[PATCH v2 0/3] Drivers: hv: add kexec support

[PATCH v2 0/3] Drivers: hv: add kexec support

[Vitaly Kuznetsov]

Changes since v1:
- rebased on top of current Greg's char-misc-next tree.

To make general-purpose kexec (not just kdump) possible for Hyper-V guests
we need to perform some additional cleanup before starting new kernel (see
[PATCH 2/3] for the detailed description).

Know limitations: kexec with balloned out memory is not possible as the
newly loaded kernel doesn't know about this memory and there is no way to
ask the host to bring all the memory back on cleanup (or at least I'm not
aware of such a way). Kexec with hotplugged memory leads to reboot (not
exactly sure why).

Vitaly Kuznetsov (3):
  Drivers: hv: vmbus: remove hv_synic_free_cpu() call from
    hv_synic_cleanup()
  Drivers: hv: vmbus: add special kexec handler
  Drivers: hv: don't do hypercalls when hypercall_page is NULL

 arch/x86/include/asm/mshyperv.h |  2 ++
 arch/x86/kernel/cpu/mshyperv.c  | 30 ++++++++++++++++++++++++++++++
 drivers/hv/hv.c                 | 15 ++++++++-------
 drivers/hv/vmbus_drv.c          | 15 +++++++++++++++
 4 files changed, 55 insertions(+), 7 deletions(-)

-- 
1.9.3

[PATCH v2 1/3] Drivers: hv: vmbus: remove hv_synic_free_cpu() call from hv_synic_cleanup()

[Vitaly Kuznetsov]

We already have hv_synic_free() which frees all per-cpu pages for all
CPUs, let's remove the hv_synic_free_cpu() call from hv_synic_cleanup()
so it will be possible to do separate cleanup (writing to MSRs) and final
freeing. This is going to be used to assist kexec.

Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
---
 drivers/hv/hv.c        | 2 --
 drivers/hv/vmbus_drv.c | 1 +
 2 files changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/hv/hv.c b/drivers/hv/hv.c
index d3943bc..5b87042 100644
--- a/drivers/hv/hv.c
+++ b/drivers/hv/hv.c
@@ -530,6 +530,4 @@ void hv_synic_cleanup(void *arg)
 	rdmsrl(HV_X64_MSR_SCONTROL, sctrl.as_uint64);
 	sctrl.enable = 0;
 	wrmsrl(HV_X64_MSR_SCONTROL, sctrl.as_uint64);
-
-	hv_synic_free_cpu(cpu);
 }
diff --git a/drivers/hv/vmbus_drv.c b/drivers/hv/vmbus_drv.c
index cf20400..00d5158 100644
--- a/drivers/hv/vmbus_drv.c
+++ b/drivers/hv/vmbus_drv.c
@@ -1120,6 +1120,7 @@ static void __exit vmbus_exit(void)
 		tasklet_kill(hv_context.event_dpc[cpu]);
 		smp_call_function_single(cpu, hv_synic_cleanup, NULL, 1);
 	}
+	hv_synic_free();
 	acpi_bus_unregister_driver(&vmbus_acpi_driver);
 	hv_cpu_hotplug_quirk(false);
 }
-- 
1.9.3

[PATCH v2 2/3] Drivers: hv: vmbus: add special kexec handler

[Vitaly Kuznetsov]

When general-purpose kexec (not kdump) is being performed in Hyper-V guest
the newly booted kernel fails with an MCE error coming from the host. It
is the same error which was fixed in the "Drivers: hv: vmbus: Implement
the protocol for tearing down vmbus state" commit - monitor pages remain
special and when they're being written to (as the new kernel doesn't know
these pages are special) bad things happen. We need to perform some
minimalistic cleanup before booting a new kernel on kexec. To do so we
need to register a special machine_ops.shutdown handler to be executed
before the native_machine_shutdown(). Registering a shutdown notification
handler via the register_reboot_notifier() call is not sufficient as it
happens to early for our purposes. machine_ops is not being exported to
modules (and I don't think we want to export it) so let's do this in
mshyperv.c

The minimalistic cleanup consists of cleaning up clockevents, synic MSRs,
guest os id MSR, and hypercall MSR.

Kdump doesn't require all this stuff as it lives in a separate memory
space.

Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
---
 arch/x86/include/asm/mshyperv.h |  2 ++
 arch/x86/kernel/cpu/mshyperv.c  | 30 ++++++++++++++++++++++++++++++
 drivers/hv/vmbus_drv.c          | 14 ++++++++++++++
 3 files changed, 46 insertions(+)

diff --git a/arch/x86/include/asm/mshyperv.h b/arch/x86/include/asm/mshyperv.h
index c163215..d3db910 100644
--- a/arch/x86/include/asm/mshyperv.h
+++ b/arch/x86/include/asm/mshyperv.h
@@ -20,4 +20,6 @@ void hyperv_vector_handler(struct pt_regs *regs);
 void hv_setup_vmbus_irq(void (*handler)(void));
 void hv_remove_vmbus_irq(void);
 
+void hv_setup_kexec_handler(void (*handler)(void));
+void hv_remove_kexec_handler(void);
 #endif
diff --git a/arch/x86/kernel/cpu/mshyperv.c b/arch/x86/kernel/cpu/mshyperv.c
index 939155f..efc20a0 100644
--- a/arch/x86/kernel/cpu/mshyperv.c
+++ b/arch/x86/kernel/cpu/mshyperv.c
@@ -18,6 +18,9 @@
 #include <linux/efi.h>
 #include <linux/interrupt.h>
 #include <linux/irq.h>
+#ifdef CONFIG_KEXEC
+#include <linux/kexec.h>
+#endif
 #include <asm/processor.h>
 #include <asm/hypervisor.h>
 #include <asm/hyperv.h>
@@ -28,12 +31,14 @@
 #include <asm/i8259.h>
 #include <asm/apic.h>
 #include <asm/timer.h>
+#include <asm/reboot.h>
 
 struct ms_hyperv_info ms_hyperv;
 EXPORT_SYMBOL_GPL(ms_hyperv);
 
