> -----Original Message-----
> From: Sunil Muthuswamy
> Sent: Friday, June 8, 2018 11:39 AM
> To: Haiyang Zhang <haiya...@microsoft.com>
> Cc: de...@linuxdriverproject.org; Sunil Muthuswamy
> <sunil...@microsoft.com>; KY Srinivasan <k...@microsoft.com>; Stephen
> Hemminger <sthem...@microsoft.com>
> Subject: [PATCH v5] Drivers: HV: Send one page worth of kmsg dump over
> Hyper-V during panic
>
> In the VM mode on Hyper-V, currently, when the kernel panics, an error
> code and few register values are populated in an MSR and the Hypervisor
> notified. This information is collected on the host. The amount of
> information currently collected is found to be limited and not very
> actionable. To gather more actionable data, such as stack trace, the
> proposal is to write one page worth of kmsg data on an allocated page
> and the Hypervisor notified of the page address through the MSR.
>
> - Sysctl option to control the behavior, with ON by default.
>
> Cc: K. Y. Srinivasan <k...@microsoft.com>
> Cc: Stephen Hemminger <sthem...@microsoft.com>
> Signed-off-by: Sunil Muthuswamy <sunil...@microsoft.com>
> ---
> Changes since v1:
> - Added a sysctl config option to allow the user to control if panic message
> should be reported to Hyper-V.
>
> Changes since v2:
> - Addressed comments from Greg KH.
> - Moved the sysctl config option from 'kernel.hyperv' to 'kernel.' since the
> likelihood of having many Hyper-V specific sysctl config options seemed less
> and so having a specific dir for Hyper-V seemed unnecessary overhead.
>
> Changes since v3:
> - The feature is only enabled if the capability is supported by the
> hypervisor.
> This is done by reading the crash control MSR, in addition to the
> pre-existing
> check for the hypervisor feature using the CPUID.
>
> Changes since v4:
> - Addressed comments from Dexuan Cui: moved the global variable to static.
> ---
> Documentation/sysctl/kernel.txt | 11 ++++
> arch/x86/hyperv/hv_init.c | 27 +++++++++
> arch/x86/include/asm/hyperv-tlfs.h | 5 +-
> arch/x86/include/asm/mshyperv.h | 1 +
> drivers/hv/vmbus_drv.c | 116
> +++++++++++++++++++++++++++++++++++++
> 5 files changed, 158 insertions(+), 2 deletions(-)
>
> diff --git a/Documentation/sysctl/kernel.txt
> b/Documentation/sysctl/kernel.txt
> index eded671d..5958503 100644
> --- a/Documentation/sysctl/kernel.txt
> +++ b/Documentation/sysctl/kernel.txt
> @@ -39,6 +39,7 @@ show up in /proc/sys/kernel:
> - hung_task_check_count
> - hung_task_timeout_secs
> - hung_task_warnings
> +- hyperv_record_panic_msg
> - kexec_load_disabled
> - kptr_restrict
> - l2cr [ PPC only ]
> @@ -374,6 +375,16 @@ This file shows up if CONFIG_DETECT_HUNG_TASK is
> enabled.
>
>
> ==========================================================
> ====
>
> +hyperv_record_panic_msg:
> +
> +Controls whether the panic kmsg data should be reported to Hyper-V.
> +
> +0: do not report panic kmsg data.
> +
> +1: report the panic kmsg data. This is the default behavior.
> +
> +=========================================================
> =====
> +
> kexec_load_disabled:
>
> A toggle indicating if the kexec_load syscall has been disabled. This
> diff --git a/arch/x86/hyperv/hv_init.c b/arch/x86/hyperv/hv_init.c
> index cfecc22..af57162 100644
> --- a/arch/x86/hyperv/hv_init.c
> +++ b/arch/x86/hyperv/hv_init.c
> @@ -395,6 +395,33 @@ void hyperv_report_panic(struct pt_regs *regs, long
> err)
> }
> EXPORT_SYMBOL_GPL(hyperv_report_panic);
>
> +/**
> + * hyperv_report_panic_msg - report panic message to Hyper-V
> + * @pa: physical address of the panic page containing the message
> + * @size: size of the message in the page
> + */
> +void hyperv_report_panic_msg(phys_addr_t pa, size_t size)
> +{
> + /*
> + * P3 to contain the physical address of the panic page & P4 to
> + * contain the size of the panic data in that page. Rest of the
> + * registers are no-op when the NOTIFY_MSG flag is set.
> + */
> + wrmsrl(HV_X64_MSR_CRASH_P0, 0);
> + wrmsrl(HV_X64_MSR_CRASH_P1, 0);
> + wrmsrl(HV_X64_MSR_CRASH_P2, 0);
> + wrmsrl(HV_X64_MSR_CRASH_P3, pa);
> + wrmsrl(HV_X64_MSR_CRASH_P4, size);
> +
> + /*
> + * Let Hyper-V know there is crash data available along with
> + * the panic message.
> + */
> + wrmsrl(HV_X64_MSR_CRASH_CTL,
> + (HV_CRASH_CTL_CRASH_NOTIFY |
> HV_CRASH_CTL_CRASH_NOTIFY_MSG));
> +}
> +EXPORT_SYMBOL_GPL(hyperv_report_panic_msg);
> +
> bool hv_is_hyperv_initialized(void)
> {
> union hv_x64_msr_hypercall_contents hypercall_msr;
> diff --git a/arch/x86/include/asm/hyperv-tlfs.h
> b/arch/x86/include/asm/hyperv-tlfs.h
> index 416cb0e..76c58af 100644
> --- a/arch/x86/include/asm/hyperv-tlfs.h
> +++ b/arch/x86/include/asm/hyperv-tlfs.h
> @@ -171,9 +171,10 @@
> #define HV_X64_ENLIGHTENED_VMCS_RECOMMENDED (1 << 14)
>
> /*
> - * Crash notification flag.
> + * Crash notification flags.
> */
> -#define HV_CRASH_CTL_CRASH_NOTIFY (1ULL << 63)
> +#define HV_CRASH_CTL_CRASH_NOTIFY_MSG BIT_ULL(62)
> +#define HV_CRASH_CTL_CRASH_NOTIFY BIT_ULL(63)
>
> /* MSR used to identify the guest OS. */
> #define HV_X64_MSR_GUEST_OS_ID 0x40000000
> diff --git a/arch/x86/include/asm/mshyperv.h
> b/arch/x86/include/asm/mshyperv.h
> index b90e796..ac83f2d 100644
> --- a/arch/x86/include/asm/mshyperv.h
> +++ b/arch/x86/include/asm/mshyperv.h
> @@ -262,6 +262,7 @@ void hyperv_init(void);
> void hyperv_setup_mmu_ops(void);
> void hyper_alloc_mmu(void);
> void hyperv_report_panic(struct pt_regs *regs, long err);
> +void hyperv_report_panic_msg(phys_addr_t pa, size_t size);
> bool hv_is_hyperv_initialized(void);
> void hyperv_cleanup(void);
>
> diff --git a/drivers/hv/vmbus_drv.c b/drivers/hv/vmbus_drv.c
> index b10fe26..eec3b235 100644
> --- a/drivers/hv/vmbus_drv.c
> +++ b/drivers/hv/vmbus_drv.c
> @@ -56,6 +56,8 @@ static struct completion probe_event;
>
> static int hyperv_cpuhp_online;
>
> +static void *hv_panic_page;
> +
> static int hyperv_panic_event(struct notifier_block *nb, unsigned long val,
> void *args)
> {
> @@ -1018,6 +1020,75 @@ static void vmbus_isr(void)
> add_interrupt_randomness(HYPERVISOR_CALLBACK_VECTOR, 0);
> }
>
> +/*
> + * Boolean to control whether to report panic messages over Hyper-V.
> + *
> + * It can be set via /proc/sys/kernel/hyperv/record_panic_msg
> + */
> +static int sysctl_record_panic_msg = 1;
> +
> +/*
> + * Callback from kmsg_dump. Grab as much as possible from the end of the
> kmsg
> + * buffer and call into Hyper-V to transfer the data.
> + */
> +static void hv_kmsg_dump(struct kmsg_dumper *dumper,
> + enum kmsg_dump_reason reason)
> +{
> + size_t bytes_written;
> + phys_addr_t panic_pa;
> +
> + /* We are only interested in panics. */
> + if ((reason != KMSG_DUMP_PANIC) || (!sysctl_record_panic_msg))
> + return;
> +
> + panic_pa = virt_to_phys(hv_panic_page);
> +
> + /*
> + * Write dump contents to the page. No need to synchronize; panic
> should
> + * be single-threaded.
> + */
> + if (!kmsg_dump_get_buffer(dumper, true, hv_panic_page,
> + PAGE_SIZE, &bytes_written)) {
> + pr_err("Hyper-V: Unable to get kmsg data for panic\n");
> + return;
> + }
> +
> + hyperv_report_panic_msg(panic_pa, bytes_written);
> +}
> +
> +static struct kmsg_dumper hv_kmsg_dumper = {
> + .dump = hv_kmsg_dump,
> +};
> +
> +static struct ctl_table_header *hv_ctl_table_hdr;
> +static int zero;
> +static int one = 1;
> +
> +/*
> + * sysctl option to allow the user to control whether kmsg data should be
> + * reported to Hyper-V on panic.
> + */
> +static struct ctl_table hv_ctl_table[] = {
> + {
> + .procname = "hyperv_record_panic_msg",
> + .data = &sysctl_record_panic_msg,
> + .maxlen = sizeof(int),
> + .mode = 0644,
> + .proc_handler = proc_dointvec_minmax,
> + .extra1 = &zero,
> + .extra2 = &one
> + },
> + {}
> +};
> +
> +static struct ctl_table hv_root_table[] = {
> + {
> + .procname = "kernel",
> + .mode = 0555,
> + .child = hv_ctl_table
> + },
> + {}
> +};
>
> /*
> * vmbus_bus_init -Main vmbus driver initialization routine.
> @@ -1065,6 +1136,38 @@ static int vmbus_bus_init(void)
> * Only register if the crash MSRs are available
> */
> if (ms_hyperv.misc_features &
> HV_FEATURE_GUEST_CRASH_MSR_AVAILABLE) {
> + u64 hyperv_crash_ctl;
> + if (!hv_ctl_table_hdr) {
When vmus_bus_init is invoked, would hv_ctl_table_hdr ever be
Non-NULL?
> + /*
> + * Sysctl registration is not fatal, since default is
> + * for operation.
> + */
> + hv_ctl_table_hdr =
> + register_sysctl_table(hv_root_table);
> + if (!hv_ctl_table_hdr)
> + pr_err("Hyper-V: sysctl table register error");
> + }
> +
> + /*
> + * Register for panic kmsg callback only if the right
> + * capability is supported by the hypervisor.
> + */
> + rdmsrl(HV_X64_MSR_CRASH_CTL, hyperv_crash_ctl);
> + if (hyperv_crash_ctl &
> HV_CRASH_CTL_CRASH_NOTIFY_MSG) {
> + hv_panic_page = (void
> *)get_zeroed_page(GFP_KERNEL);
> + if (!hv_panic_page) {
Why is this allocation failure fatal in terms failing the load of the driver?
If we failed to allocate why can't we revert to the existing panic notification
scheme.
> + ret = -ENOMEM;
> + goto err_connect;
> + }
> +
> + ret = kmsg_dump_register(&hv_kmsg_dumper);
> + if (ret) {
> + pr_err("Hyper-V: kmsg dump register error "
> + "0x%x\n", ret);
> + goto err_connect;
> + }
> + }
> +
> register_die_notifier(&hyperv_die_block);
> atomic_notifier_chain_register(&panic_notifier_list,
> &hyperv_panic_block);
> @@ -1081,6 +1184,11 @@ static int vmbus_bus_init(void)
> hv_remove_vmbus_irq();
>
> bus_unregister(&hv_bus);
> + free_page((unsigned long)hv_panic_page);
> + if (!hv_ctl_table_hdr) {
> + unregister_sysctl_table(hv_ctl_table_hdr);
> + hv_ctl_table_hdr = NULL;
> + }
>
> return ret;
> }
> @@ -1785,10 +1893,18 @@ static void __exit vmbus_exit(void)
> vmbus_free_channels();
>
> if (ms_hyperv.misc_features &
> HV_FEATURE_GUEST_CRASH_MSR_AVAILABLE) {
> + kmsg_dump_unregister(&hv_kmsg_dumper);
> unregister_die_notifier(&hyperv_die_block);
> atomic_notifier_chain_unregister(&panic_notifier_list,
> &hyperv_panic_block);
> }
> +
> + free_page((unsigned long)hv_panic_page);
> + if (!hv_ctl_table_hdr) {
> + unregister_sysctl_table(hv_ctl_table_hdr);
> + hv_ctl_table_hdr = NULL;
> + }
> +
> bus_unregister(&hv_bus);
>
> cpuhp_remove_state(hyperv_cpuhp_online);
> --
> 2.7.4
_______________________________________________
devel mailing list
de...@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel