2006-03-13 18:10:20

by Zachary Amsden

[permalink] [raw]
Subject: [RFC, PATCH 14/24] i386 Vmi reboot fixes

Fix reboot to work with the VMI. We must support fallback to the standard
BIOS reboot mechanism. Turns out that this is required by kexec, and a good
idea for native hardware. We simply insert the NOP VMI reboot hook before
calling the BIOS reboot. While here, fix SMP reboot issues as well. The
problem is the halt() macro in VMI has been defined to be equivalent to
safe_halt(), which enables interrupts. Several call sites actually want to
disable interrupts and shutdown the processor, which is what VMI_Shutdown()
does.

Signed-off-by: Zachary Amsden <[email protected]>

Index: linux-2.6.16-rc5/arch/i386/kernel/crash.c
===================================================================
--- linux-2.6.16-rc5.orig/arch/i386/kernel/crash.c 2006-03-08 11:34:53.000000000 -0800
+++ linux-2.6.16-rc5/arch/i386/kernel/crash.c 2006-03-08 11:38:09.000000000 -0800
@@ -113,7 +113,7 @@ static int crash_nmi_callback(struct pt_
disable_local_APIC();
atomic_dec(&waiting_for_crash_ipi);
/* Assume hlt works */
- halt();
+ shutdown_halt();
for(;;);

return 1;
Index: linux-2.6.16-rc5/arch/i386/kernel/process.c
===================================================================
--- linux-2.6.16-rc5.orig/arch/i386/kernel/process.c 2006-03-08 11:38:06.000000000 -0800
+++ linux-2.6.16-rc5/arch/i386/kernel/process.c 2006-03-08 11:38:09.000000000 -0800
@@ -156,7 +156,7 @@ static inline void play_dead(void)
*/
local_irq_disable();
while (1)
- halt();
+ shutdown_halt();
}
#else
static inline void play_dead(void)
Index: linux-2.6.16-rc5/arch/i386/kernel/reboot.c
===================================================================
--- linux-2.6.16-rc5.orig/arch/i386/kernel/reboot.c 2006-03-08 11:34:53.000000000 -0800
+++ linux-2.6.16-rc5/arch/i386/kernel/reboot.c 2006-03-08 11:38:09.000000000 -0800
@@ -146,14 +146,10 @@ real_mode_gdt_entries [3] =
0x000092000100ffffULL /* 16-bit real-mode 64k data at 0x00000100 */
};

-static struct
-{
- unsigned short size __attribute__ ((packed));
- unsigned long long * base __attribute__ ((packed));
-}
-real_mode_gdt = { sizeof (real_mode_gdt_entries) - 1, real_mode_gdt_entries },
-real_mode_idt = { 0x3ff, NULL },
-no_idt = { 0, NULL };
+static struct Xgt_desc_struct
+real_mode_gdt = { sizeof (real_mode_gdt_entries) - 1, (unsigned long)real_mode_gdt_entries },
+real_mode_idt = { 0x3ff, 0 },
+no_idt = { 0, 0 };


/* This is 16-bit protected mode code to disable paging and the cache,
@@ -322,6 +318,9 @@ void machine_shutdown(void)

void machine_emergency_restart(void)
{
+#ifdef CONFIG_X86_VMI
+ vmi_reboot(!reboot_thru_bios);
+#endif
if (!reboot_thru_bios) {
if (efi_enabled) {
efi.reset_system(EFI_RESET_COLD, EFI_SUCCESS, 0, NULL);
@@ -352,6 +351,7 @@ void machine_restart(char * __unused)

void machine_halt(void)
{
+ shutdown_halt();
}

void machine_power_off(void)
Index: linux-2.6.16-rc5/arch/i386/kernel/smp.c
===================================================================
--- linux-2.6.16-rc5.orig/arch/i386/kernel/smp.c 2006-03-08 11:34:53.000000000 -0800
+++ linux-2.6.16-rc5/arch/i386/kernel/smp.c 2006-03-08 11:38:09.000000000 -0800
@@ -575,7 +575,7 @@ static void stop_this_cpu (void * dummy)
local_irq_disable();
disable_local_APIC();
if (cpu_data[smp_processor_id()].hlt_works_ok)
- for(;;) halt();
+ for(;;) shutdown_halt();
for (;;);
}

Index: linux-2.6.16-rc5/arch/i386/mach-vmi/setup.c
===================================================================
--- linux-2.6.16-rc5.orig/arch/i386/mach-vmi/setup.c 2006-03-08 11:37:59.000000000 -0800
+++ linux-2.6.16-rc5/arch/i386/mach-vmi/setup.c 2006-03-08 11:38:32.000000000 -0800
@@ -31,6 +31,7 @@
#include <linux/interrupt.h>
#include <linux/bootmem.h>
#include <linux/mm.h>
+#include <linux/pm.h>
#include <asm/acpi.h>
#include <asm/arch_hooks.h>
#include <asm/processor.h>
@@ -205,7 +206,6 @@ void __init intr_init_hook(void)
setup_irq(2, &irq2);
}

-
/*
* Probe for the VMI option ROM
*/
@@ -238,6 +238,13 @@ void __init probe_vmi_rom(void)
}
}

+/* Simple shutdown routine to power down in a VM */
+void
+vmi_power_off(void)
+{
+ /* Shutdown halt powers off the CPU */
+ shutdown_halt();
+}

/*
* Activate the VMI interfaces
@@ -267,6 +274,9 @@ void __init vmi_init(void)
"(kernel requires version >= %d.%d) "
" - falling back to native mode\n",
VMI_API_REV_MAJOR, MIN_VMI_API_REV_MINOR);
+
+ if (hypervisor_found)
+ pm_power_off = vmi_power_off;
}



2006-03-15 21:35:56

by Eric W. Biederman

[permalink] [raw]
Subject: Re: [RFC, PATCH 14/24] i386 Vmi reboot fixes

Zachary Amsden <[email protected]> writes:

> Fix reboot to work with the VMI. We must support fallback to the standard
> BIOS reboot mechanism. Turns out that this is required by kexec, and a good
> idea for native hardware.

Huh? Rebooting through the BIOS and kexec are pretty much mutually exclusive.
Looking at the patch I can't see what you are talking about either.

Does kexec successfully work under VMWare?


> We simply insert the NOP VMI reboot hook before
> calling the BIOS reboot. While here, fix SMP reboot issues as well. The
> problem is the halt() macro in VMI has been defined to be equivalent to
> safe_halt(), which enables interrupts. Several call sites actually want to
> disable interrupts and shutdown the processor, which is what VMI_Shutdown()
> does.

machine_halt actually is not one of those places.

machine_halt does not want to stop the processor. It is very much
about killing the kernel and user space but having the software still
linger a little.

This needs a cleaner abstraction to make sense or go in.

Eric

2006-03-15 21:35:35

by Eric W. Biederman

[permalink] [raw]
Subject: Re: [RFC, PATCH 14/24] i386 Vmi reboot fixes

Zachary Amsden <[email protected]> writes:

> Fix reboot to work with the VMI. We must support fallback to the standard
> BIOS reboot mechanism. Turns out that this is required by kexec, and a good
> idea for native hardware.

Huh? Rebooting through the BIOS and kexec are pretty much mutually exclusive.
Looking at the patch I can't see what you are talking about either.

Does kexec successfully work under VMWare?


> We simply insert the NOP VMI reboot hook before
> calling the BIOS reboot. While here, fix SMP reboot issues as well. The
> problem is the halt() macro in VMI has been defined to be equivalent to
> safe_halt(), which enables interrupts. Several call sites actually want to
> disable interrupts and shutdown the processor, which is what VMI_Shutdown()
> does.

machine_halt actually is not one of those places.

machine_halt does not want to stop the processor. It is very much
about killing the kernel and user space but having the software still
linger a little.

This needs a cleaner abstraction to make sense or go in.

Eric

2006-03-15 23:25:44

by Zachary Amsden

[permalink] [raw]
Subject: Re: [RFC, PATCH 14/24] i386 Vmi reboot fixes

Eric W. Biederman wrote:
> Zachary Amsden <[email protected]> writes:
> Huh? Rebooting through the BIOS and kexec are pretty much mutually exclusive.
> Looking at the patch I can't see what you are talking about either.
>

Let me rephrase - kexec doesn't define calls for machine_shutdown and
others that are in arch/i386/kernel/reboot.c. So kexec requires BIOS
reboot code to be compiled in, even though the usage of the two is
mutually exclusive.

> Does kexec successfully work under VMWare?
>

It should work just fine. But it could expose bugs on either end. I've
been monitoring our kexec testing, and I'll be able to help you with any
issues that we might find on the Linux side. :)

> machine_halt does not want to stop the processor. It is very much
> about killing the kernel and user space but having the software still
> linger a little.
>

I was afraid of that. I can back that change out. The problem I had
was that the shutdown code I was running in userspace would not make the
syscalls to actually call machine_power_off, but machine_halt instead.
Will fix.

Zach

2006-03-16 05:12:22

by Eric W. Biederman

[permalink] [raw]
Subject: Re: [RFC, PATCH 14/24] i386 Vmi reboot fixes

Zachary Amsden <[email protected]> writes:

> Eric W. Biederman wrote:
>> Zachary Amsden <[email protected]> writes:
>> Huh? Rebooting through the BIOS and kexec are pretty much mutually exclusive.
>> Looking at the patch I can't see what you are talking about either.
>>
>
> Let me rephrase - kexec doesn't define calls for machine_shutdown and others
> that are in arch/i386/kernel/reboot.c. So kexec requires BIOS reboot code to be
> compiled in, even though the usage of the two is mutually exclusive.

Partially true. Basically it has never been optional to compile in the
BIOS reboot code and kexec did not change that situation. Although it
did provide a similar mechanism.

>> Does kexec successfully work under VMWare?
>>
>
> It should work just fine. But it could expose bugs on either end. I've been
> monitoring our kexec testing, and I'll be able to help you with any issues that
> we might find on the Linux side. :)

Ok.

>> machine_halt does not want to stop the processor. It is very much
>> about killing the kernel and user space but having the software still
>> linger a little.
>>
>
> I was afraid of that. I can back that change out. The problem I had was that
> the shutdown code I was running in userspace would not make the syscalls to
> actually call machine_power_off, but machine_halt instead. Will fix.

/sbin/halt -p will call machine_power_off if pm_power_off is defined.
otherwise it will call machine_halt.

Eric