From 5f94e09f9c3f4887d7183c8c123213e94e6ede31 Mon Sep 17 00:00:00 2001
From: Christian Hergert <chergert@redhat.com>
Date: Fri, 1 Sep 2023 14:09:17 -0700
Subject: [PATCH] libsysprof: avoid duplicate process records

We want the first process record, but subsequent records should
be ignored, except that we want to take the appropriate comm[]
action from them.
---
 .../sysprof-document-process-private.h         |  1 +
 src/libsysprof/sysprof-document-process.c      | 18 +++++++++++++++++-
 src/libsysprof/sysprof-document.c              | 17 ++++++++++++++++-
 3 files changed, 34 insertions(+), 2 deletions(-)

diff --git a/src/libsysprof/sysprof-document-process-private.h b/src/libsysprof/sysprof-document-process-private.h
index e345e15a..78eb5f2d 100644
--- a/src/libsysprof/sysprof-document-process-private.h
+++ b/src/libsysprof/sysprof-document-process-private.h
@@ -28,5 +28,6 @@ G_BEGIN_DECLS
 SysprofProcessInfo *_sysprof_document_process_get_info (SysprofDocumentProcess *self);
 void                _sysprof_document_process_set_info (SysprofDocumentProcess *self,
                                                         SysprofProcessInfo     *process_info);
+const char         *_sysprof_document_process_get_comm (SysprofDocumentProcess *self);
 
 G_END_DECLS
diff --git a/src/libsysprof/sysprof-document-process.c b/src/libsysprof/sysprof-document-process.c
index ca348983..0e0108b8 100644
--- a/src/libsysprof/sysprof-document-process.c
+++ b/src/libsysprof/sysprof-document-process.c
@@ -25,6 +25,7 @@
 #include "sysprof-document-frame-private.h"
 #include "sysprof-document-process-private.h"
 #include "sysprof-mount.h"
+#include "sysprof-symbol-private.h"
 #include "sysprof-thread-info.h"
 
 struct _SysprofDocumentProcess
@@ -159,7 +160,7 @@ sysprof_document_process_init (SysprofDocumentProcess *self)
 }
 
 const char *
-sysprof_document_process_get_command_line (SysprofDocumentProcess *self)
+_sysprof_document_process_get_comm (SysprofDocumentProcess *self)
 {
   const SysprofCaptureProcess *proc;
 
@@ -170,6 +171,21 @@ sysprof_document_process_get_command_line (SysprofDocumentProcess *self)
   return SYSPROF_DOCUMENT_FRAME_CSTRING (self, proc->cmdline);
 }
 
+const char *
+sysprof_document_process_get_command_line (SysprofDocumentProcess *self)
+{
+  g_return_val_if_fail (SYSPROF_IS_DOCUMENT_PROCESS (self), 0);
+
+  /* Prefer ProcessInfo symbol name which gets updated from
+   * subsequent comm[] updates for process records.
+   */
+  if (self->process_info != NULL &&
+      self->process_info->symbol != NULL)
+    return self->process_info->symbol->name;
+
+  return _sysprof_document_process_get_comm (self);
+}
+
 /**
  * sysprof_document_process_list_memory_maps:
  * @self: a #SysprofDocumentProcess
diff --git a/src/libsysprof/sysprof-document.c b/src/libsysprof/sysprof-document.c
index a147071c..61d01ada 100644
--- a/src/libsysprof/sysprof-document.c
+++ b/src/libsysprof/sysprof-document.c
@@ -803,11 +803,16 @@ static void
 sysprof_document_load_processes (SysprofDocument *self)
 {
   G_GNUC_UNUSED SysprofProcessInfo *pid0;
+  g_autoptr(GHashTable) seen = NULL;
+  g_autoptr(EggBitset) duplicate = NULL;
   EggBitsetIter iter;
   guint i;
 
   g_assert (SYSPROF_IS_DOCUMENT (self));
 
+  seen = g_hash_table_new (NULL, NULL);
+  duplicate = egg_bitset_new_empty ();
+
   /* Always create PID 0 info */
   pid0 = _sysprof_document_process_info (self, 0, TRUE);
 
@@ -818,7 +823,7 @@ sysprof_document_load_processes (SysprofDocument *self)
           g_autoptr(SysprofDocumentProcess) process = g_list_model_get_item (G_LIST_MODEL (self), i);
           int pid = sysprof_document_frame_get_pid (SYSPROF_DOCUMENT_FRAME (process));
           SysprofProcessInfo *process_info = _sysprof_document_process_info (self, pid, TRUE);
-          const char *cmdline = sysprof_document_process_get_command_line (process);
+          const char *cmdline = _sysprof_document_process_get_comm (process);
 
           if (cmdline != NULL)
             {
@@ -836,9 +841,19 @@ sysprof_document_load_processes (SysprofDocument *self)
                                      NULL, NULL, 0, 0,
                                      SYSPROF_SYMBOL_KIND_PROCESS);
             }
+
+          /* Only include the first process discover for this process,
+           * ignore all secondary notations which are just comm[] updates.
+           */
+          if (g_hash_table_contains (seen, GINT_TO_POINTER (pid)))
+            egg_bitset_add (duplicate, i);
+          else
+            g_hash_table_add (seen, GINT_TO_POINTER (pid));
         }
       while (egg_bitset_iter_next (&iter, &i));
     }
+
+  egg_bitset_subtract (self->processes, duplicate);
 }
 
 static void