[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <200810250026.EGB05240.HJtQFOMSOVOFFL@I-love.SAKURA.ne.jp>
Date: Sat, 25 Oct 2008 00:26:15 +0900
From: Tetsuo Handa <penguin-kernel@...ove.SAKURA.ne.jp>
To: greg@...ah.com
Cc: linux-usb@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 2.6.27.3 usbcore] Move __module_param_call(nousb) toimmediately after declaration of nousb.
Hello.
I succeeded to produce this problem with 2.6.26 .
Thus, at least, it is not a new bug introduced in 2.6.27 .
Greg KH wrote:
> This really sounds like a compiler bug somewhere, reordering this
> shouldn't "fix" this issue.
I think so too.
> Is 3.3.5 still supported by the Debian team?
I think Security Support for Debian Sarge terminated on March 31st 2008.
> I know the minimum gcc version is 3.2 for the kernel, so it's
> strange that no one else sees this.
Thus, it is likely that nobody is using gcc (GCC) 3.3.5 (Debian 1:3.3.5-13).
> Can you duplicate this in a stand-alone test module?
Here is the patch for 2.6.26.7 and 2.6.27.3 .
But... please check this patch carefully. It seems that something is wrong.
----------------------------------------
---
drivers/usb/Makefile | 2 ++
drivers/usb/paramtest.c | 38 ++++++++++++++++++++++++++++++++++++++
2 files changed, 40 insertions(+)
--- linux-2.6.26.7.orig/drivers/usb/Makefile
+++ linux-2.6.26.7/drivers/usb/Makefile
@@ -2,6 +2,8 @@
# Makefile for the kernel USB device drivers.
#
+obj-y += paramtest.o
+
# Object files in subdirectories
obj-$(CONFIG_USB) += core/
--- /dev/null
+++ linux-2.6.26.7/drivers/usb/paramtest.c
@@ -0,0 +1,38 @@
+#include <linux/module.h>
+#include <linux/moduleparam.h>
+
+static int param1;
+module_param(param1, bool, S_IRUGO | S_IWUSR);
+MODULE_PARM_DESC(param1, "param1");
+
+static int param2;
+module_param (param2, bool, S_IRUGO);
+MODULE_PARM_DESC (param2, "param2");
+
+static int param3;
+module_param(param3, bool, S_IRUGO | S_IWUSR);
+MODULE_PARM_DESC(param3, "param3");
+
+static int param4;
+module_param(param4, bool, S_IRUGO | S_IWUSR);
+MODULE_PARM_DESC(param4, "param4");
+
+static int param5;
+__module_param_call("", param5, param_set_bool, param_get_bool, ¶m5, 0444);
+
+static int param6;
+module_param_named(paramtest, param6, int, 0644);
+MODULE_PARM_DESC(paramtest, "param6");
+
+static int __init paramtest_init(void)
+{
+ return 0;
+}
+
+static void __exit paramtest_exit(void)
+{
+}
+
+subsys_initcall(paramtest_init);
+module_exit(paramtest_exit);
+MODULE_LICENSE("GPL");
----------------------------------------
Did I make a mistake in the above patch?
I tested after applying the above patch, and all results are wrong.
----------------------------------------
# cat /proc/version
Linux version 2.6.27.3 (root@...oyo) (gcc version 4.1.2 20071124 (Red Hat 4.1.2-42)) #2 SMP Fri Oct 24 23:45:52 JST 2008
# ls -ail /sys/module/paramtest/parameters/
total 0
197 drwxr-xr-x 2 root root 0 Oct 24 23:49 .
196 drwxr-xr-x 3 root root 0 Oct 24 23:49 ..
199 -r--r--r-- 1 root root 4096 Oct 24 23:49 amtest.param4
202 -r--r--r-- 1 root root 4096 Oct 24 23:49 param2
201 -rw-r--r-- 1 root root 4096 Oct 24 23:49 param3
200 -rw-r--r-- 1 root root 4096 Oct 24 23:49 param4
198 -rw-r--r-- 1 root root 4096 Oct 24 23:49 paramtest
----------
# cat /proc/version
Linux version 2.6.26.7 (root@...oyo) (gcc version 4.1.2 20071124 (Red Hat 4.1.2-42)) #2 SMP Fri Oct 24 23:47:07 JST 2008
# ls -ail /sys/module/paramtest/parameters/
total 0
197 drwxr-xr-x 2 root root 0 Oct 24 23:52 .
196 drwxr-xr-x 3 root root 0 Oct 24 23:52 ..
199 -r--r--r-- 1 root root 4096 Oct 24 23:52 amtest.param4
202 -r--r--r-- 1 root root 4096 Oct 24 23:52 param2
201 -rw-r--r-- 1 root root 4096 Oct 24 23:52 param3
200 -rw-r--r-- 1 root root 4096 Oct 24 23:52 param4
198 -rw-r--r-- 1 root root 4096 Oct 24 23:52 paramtest
----------
# cat /proc/version
Linux version 2.6.27.3 (root@...oyo) (gcc version 3.3.5 (Debian 1:3.3.5-13)) #2 SMP Fri Oct 24 22:12:09 JST 2008
tomoyo:~# ls -ail /sys/module/paramtest/parameters/
total 0
197 drwxr-xr-x 2 root root 0 Oct 24 22:14 .
196 drwxr-xr-x 3 root root 0 Oct 24 22:14 ..
