[<prev] [next>] [day] [month] [year] [list]
Message-Id: <20220208195421.969326-1-jsyoo5b@gmail.com>
Date: Wed, 9 Feb 2022 04:54:22 +0900
From: JaeSang Yoo <js.yoo.5b@...il.com>
To: rostedt@...dmis.org
Cc: mingo@...hat.com, linux-kernel@...r.kernel.org,
JaeSang Yoo <jsyoo5b@...il.com>
Subject: [PATCH v3] trace: param: fix tp_printk option related with tp_printk_stop_on_boot
Kernel param "tp_printk_stop_on_boot" starts with "tp_printk" which is
the exact as the other kernel param "tp_printk".
In compile & build process, It may not guaranteed that
"tp_printk_stop_on_boot" always checked before "tp_printk".
(By swapping its __setup() macro order, it may not work as expected.)
Some kernel params which starts with other kernel params consider this
problem. See commit 745a600cf1a6 ("um: console: Ignore console= option")
or init/do_mounts.c:45 (setup function of "ro" kernel param)
Kernel param "tp_printk" can be handled with its value(0 or off) or
it can be handled without its value. (maybe it won't effect anything)
Fix setup function to ignore when the "tp_printk" becomes prefix of
other kernel param.
Signed-off-by: JaeSang Yoo <jsyoo5b@...il.com>
---
from v2, I thought the changes were obvious that I didn't even tried to
check whether it has obvious typo. Sorry about that.
kernel/trace/trace.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
index c860f582b078..48f34ab29db2 100644
--- a/kernel/trace/trace.c
+++ b/kernel/trace/trace.c
@@ -252,6 +252,9 @@ __setup("trace_clock=", set_trace_boot_clock);
static int __init set_tracepoint_printk(char *str)
{
+ /* Ignore the "tp_printk_stop_on_boot" param */
+ if (*str == '_')
+ return 0;
if ((strcmp(str, "=0") != 0 && strcmp(str, "=off") != 0))
tracepoint_printk = 1;
return 1;
--
2.25.1
Powered by blists - more mailing lists