 #if IS_ENABLED(CONFIG_HYPERV)
 static void (*vmbus_handler)(void);
+static void (*hv_kexec_handler)(void);
 
 void hyperv_vector_handler(struct pt_regs *regs)
 {
@@ -69,6 +74,28 @@ void hv_remove_vmbus_irq(void)
 }
 EXPORT_SYMBOL_GPL(hv_setup_vmbus_irq);
 EXPORT_SYMBOL_GPL(hv_remove_vmbus_irq);
+
+void hv_setup_kexec_handler(void (*handler)(void))
+{
+	hv_kexec_handler = handler;
+}
+EXPORT_SYMBOL_GPL(hv_setup_kexec_handler);
+
+void hv_remove_kexec_handler(void)
+{
+	/* We have no way to deallocate the interrupt gate */
+	hv_kexec_handler = NULL;
+}
+EXPORT_SYMBOL_GPL(hv_remove_kexec_handler);
+
+static void hv_machine_shutdown(void)
+{
+#ifdef CONFIG_KEXEC
+	if (kexec_in_progress && hv_kexec_handler)
+		hv_kexec_handler();
+#endif
+	native_machine_shutdown();
+}
 #endif
 
 static uint32_t  __init ms_hyperv_platform(void)
@@ -143,6 +170,9 @@ static void __init ms_hyperv_init_platform(void)
 	no_timer_check = 1;
 #endif
 
+#if IS_ENABLED(CONFIG_HYPERV)
+	machine_ops.shutdown = hv_machine_shutdown;
+#endif
 }
 
 const __refconst struct hypervisor_x86 x86_hyper_ms_hyperv = {
diff --git a/drivers/hv/vmbus_drv.c b/drivers/hv/vmbus_drv.c
index 00d5158..31748a2 100644
--- a/drivers/hv/vmbus_drv.c
+++ b/drivers/hv/vmbus_drv.c
@@ -1060,6 +1060,17 @@ static struct acpi_driver vmbus_acpi_driver = {
 	},
 };
 
+static void hv_kexec_handler(void)
+{
+	int cpu;
+
+	hv_synic_clockevents_cleanup();
+	vmbus_initiate_unload();
+	for_each_online_cpu(cpu)
+		smp_call_function_single(cpu, hv_synic_cleanup, NULL, 1);
+	hv_cleanup();
+};
+
 static int __init hv_acpi_init(void)
 {
 	int ret, t;
@@ -1092,6 +1103,8 @@ static int __init hv_acpi_init(void)
 	if (ret)
 		goto cleanup;
 
+	hv_setup_kexec_handler(hv_kexec_handler);
+
 	return 0;
 
 cleanup:
@@ -1104,6 +1117,7 @@ static void __exit vmbus_exit(void)
 {
 	int cpu;
 
+	hv_remove_kexec_handler();
 	vmbus_connection.conn_state = DISCONNECTED;
 	hv_synic_clockevents_cleanup();
 	vmbus_disconnect();
-- 
1.9.3

[PATCH v2 3/3] Drivers: hv: don't do hypercalls when hypercall_page is NULL

[Vitaly Kuznetsov]

At the very late stage of kexec a driver (which are not being unloaded) can
try to post a message or signal an event. This will crash the kernel as we
already did hv_cleanup() and the hypercall page is NULL.

Move all common (between 32 and 64 bit code) declarations to the beginning
of the do_hypercall() function. Unfortunately we have to write the
!hypercall_page check twice to not mix declarations and code.

Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
---
 drivers/hv/hv.c | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/drivers/hv/hv.c b/drivers/hv/hv.c
index 5b87042..41d8072 100644
--- a/drivers/hv/hv.c
+++ b/drivers/hv/hv.c
@@ -93,11 +93,14 @@ static int query_hypervisor_info(void)
  */
 static u64 do_hypercall(u64 control, void *input, void *output)
 {
-#ifdef CONFIG_X86_64
-	u64 hv_status = 0;
 	u64 input_address = (input) ? virt_to_phys(input) : 0;
 	u64 output_address = (output) ? virt_to_phys(output) : 0;
 	void *hypercall_page = hv_context.hypercall_page;
+#ifdef CONFIG_X86_64
+	u64 hv_status = 0;
+
+	if (!hypercall_page)
+		return (u64)ULLONG_MAX;
 
 	__asm__ __volatile__("mov %0, %%r8" : : "r" (output_address) : "r8");
 	__asm__ __volatile__("call *%3" : "=a" (hv_status) :
@@ -112,13 +115,13 @@ static u64 do_hypercall(u64 control, void *input, void *output)
 	u32 control_lo = control & 0xFFFFFFFF;
 	u32 hv_status_hi = 1;
 	u32 hv_status_lo = 1;
-	u64 input_address = (input) ? virt_to_phys(input) : 0;
 	u32 input_address_hi = input_address >> 32;
 	u32 input_address_lo = input_address & 0xFFFFFFFF;
-	u64 output_address = (output) ? virt_to_phys(output) : 0;
 	u32 output_address_hi = output_address >> 32;
 	u32 output_address_lo = output_address & 0xFFFFFFFF;
-	void *hypercall_page = hv_context.hypercall_page;
+
+	if (!hypercall_page)
+		return (u64)ULLONG_MAX;
 
 	__asm__ __volatile__ ("call *%8" : "=d"(hv_status_hi),
 			      "=a"(hv_status_lo) : "d" (control_hi),
-- 
1.9.3