202 -r--r--r-- 1 root root 4096 Oct 24 22:14 amtest.paramtest
198 -rw-r--r-- 1 root root 4096 Oct 24 22:14 param1
199 -r--r--r-- 1 root root 4096 Oct 24 22:14 param2
200 -rw-r--r-- 1 root root 4096 Oct 24 22:14 param3
201 -rw-r--r-- 1 root root 4096 Oct 24 22:14 param4
----------
# cat /proc/version
Linux version 2.6.26.7 (root@...oyo) (gcc version 3.3.5 (Debian 1:3.3.5-13)) #2 SMP Fri Oct 24 22:10:11 JST 2008
tomoyo:~# ls -ail /sys/module/paramtest/parameters/
total 0
197 drwxr-xr-x 2 root root 0 Oct 24 22:16 .
196 drwxr-xr-x 3 root root 0 Oct 24 22:16 ..
202 -r--r--r-- 1 root root 4096 Oct 24 22:16 amtest.paramtest
198 -rw-r--r-- 1 root root 4096 Oct 24 22:16 param1
199 -r--r--r-- 1 root root 4096 Oct 24 22:16 param2
200 -rw-r--r-- 1 root root 4096 Oct 24 22:16 param3
201 -rw-r--r-- 1 root root 4096 Oct 24 22:16 param4
----------------------------------------
However, applying the below patch after the above patch
prevents this problem from appearing.
----------------------------------------
---
drivers/usb/paramtest.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
--- linux-2.6.26.7.orig/drivers/usb/paramtest.c
+++ linux-2.6.26.7/drivers/usb/paramtest.c
@@ -18,12 +18,13 @@ module_param(param4, bool, S_IRUGO | S_I
MODULE_PARM_DESC(param4, "param4");
static int param5;
-__module_param_call("", param5, param_set_bool, param_get_bool, ¶m5, 0444);
static int param6;
module_param_named(paramtest, param6, int, 0644);
MODULE_PARM_DESC(paramtest, "param6");
+__module_param_call("", param5, param_set_bool, param_get_bool, ¶m5, 0444);
+
static int __init paramtest_init(void)
{
return 0;
----------------------------------------
----------------------------------------
# cat /proc/version
Linux version 2.6.27.3 (root@...oyo) (gcc version 4.1.2 20071124 (Red Hat 4.1.2-42)) #3 SMP Fri Oct 24 23:55:21 JST 2008
# ls -ail /sys/module/paramtest/parameters/
total 0
197 drwxr-xr-x 2 root root 0 Oct 25 00:00 .
196 drwxr-xr-x 3 root root 0 Oct 25 00:00 ..
202 -rw-r--r-- 1 root root 4096 Oct 25 00:00 param1
201 -r--r--r-- 1 root root 4096 Oct 25 00:00 param2
200 -rw-r--r-- 1 root root 4096 Oct 25 00:00 param3
199 -rw-r--r-- 1 root root 4096 Oct 25 00:00 param4
198 -rw-r--r-- 1 root root 4096 Oct 25 00:00 paramtest
----------
# cat /proc/version
Linux version 2.6.26.7 (root@...oyo) (gcc version 4.1.2 20071124 (Red Hat 4.1.2-42)) #3 SMP Fri Oct 24 23:56:17 JST 2008
# ls -ail /sys/module/paramtest/parameters/
total 0
197 drwxr-xr-x 2 root root 0 Oct 25 00:01 .
196 drwxr-xr-x 3 root root 0 Oct 25 00:01 ..
202 -rw-r--r-- 1 root root 4096 Oct 25 00:01 param1
201 -r--r--r-- 1 root root 4096 Oct 25 00:01 param2
200 -rw-r--r-- 1 root root 4096 Oct 25 00:01 param3
199 -rw-r--r-- 1 root root 4096 Oct 25 00:01 param4
198 -rw-r--r-- 1 root root 4096 Oct 25 00:01 paramtest
----------
# cat /proc/version
Linux version 2.6.27.3 (root@...oyo) (gcc version 3.3.5 (Debian 1:3.3.5-13)) #3 SMP Fri Oct 24 22:18:06 JST 2008
tomoyo:~# ls -ail /sys/module/paramtest/parameters/
total 0
197 drwxr-xr-x 2 root root 0 Oct 24 22:21 .
196 drwxr-xr-x 3 root root 0 Oct 24 22:21 ..
198 -rw-r--r-- 1 root root 4096 Oct 24 22:21 param1
199 -r--r--r-- 1 root root 4096 Oct 24 22:21 param2
200 -rw-r--r-- 1 root root 4096 Oct 24 22:21 param3
201 -rw-r--r-- 1 root root 4096 Oct 24 22:21 param4
202 -rw-r--r-- 1 root root 4096 Oct 24 22:21 paramtest
----------
# cat /proc/version
Linux version 2.6.26.7 (root@...oyo) (gcc version 3.3.5 (Debian 1:3.3.5-13)) #3 SMP Fri Oct 24 22:19:26 JST 2008
tomoyo:~# ls -ail /sys/module/paramtest/parameters/
total 0
197 drwxr-xr-x 2 root root 0 Oct 24 22:24 .
196 drwxr-xr-x 3 root root 0 Oct 24 22:24 ..
198 -rw-r--r-- 1 root root 4096 Oct 24 22:24 param1
199 -r--r--r-- 1 root root 4096 Oct 24 22:24 param2
200 -rw-r--r-- 1 root root 4096 Oct 24 22:24 param3
201 -rw-r--r-- 1 root root 4096 Oct 24 22:24 param4
202 -rw-r--r-- 1 root root 4096 Oct 24 22:24 paramtest
----------------------------------------
Well, I'm getting more and more confused.
It may be gcc3's bug, it may not be gcc's bug.
Regards.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists