[glib/th/g-close-ebadf: 6/6] gstdio: fail assertion in g_close() for invalid file descriptor (EBADF)

[Philip Withnall <pwithnall src gnome org>]

commit d5dc7d266f2b8d0f7d12c878c5a8dc179af14a97
Author: Thomas Haller <thaller redhat com>
Date:   Mon Oct 17 16:23:35 2022 +0200

    gstdio: fail assertion in g_close() for invalid file descriptor (EBADF)
    
    An application must keep track of the file descriptors that it
    has. Closing an invalid, non-negative file descriptor is usually
    a bug, because it indicates somebody messed up the tracking.
    
    On a single threaded application it may be fine, but EBADF is always a bug
    in a multi threaded application because another thread might race
    reusing the bad file descriptor. With GDBus and other glib API, it is very
    common that your application has multiple threads running and this is
    in fact a bug.
    
    The assertion failure does not necessarily indicate that the bug
    is in the caller. It could have been another part of the application
    that wrongly closed the file descriptor.

 glib/gstdio.c | 54 ++++++++++++++++++++++++++++++++++++++++++------------
 1 file changed, 42 insertions(+), 12 deletions(-)
---
diff --git a/glib/gstdio.c b/glib/gstdio.c
index f2d58134ef..ed46bdf2dd 100644
--- a/glib/gstdio.c
+++ b/glib/gstdio.c
@@ -1757,6 +1757,8 @@ g_utime (const gchar    *filename,
  * attempt to correctly handle %EINTR, which has platform-specific
  * semantics.
  *
+ * It is a bug to call this function with an invalid file descriptor.
+ *
  * Returns: %TRUE on success, %FALSE if there was an error.
  *
  * Since: 2.36
@@ -1766,26 +1768,54 @@ g_close (gint       fd,
          GError   **error)
 {
   int res;
+
   res = close (fd);
-  /* Just ignore EINTR for now; a retry loop is the wrong thing to do
-   * on Linux at least.  Anyone who wants to add a conditional check
-   * for e.g. HP-UX is welcome to do so later...
-   *
-   * http://lkml.indiana.edu/hypermail/linux/kernel/0509.1/0877.html
-   * https://bugzilla.gnome.org/show_bug.cgi?id=682819
-   * http://utcc.utoronto.ca/~cks/space/blog/unix/CloseEINTR
-   * 
https://sites.google.com/site/michaelsafyan/software-engineering/checkforeintrwheninvokingclosethinkagain
-   */
-  if (G_UNLIKELY (res == -1 && errno == EINTR))
-    return TRUE;
-  else if (res == -1)
+
+  if (res == -1)
     {
       int errsv = errno;
+
+      if (errsv == EINTR)
+        {
+          /* Just ignore EINTR for now; a retry loop is the wrong thing to do
+           * on Linux at least.  Anyone who wants to add a conditional check
+           * for e.g. HP-UX is welcome to do so later...
+           *
+           * https://lwn.net/Articles/576478/
+           * http://lkml.indiana.edu/hypermail/linux/kernel/0509.1/0877.html
+           * https://bugzilla.gnome.org/show_bug.cgi?id=682819
+           * http://utcc.utoronto.ca/~cks/space/blog/unix/CloseEINTR
+           * 
https://sites.google.com/site/michaelsafyan/software-engineering/checkforeintrwheninvokingclosethinkagain
+           */
+          return TRUE;
+        }
+
       g_set_error_literal (error, G_FILE_ERROR,
                            g_file_error_from_errno (errsv),
                            g_strerror (errsv));
+
+      if (errsv == EBADF)
+        {
+          if (fd >= 0)
+            {
+              /* Closing an non-negative, invalid file descriptor is a bug. The bug is
+               * not necessarily in the caller of g_close(), but somebody else
+               * might have wrongly closed fd. In any case, there is a serious bug
+               * somewhere. */
+              g_critical ("g_close(fd:%d) failed with EBADF. The tracking of file descriptors got messed 
up", fd);
+            }
+          else
+            {
+              /* Closing a negative "file descriptor" is less problematic. It's still a nonsensical action
+               * from the caller. Assert against that too. */
+              g_critical ("g_close(fd:%d) failed with EBADF. This is not a valid file descriptor", fd);
+            }
+        }
+
       errno = errsv;
+
       return FALSE;
     }
+
   return TRUE;
 }