1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
|
* A problem with non-interactive signal handling has been fixed.
Patch: fix-batch-mode-signal-handling.diff
Added-by: Rob Browning <rlb@defaultvalue.org>
Status: patch backported from upstream
An upstream patch from 2003-06-21 was backported to fix Bug#253887.
Under certain conditions running "emacs --batch ... < /dev/null"
would trigger a process exit via SIGIO or SIGHUP.
diff -urNad /home/rlb/deb/emacs/trunk-21/src/keyboard.c trunk-21/src/keyboard.c
--- /home/rlb/deb/emacs/trunk-21/src/keyboard.c 2003-09-09 17:26:14.000000000 -0500
+++ trunk-21/src/keyboard.c 2004-07-18 17:10:35.000000000 -0500
@@ -6186,7 +6186,12 @@
/* ??? Is it really right to send the signal just to this process
rather than to the whole process group?
Perhaps on systems with FIONREAD Emacs is alone in its group. */
- kill (getpid (), SIGHUP);
+ {
+ if (! noninteractive)
+ kill (getpid (), SIGHUP);
+ else
+ n_to_read = 0;
+ }
if (n_to_read == 0)
return 0;
if (n_to_read > sizeof cbuf)
diff -urNad /home/rlb/deb/emacs/trunk-21/src/process.c trunk-21/src/process.c
--- /home/rlb/deb/emacs/trunk-21/src/process.c 2003-09-09 17:28:35.000000000 -0500
+++ trunk-21/src/process.c 2004-07-18 17:11:21.000000000 -0500
@@ -2734,7 +2734,7 @@
but select says there is input. */
if (XINT (read_kbd) && interrupt_input
- && keyboard_bit_set (&Available))
+ && keyboard_bit_set (&Available) && ! noninteractive)
kill (getpid (), SIGIO);
#endif
|
