[git commit] ash: INT_OFF/INT_ON around run_nofork_applet()

Denys Vlasenko vda.linux at googlemail.com
Fri Aug 4 12:50:03 UTC 2017


commit: https://git.busybox.net/busybox/commit/?id=d329e34c96e0429602fe39489586cd61f97a2877
branch: https://git.busybox.net/busybox/commit/?id=refs/heads/master

function                                             old     new   delta
evalcommand                                         1441    1447      +6

Signed-off-by: Denys Vlasenko <vda.linux at googlemail.com>
---
 shell/ash.c | 22 +++++++++++++++++++---
 1 file changed, 19 insertions(+), 3 deletions(-)

diff --git a/shell/ash.c b/shell/ash.c
index ca9926b..2afa5e8 100644
--- a/shell/ash.c
+++ b/shell/ash.c
@@ -9919,10 +9919,26 @@ evalcommand(union node *cmd, int flags)
 		int applet_no = (- cmdentry.u.index - 2);
 		if (applet_no >= 0 && APPLET_IS_NOFORK(applet_no)) {
 			listsetvar(varlist.list, VEXPORT|VSTACK);
-			/* run <applet>_main() */
-//FIXME: do we need INT_OFF / INT_ON here?
-//wouldn't open files and allocations leak on ^C otherwise?
+			/*
+			 * Run <applet>_main().
+			 * Signals (^C) can't interrupt here.
+			 * Otherwise we can mangle stdio or malloc internal state.
+			 * This makes applets which can run for a long time
+			 * and/or wait for user input ineligible for NOFORK:
+			 * for example, "yes" or "rm" (rm -i waits for input).
+			 */
+			INT_OFF;
 			status = run_nofork_applet(applet_no, argv);
+			/*
+			 * Try enabling NOFORK for "yes" applet.
+			 * ^C _will_ stop it (write returns EINTR),
+			 * but this causes stdout FILE to be stuck
+			 * and needing clearerr(). What if other applets
+			 * also can get EINTRs? Do we need to switch
+			 * our signals to SA_RESTART?
+			 */
+			/*clearerr(stdout);*/
+			INT_ON;
 			break;
 		}
 #endif


More information about the busybox-cvs mailing list