gpu: nvgpu: process granularity for FECS traces

When processing FECS traces, a hash table is used
to retrieve the 'pid' of the process that created
the channel/TSG. Report process identifer (aka
tgid in kernel) instead of thread identifier (aka
pid) for FECS traces.

Bug 1736423

Change-Id: I54cb9d298b9fe3e1cccdd7145604cd01c5758c9d
Signed-off-by: Thomas Fleury <tfleury@nvidia.com>
Reviewed-on: http://git-master/r/1166501
(cherry picked from commit f7fd1f6d7ad0753b787ec20604a08a1f4882fe6f)
Reviewed-on: http://git-master/r/1168728
(cherry picked from commit 97a62e5b89352fce576f1bca71b38bf2242ff047)
Reviewed-on: http://git-master/r/1177823
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Richard Zhao <rizhao@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>

6 files changed, 10 insertions, 4 deletions

diff --git a/drivers/gpu/nvgpu/gk20a/channel_gk20a.c b/drivers/gpu/nvgpu/gk20a/channel_gk20a.c
index e631b836..d5457d10 100644
--- a/drivers/gpu/nvgpu/gk20a/channel_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/channel_gk20a.c
@@ -1171,6 +1171,7 @@ struct channel_gk20a *gk20a_open_new_channel(struct gk20a *g,
          * alive and used (i.e. get-able) yet */
 
         ch->pid = current->pid;
+        ch->tgid = current->tgid;  /* process granularity for FECS traces */
 
         /* By default, channel is regular (non-TSG) channel */
         ch->tsgid = NVGPU_INVALID_TSG_ID;
diff --git a/drivers/gpu/nvgpu/gk20a/channel_gk20a.h b/drivers/gpu/nvgpu/gk20a/channel_gk20a.h
index c5a1bd24..4b5fe1b3 100644
--- a/drivers/gpu/nvgpu/gk20a/channel_gk20a.h
+++ b/drivers/gpu/nvgpu/gk20a/channel_gk20a.h
@@ -117,6 +117,7 @@ struct channel_gk20a {
         bool vpr;
         bool cde;
         pid_t pid;
+        pid_t tgid;
         struct mutex ioctl_lock;
 
         int tsgid;
diff --git a/drivers/gpu/nvgpu/gk20a/ctxsw_trace_gk20a.c b/drivers/gpu/nvgpu/gk20a/ctxsw_trace_gk20a.c
index d43c06be..1f18d9cf 100644
--- a/drivers/gpu/nvgpu/gk20a/ctxsw_trace_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/ctxsw_trace_gk20a.c
@@ -657,7 +657,7 @@ void gk20a_ctxsw_trace_channel_reset(struct gk20a *g, struct channel_gk20a *ch)
                 .vmid = 0,
                 .tag = NVGPU_CTXSW_TAG_ENGINE_RESET,
                 .context_id = 0,
-                .pid = ch->pid,
+                .pid = ch->tgid,
         };
 
         if (!g->ctxsw_trace)
diff --git a/drivers/gpu/nvgpu/gk20a/fecs_trace_gk20a.c b/drivers/gpu/nvgpu/gk20a/fecs_trace_gk20a.c
index 15e645f2..abf1cc55 100644
--- a/drivers/gpu/nvgpu/gk20a/fecs_trace_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/fecs_trace_gk20a.c
@@ -631,10 +631,14 @@ static int gk20a_fecs_trace_bind_channel(struct gk20a *g,
                         GK20A_FECS_TRACE_NUM_RECORDS));
 
         gk20a_mem_end(g, mem);
+
+        /* pid (process identifier) in user space, corresponds to tgid (thread
+         * group id) in kernel space.
+         */
         if (gk20a_is_channel_marked_as_tsg(ch))
                 pid = tsg_gk20a_from_ch(ch)->tgid;
         else
-                pid = ch->pid;
+                pid = ch->tgid;
         gk20a_fecs_trace_hash_add(g, context_ptr, pid);
 
         return 0;
diff --git a/drivers/gpu/nvgpu/gk20a/fifo_gk20a.c b/drivers/gpu/nvgpu/gk20a/fifo_gk20a.c
index 8c82fc68..5133f86a 100644
--- a/drivers/gpu/nvgpu/gk20a/fifo_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/fifo_gk20a.c
@@ -2999,7 +2999,7 @@ static int gk20a_fifo_sched_debugfs_seq_show(
                 seq_printf(s, "%-8d %-8d %-8d %-9d %-8d %-10d %-8d %-8d\n",
                                 ch->hw_chid,
                                 ch->tsgid,
-                                ch->pid,
+                                ch->tgid,
                                 tsg ? tsg->timeslice_us : ch->timeslice_us,
                                 ch->timeout_ms_max,
                                 ch->interleave_level,
diff --git a/drivers/gpu/nvgpu/vgpu/fifo_vgpu.c b/drivers/gpu/nvgpu/vgpu/fifo_vgpu.c
index 29692573..8d3a5e9f 100644
--- a/drivers/gpu/nvgpu/vgpu/fifo_vgpu.c
+++ b/drivers/gpu/nvgpu/vgpu/fifo_vgpu.c
@@ -74,7 +74,7 @@ static int vgpu_channel_alloc_inst(struct gk20a *g, struct channel_gk20a *ch)
         msg.cmd = TEGRA_VGPU_CMD_CHANNEL_ALLOC_HWCTX;
         msg.handle = platform->virt_handle;
         p->id = ch->hw_chid;
-        p->pid = (u64)current->pid;
+        p->pid = (u64)current->tgid;
         err = vgpu_comm_sendrecv(&msg, sizeof(msg), sizeof(msg));
         if (err || msg.ret) {
                 gk20a_err(dev_from_gk20a(g), "fail");