* [PATCH] Eat keys on panic
@ 2003-05-31 11:56 Andi Kleen
2003-05-31 14:14 ` Alan Cox
0 siblings, 1 reply; 4+ messages in thread
From: Andi Kleen @ 2003-05-31 11:56 UTC (permalink / raw)
To: linux-kernel; +Cc: akpm
[forgot to cc linux-kernel with the other mail]
I have a weird KVM that refuses to switch consoles when the keyboard
controller doesn't process key presses. Actually it switches, but only
after a long delay which usually drives me crazy.
This patch changes the panic loop to eat keys from the keyboard controller
on i386 boxes. With that I can switch consoles after a panic.
The ioport should be available on all PC like boxes, the reset code also
depends on it being at a fixed place.
ioport.c is not the best place in the world to place the function,
but I didn't find a better one. It cannot be inlined to avoid include
hell.
-Andi
Index: linux/arch/i386/kernel/ioport.c
===================================================================
RCS file: /home/cvs/linux-2.5/arch/i386/kernel/ioport.c,v
retrieving revision 1.17
diff -u -u -r1.17 ioport.c
--- linux/arch/i386/kernel/ioport.c 30 May 2003 20:12:29 -0000 1.17
+++ linux/arch/i386/kernel/ioport.c 31 May 2003 10:24:18 -0000
@@ -15,6 +15,7 @@
#include <linux/stddef.h>
#include <linux/slab.h>
#include <linux/thread_info.h>
+#include <asm/io.h>
/* Set EXTENT bits starting at BASE in BITMAP to value TURN_ON. */
static void set_bitmap(unsigned long *bitmap, short base, short extent, int new_value)
@@ -128,4 +129,10 @@
/* Make sure we return the long way (not sysenter) */
set_thread_flag(TIF_IRET);
return 0;
+}
+
+void eat_key(void)
+{
+ if (inb(0x60) & 1)
+ inb(0x64);
}
Index: linux/include/asm-i386/system.h
===================================================================
RCS file: /home/cvs/linux-2.5/include/asm-i386/system.h,v
retrieving revision 1.37
diff -u -u -r1.37 system.h
--- linux/include/asm-i386/system.h 30 May 2003 20:10:39 -0000 1.37
+++ linux/include/asm-i386/system.h 31 May 2003 10:24:24 -0000
@@ -476,4 +476,7 @@
#define BROKEN_PNP_BIOS 0x0004
#define BROKEN_CPUFREQ 0x0008
+#define HAVE_EAT_KEY
+void eat_key(void);
+
#endif
Index: linux/kernel/panic.c
===================================================================
RCS file: /home/cvs/linux-2.5/kernel/panic.c,v
retrieving revision 1.20
diff -u -u -r1.20 panic.c
--- linux/kernel/panic.c 30 May 2003 20:14:20 -0000 1.20
+++ linux/kernel/panic.c 31 May 2003 10:24:26 -0000
@@ -96,8 +96,11 @@
disabled_wait(caller);
#endif
local_irq_enable();
- for (;;)
- ;
+ for (;;) {
+#ifdef HAVE_EAT_KEY
+ eat_key();
+#endif
+ }
}
/**
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] Eat keys on panic
2003-05-31 11:56 [PATCH] Eat keys on panic Andi Kleen
@ 2003-05-31 14:14 ` Alan Cox
2003-05-31 15:45 ` Andi Kleen
0 siblings, 1 reply; 4+ messages in thread
From: Alan Cox @ 2003-05-31 14:14 UTC (permalink / raw)
To: Andi Kleen; +Cc: Linux Kernel Mailing List, akpm
On Sad, 2003-05-31 at 12:56, Andi Kleen wrote:
> +void eat_key(void)
> +{
> + if (inb(0x60) & 1)
> + inb(0x64);
> }
This will crash at least one of my machines. The keyboard controller
has mandatory access delays of up to 1mS. Respect them or some stuff
dies horribly.
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] Eat keys on panic
2003-05-31 14:14 ` Alan Cox
@ 2003-05-31 15:45 ` Andi Kleen
2003-05-31 15:58 ` Adrian Bunk
0 siblings, 1 reply; 4+ messages in thread
From: Andi Kleen @ 2003-05-31 15:45 UTC (permalink / raw)
To: Alan Cox; +Cc: Andi Kleen, Linux Kernel Mailing List, akpm
On Sat, May 31, 2003 at 03:14:51PM +0100, Alan Cox wrote:
> On Sad, 2003-05-31 at 12:56, Andi Kleen wrote:
>
> > +void eat_key(void)
> > +{
> > + if (inb(0x60) & 1)
> > + inb(0x64);
> > }
>
> This will crash at least one of my machines. The keyboard controller
> has mandatory access delays of up to 1mS. Respect them or some stuff
> dies horribly.
