* [PATCH] fix character subsystem initialisation panic
@ 2003-06-08 17:35 James Bottomley
2003-06-08 18:43 ` Christoph Hellwig
0 siblings, 1 reply; 2+ messages in thread
From: James Bottomley @ 2003-06-08 17:35 UTC (permalink / raw)
To: torvalds; +Cc: Andrew Morton, greg, Linux Kernel
[-- Attachment #1: Type: text/plain, Size: 733 bytes --]
In 2.5.70 bk latest, I'm getting a panic related to character device
initialisation. The problem seems to be that the new sysfs entries for
character devices require that everything now have a properly
initialised parent. However, the character subsystem is set up in
drivers/char/mem.c as
__initcall(chr_dev_init);
However, __initcall() is the same priority as module_init(), so whether
character devices are initialised before their required subsystem
depends purely on link ordering (on parisc, we initialise devices/parisc
before everything else, so it is panicing reliably with this).
I think the fix is to convert the __initcall to subsys_initcall (patch
attached). The patch allows parisc to boot properly now.
James
[-- Attachment #2: tmp.diff --]
[-- Type: text/plain, Size: 259 bytes --]
===== drivers/char/mem.c 1.39 vs edited =====
--- 1.39/drivers/char/mem.c Fri Jun 6 01:36:40 2003
+++ edited/drivers/char/mem.c Sun Jun 8 12:02:24 2003
@@ -713,4 +713,4 @@
return 0;
}
-__initcall(chr_dev_init);
+subsys_initcall(chr_dev_init);
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [PATCH] fix character subsystem initialisation panic
2003-06-08 17:35 [PATCH] fix character subsystem initialisation panic James Bottomley
@ 2003-06-08 18:43 ` Christoph Hellwig
0 siblings, 0 replies; 2+ messages in thread
From: Christoph Hellwig @ 2003-06-08 18:43 UTC (permalink / raw)
To: James Bottomley; +Cc: torvalds, Andrew Morton, greg, Linux Kernel
On Sun, Jun 08, 2003 at 12:35:25PM -0500, James Bottomley wrote:
> In 2.5.70 bk latest, I'm getting a panic related to character device
> initialisation. The problem seems to be that the new sysfs entries for
> character devices require that everything now have a properly
> initialised parent. However, the character subsystem is set up in
> drivers/char/mem.c as
>
> __initcall(chr_dev_init);
>
> However, __initcall() is the same priority as module_init(), so whether
> character devices are initialised before their required subsystem
> depends purely on link ordering (on parisc, we initialise devices/parisc
> before everything else, so it is panicing reliably with this).
>
> I think the fix is to convert the __initcall to subsys_initcall (patch
> attached). The patch allows parisc to boot properly now.
This means that all the drivers initialized from chr_dev_init are
also at subsystem level now which doesn't seem to break something
(yet) but still is v ery confusing. This patch converts all of
them (except fbmem_init which is in a different directory so
we could get link order problems) to module_init.
--- 1.40/drivers/char/mem.c Sun Jun 8 14:02:24 2003
+++ edited/drivers/char/mem.c Sat Jun 7 22:30:30 2003
@@ -698,17 +698,8 @@
S_IFCHR | devlist[i].mode, devlist[i].name);
}
- rand_initialize();
#if defined (CONFIG_FB)
fbmem_init();
-#endif
- tty_init();
-#ifdef CONFIG_M68K_PRINTER
- lp_m68k_init();
-#endif
- misc_init();
-#ifdef CONFIG_FTAPE
- ftape_init();
#endif
return 0;
}
--- 1.20/drivers/char/misc.c Sat May 17 21:39:13 2003
+++ edited/drivers/char/misc.c Sat Jun 7 22:25:37 2003
@@ -244,7 +244,7 @@
EXPORT_SYMBOL(misc_register);
EXPORT_SYMBOL(misc_deregister);
-int __init misc_init(void)
+static int __init misc_init(void)
{
create_proc_read_entry("misc", 0, 0, misc_read_proc, NULL);
#ifdef CONFIG_MVME16x
@@ -281,3 +281,4 @@
}
return 0;
}
+module_init(misc_init);
--- 1.33/drivers/char/random.c Mon May 5 07:49:54 2003
+++ edited/drivers/char/random.c Sat Jun 7 22:23:12 2003
@@ -1420,7 +1420,7 @@
}
}
-void __init rand_initialize(void)
+static void __init rand_initialize(void)
{
int i;
@@ -1443,6 +1443,7 @@
memset(&extract_timer_state, 0, sizeof(struct timer_rand_state));
extract_timer_state.dont_count_entropy = 1;
}
+module_init(rand_initialize);
void rand_initialize_irq(int irq)
{
--- 1.106/drivers/char/tty_io.c Fri Jun 6 08:36:47 2003
+++ edited/drivers/char/tty_io.c Sat Jun 7 22:31:23 2003
@@ -2364,7 +2364,7 @@
* Ok, now we can initialize the rest of the tty devices and can count
* on memory allocations, interrupts etc..
