Bug 203437

Summary: scripts/kconfig/nconf.c stuck in infinite loop in search
Product: Tools Reporter: GYt2bW (howaboutsynergy)
Component: OtherAssignee: Tools.Other (tools_other)
Status: CLOSED CODE_FIX    
Severity: normal CC: adjudicatordarren, howaboutsynergy, rdunlap
Priority: P1    
Hardware: x86-64   
OS: Linux   
Kernel Version: v5.1-rc6 Subsystem:
Regression: No Bisected commit-id:
Attachments: naive workaround

Description GYt2bW 2019-04-26 15:04:21 UTC 
Created attachment 282549 [details]
naive workaround

Repro. steps:
1. $ make nconfig
2. Go to `File Systems`
3. Press `/` to search and type `cp`
4. Press and release DownArrow key, then UpArrow key
The `nconf` process is now stuck in an infinite loop using 100% CPU (one full core) 
5. press C-c to stop



This happens because `match_start` is `-1` and thus, since `index` will not reach values less than `0` (apparently), it won't ever exit the `while (true)` loop.

naive workaround(also attached):

```diff
diff --git a/scripts/kconfig/nconf.c b/scripts/kconfig/nconf.c
index ac92c0ded6c5..9f1d85acc572 100644
--- a/scripts/kconfig/nconf.c
+++ b/scripts/kconfig/nconf.c
@@ -505,7 +505,7 @@ static int get_mext_match(const char *match_str, match_f flag)
 
 	index = match_start;
 	index = (index + items_num) % items_num;
-	while (true) {
+	while (match_start != -1) {
 		char *str = k_menu_items[index].str;
 		if (strcasestr(str, match_str) != NULL)
 			return index;
@@ -516,8 +516,9 @@ static int get_mext_match(const char *match_str, match_f flag)
 			++index;
 		index = (index + items_num) % items_num;
 		if (index == match_start)
-			return -1;
+			break;
 	}
+	return -1;
 }
 
 /* Make a new item. */
```
Comment 1 Randy Dunlap 2019-09-26 03:13:42 UTC 
Hi,
Please report this problem and your solution to
linux-kbuild@vger.kernel.org .  I would expect the kbuild/kconfig
maintainer to reply to it there.
Thanks.
Comment 2 GYt2bW 2019-09-27 16:33:16 UTC 
(In reply to Randy Dunlap from comment #1)
yeah, I was afraid it wasn't that simple :)
now I'm being asked to do things from this: https://github.com/torvalds/linux/blob/master/Documentation/process/submitting-patches.rst

and let me just say it's never gonna happen.
Comment 3 GYt2bW 2019-09-27 16:36:21 UTC 
besides, it is a `naive workaround`, I expect actual devs to fix it properly :D
Comment 4 Randy Dunlap 2019-09-27 16:40:49 UTC 
The kbuild/kconfig maintainer is just asking for an email with the problem description and patch.  It's pretty easy to do that.  Although the patch would need a Signed-off-by: line with your real name and email address in it.

We mostly work through email, not bugzilla.  Anyway, thanks for the report and analysis.