUADD_MAX_COUNTERS      32
+#define QUADD_MAX_PROCESS       64
+
+#define QUADD_DEVICE_NAME               "quadd"
+#define QUADD_AUTH_DEVICE_NAME          "quadd_auth"
+
+#define QUADD_MOD_DEVICE_NAME           "quadd_mod"
+#define QUADD_MOD_AUTH_DEVICE_NAME      "quadd_mod_auth"
 
 #define QUADD_IOCTL     100
 
@@ -345,6 +346,17 @@ enum {
         QM_DEBUG_SAMPLE_TYPE_SOURCE_STOP,
 };
 
+#define QUADD_COMM_FLAG_EXEC    (1U << 0)
+
+struct quadd_comm_data {
+        __u32 pid;
+        __u32 tgid;
+        __u64 time;
+
+        __u16 length;
+        __u32 flags;
+};
+
 struct quadd_debug_data {
         __u8 type;
 
@@ -414,6 +426,7 @@ struct quadd_record_data {
                 struct quadd_power_rate_data    power_rate;
                 struct quadd_hotplug_data       hotplug;
                 struct quadd_sched_data         sched;
+                struct quadd_comm_data          comm;
                 struct quadd_additional_sample  additional_sample;
         };
 } __aligned(4);
@@ -457,34 +470,34 @@ struct quadd_event {
 };
 
 struct quadd_parameters {
-        __u32 freq;
-        __u32 ma_freq;
-        __u32 power_rate_freq;
+        __u32 freq;
+        __u32 ma_freq;
+        __u32 power_rate_freq;
 
-        __u64   backtrace:1,
-                use_freq:1,
-                system_wide:1,
-                debug_samples:1,
-                trace_all_tasks:1;
+        __u64   backtrace:1,
+                use_freq:1,
+                system_wide:1,
+                debug_samples:1,
+                trace_all_tasks:1;
 
-        __u32 pids[QUADD_MAX_PROCESS];
-        __u32 nr_pids;
+        __u32 pids[QUADD_MAX_PROCESS];
+        __u32 nr_pids;
 
-        __u8 package_name[QUADD_MAX_PACKAGE_NAME];
+        __u8 package_name[QUADD_MAX_PACKAGE_NAME];
 
-        struct quadd_event events[QUADD_MAX_COUNTERS];
-        __u32 nr_events;
+        struct quadd_event events[QUADD_MAX_COUNTERS];
+        __u32 nr_events;
 
-        __u32 reserved[16];     /* reserved fields for future extensions */
+        __u32 reserved[16];     /* reserved fields for future extensions */
 };
 
 struct quadd_pmu_setup_for_cpu {
-        __u32 cpuid;
+        __u32 cpuid;
 
-        struct quadd_event events[QUADD_MAX_COUNTERS];
-        __u32 nr_events;
+        struct quadd_event events[QUADD_MAX_COUNTERS];
+        __u32 nr_events;
 
-        __u32 reserved[16];
+        __u32 reserved[16];
 };
 
 struct quadd_events_cap {