*/
-void __init tty_init(void)
+static void __init tty_init(void)
{
strcpy(tty_cdev.kobj.name, "dev.tty");
cdev_init(&tty_cdev, &tty_fops);
@@ -2454,3 +2454,4 @@
a2232board_init();
#endif
}
+module_init(tty_init);
--- 1.4/drivers/char/ftape/lowlevel/ftape-init.c Mon Feb 3 21:19:37 2003
+++ edited/drivers/char/ftape/lowlevel/ftape-init.c Sat Jun 7 22:28:33 2003
@@ -55,14 +55,24 @@
char ft_dat[] __initdata = "$Date: 1997/11/06 00:38:08 $";
+#ifndef CONFIG_FT_NO_TRACE_AT_ALL
+static int ft_tracing = -1;
+#endif
+
+
/* Called by modules package when installing the driver
* or by kernel during the initialization phase
*/
-int __init ftape_init(void)
+static int __init ftape_init(void)
{
TRACE_FUN(ft_t_flow);
#ifdef MODULE
+#ifndef CONFIG_FT_NO_TRACE_AT_ALL
+ if (ft_tracing != -1) {
+ ftape_tracing = ft_tracing;
+ }
+#endif
printk(KERN_INFO FTAPE_VERSION "\n");
if (TRACE_LEVEL >= ft_t_info) {
printk(
@@ -112,13 +122,6 @@
#endif
TRACE_EXIT 0;
}
-
-#ifdef MODULE
-
-#ifndef CONFIG_FT_NO_TRACE_AT_ALL
-static int ft_tracing = -1;
-#endif
-
#define FT_MOD_PARM(var,type,desc) \
MODULE_PARM(var,type); MODULE_PARM_DESC(var,desc)
@@ -141,21 +144,7 @@
"QIC-117 driver for QIC-40/80/3010/3020 floppy tape drives.");
MODULE_LICENSE("GPL");
-/* Called by modules package when installing the driver
- */
-int init_module(void)
-{
-#ifndef CONFIG_FT_NO_TRACE_AT_ALL
- if (ft_tracing != -1) {
- ftape_tracing = ft_tracing;
- }
-#endif
- return ftape_init();
-}
-
-/* Called by modules package when removing the driver
- */
-void cleanup_module(void)
+static void __exit ftape_exit(void)
{
TRACE_FUN(ft_t_flow);
@@ -167,3 +156,6 @@
TRACE_EXIT;
}
#endif /* MODULE */
+
+module_init(ftape_init);
+module_exit(ftape_exit);
--- 1.3/include/linux/ftape.h Tue Apr 1 01:55:26 2003
+++ edited/include/linux/ftape.h Sat Jun 7 22:26:34 2003
@@ -199,8 +199,6 @@
#define ABS(a) ((a) < 0 ? -(a) : (a))
#define NR_ITEMS(x) (int)(sizeof(x)/ sizeof(*x))
-extern int ftape_init(void);
-
#endif /* __KERNEL__ */
#endif
--- 1.7/include/linux/miscdevice.h Mon Apr 21 01:21:19 2003
+++ edited/include/linux/miscdevice.h Sat Jun 7 22:26:22 2003
@@ -34,8 +34,6 @@
#define TUN_MINOR 200
-extern int misc_init(void);
-
struct miscdevice
{
int minor;
--- 1.2/include/linux/random.h Mon Oct 28 20:57:55 2002
+++ edited/include/linux/random.h Sat Jun 7 22:24:29 2003
@@ -42,7 +42,6 @@
#ifdef __KERNEL__
-extern void rand_initialize(void);
extern void rand_initialize_irq(int irq);
extern void batch_entropy_store(u32 a, u32 b, int num);
--- 1.18/include/linux/tty.h Mon May 26 08:19:13 2003
+++ edited/include/linux/tty.h Sat Jun 7 22:24:19 2003
@@ -351,7 +351,6 @@
extern int lp_init(void);
extern int pty_init(void);
-extern void tty_init(void);
extern int mxser_init(void);
extern int moxa_init(void);
extern int ip2_init(void);
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2003-06-08 18:30 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2003-06-08 17:35 [PATCH] fix character subsystem initialisation panic James Bottomley
2003-06-08 18:43 ` Christoph Hellwig
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).