1ms? Is it clocked in Khz or something?
Here is a new version that is hopefully Alan's-broken-chipset proof.
I also added the request comment.
-Andi
P.S.: This is different from the panic blink code. The two can easily
coexist though.
Index: linux/arch/i386/kernel/ioport.c
===================================================================
RCS file: /home/cvs/linux-2.5/arch/i386/kernel/ioport.c,v
retrieving revision 1.17
diff -u -u -r1.17 ioport.c
--- linux/arch/i386/kernel/ioport.c 30 May 2003 20:12:29 -0000 1.17
+++ linux/arch/i386/kernel/ioport.c 31 May 2003 14:24:16 -0000
@@ -15,6 +15,8 @@
#include <linux/stddef.h>
#include <linux/slab.h>
#include <linux/thread_info.h>
+#include <asm/io.h>
+#include <linux/delay.h>
/* Set EXTENT bits starting at BASE in BITMAP to value TURN_ON. */
static void set_bitmap(unsigned long *bitmap, short base, short extent, int new_value)
@@ -128,4 +130,14 @@
/* Make sure we return the long way (not sysenter) */
set_thread_flag(TIF_IRET);
return 0;
+}
+
+/* Some KVMs don't switch consoles unless the keyboard is served. */
+void eat_key(void)
+{
+ if (inb(0x60) & 1) {
+ mdelay(1);
+ inb(0x64);
+ }
+ mdelay(1);
}
Index: linux/include/asm-i386/system.h
===================================================================
RCS file: /home/cvs/linux-2.5/include/asm-i386/system.h,v
retrieving revision 1.37
diff -u -u -r1.37 system.h
--- linux/include/asm-i386/system.h 30 May 2003 20:10:39 -0000 1.37
+++ linux/include/asm-i386/system.h 31 May 2003 14:24:22 -0000
@@ -476,4 +476,7 @@
#define BROKEN_PNP_BIOS 0x0004
#define BROKEN_CPUFREQ 0x0008
+#define HAVE_EAT_KEY
+void eat_key(void);
+
#endif
Index: linux/kernel/panic.c
===================================================================
RCS file: /home/cvs/linux-2.5/kernel/panic.c,v
retrieving revision 1.20
diff -u -u -r1.20 panic.c
--- linux/kernel/panic.c 30 May 2003 20:14:20 -0000 1.20
+++ linux/kernel/panic.c 31 May 2003 14:24:24 -0000
@@ -96,8 +96,11 @@
disabled_wait(caller);
#endif
local_irq_enable();
- for (;;)
- ;
+ for (;;) {
+#ifdef HAVE_EAT_KEY
+ eat_key();
+#endif
+ }
}
/**
>
>
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] Eat keys on panic
2003-05-31 15:45 ` Andi Kleen
@ 2003-05-31 15:58 ` Adrian Bunk
0 siblings, 0 replies; 4+ messages in thread
From: Adrian Bunk @ 2003-05-31 15:58 UTC (permalink / raw)
To: Andi Kleen; +Cc: Alan Cox, Linux Kernel Mailing List, akpm
On Sat, May 31, 2003 at 05:45:21PM +0200, Andi Kleen wrote:
>...
> --- linux/arch/i386/kernel/ioport.c 30 May 2003 20:12:29 -0000 1.17
> +++ linux/arch/i386/kernel/ioport.c 31 May 2003 14:24:16 -0000
> @@ -15,6 +15,8 @@
> #include <linux/stddef.h>
> #include <linux/slab.h>
> #include <linux/thread_info.h>
> +#include <asm/io.h>
> +#include <linux/delay.h>
>...
Could you move the #include <linux/delay.h> above the
#include <asm/io.h> ?
TIA
Adrian
--
"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2003-05-31 15:45 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2003-05-31 11:56 [PATCH] Eat keys on panic Andi Kleen
2003-05-31 14:14 ` Alan Cox
2003-05-31 15:45 ` Andi Kleen
2003-05-31 15:58 ` Adrian Bunk
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).