The hotplug support for kexec_load() requires changes to the userspace kexec-tools and a little extra help from the kernel. Given a kdump capture kernel loaded via kexec_load(), and a subsequent hotplug event, the crash hotplug handler finds the elfcorehdr and rewrites it to reflect the hotplug change. That is the desired outcome, however, at kernel panic time, the purgatory integrity check fails (because the elfcorehdr changed), and the capture kernel does not boot and no vmcore is generated. Therefore, the userspace kexec-tools/kexec must indicate to the kernel that the elfcorehdr can be modified (because the kexec excluded the elfcorehdr from the digest, and sized the elfcorehdr memory buffer appropriately). To facilitate hotplug support with kexec_load(): - a new kexec flag KEXEC_UPATE_ELFCOREHDR indicates that it is safe for the kernel to modify the kexec_load()'d elfcorehdr - the /sys/kernel/crash_elfcorehdr_size node communicates the preferred size of the elfcorehdr memory buffer - The sysfs crash_hotplug nodes (ie. /sys/devices/system/[cpu|memory]/crash_hotplug) dynamically take into account kexec_file_load() vs kexec_load() and KEXEC_UPDATE_ELFCOREHDR. This is critical so that the udev rule processing of crash_hotplug is all that is needed to determine if the userspace unload-then-load of the kdump image is to be skipped, or not. The proposed udev rule change looks like: # The kernel updates the crash elfcorehdr for CPU and memory changes SUBSYSTEM=="cpu", ATTRS{crash_hotplug}=="1", GOTO="kdump_reload_end" SUBSYSTEM=="memory", ATTRS{crash_hotplug}=="1", GOTO="kdump_reload_end" The table below indicates the behavior of kexec_load()'d kdump image updates (with the new udev crash_hotplug rule in place): Kernel |Kexec -------+-----+---- Old |Old |New | a | a -------+-----+---- New | a | b -------+-----+---- where kexec 'old' and 'new' delineate kexec-tools has the needed modifications for the crash hotplug feature, and kernel 'old' and 'new' delineate the kernel supports this crash hotplug feature. Behavior 'a' indicates the unload-then-reload of the entire kdump image. For the kexec 'old' column, the unload-then-reload occurs due to the missing flag KEXEC_UPDATE_ELFCOREHDR. An 'old' kernel (with 'new' kexec) does not present the crash_hotplug sysfs node, which leads to the unload-then-reload of the kdump image. Behavior 'b' indicates the desired optimized behavior of the kernel directly modifying the elfcorehdr and avoiding the unload-then-reload of the kdump image. If the udev rule is not updated with crash_hotplug node check, then no matter any combination of kernel or kexec is new or old, the kdump image continues to be unload-then-reload on hotplug changes. To fully support crash hotplug feature, there needs to be a rollout of kernel, kexec-tools and udev rule changes. However, the order of the rollout of these pieces does not matter; kexec_load()'d kdump images still function for hotplug as-is.
Unified · Split
| @@ -214,14 +214,17 @@ void arch_crash_handle_hotplug_event(struct kimage *image); · 213 unmodified lines | |||
| 214 | #define arch_crash_handle_hotplug_event arch_crash_handle_hotplug_event | 214 | #define arch_crash_handle_hotplug_event arch_crash_handle_hotplug_event |
| 215 | 215 | ||
| 216 | #ifdef CONFIG_HOTPLUG_CPU | 216 | #ifdef CONFIG_HOTPLUG_CPU |
| 217 | static inline int crash_hotplug_cpu_support(void) { return 1; } | 217 | int arch_crash_hotplug_cpu_support(void); |
| 218 | #define crash_hotplug_cpu_support crash_hotplug_cpu_support | 218 | #define crash_hotplug_cpu_support arch_crash_hotplug_cpu_support |
| 219 | #endif | 219 | #endif |
| 220 | 220 | ||
| 221 | #ifdef CONFIG_MEMORY_HOTPLUG | 221 | #ifdef CONFIG_MEMORY_HOTPLUG |
| 222 | static inline int crash_hotplug_memory_support(void) { return 1; } | 222 | int arch_crash_hotplug_memory_support(void); |
| 223 | #define crash_hotplug_memory_support crash_hotplug_memory_support | 223 | #define crash_hotplug_memory_support arch_crash_hotplug_memory_support |
| 224 | #endif | 224 | #endif |
| 225 | |||
| 226 | unsigned int arch_crash_get_elfcorehdr_size(void); | ||
| 227 | #define crash_get_elfcorehdr_size arch_crash_get_elfcorehdr_size | ||
| 225 | #endif | 228 | #endif |
| 226 | 229 | ||
| 227 | #endif /* __ASSEMBLY__ */ | 230 | #endif /* __ASSEMBLY__ */ |
| @@ -429,6 +429,33 @@ int crash_load_segments(struct kimage *image) · 428 unmodified lines | |||
| 429 | #undef pr_fmt | 429 | #undef pr_fmt |
| 430 | #define pr_fmt(fmt) "crash hp: " fmt | 430 | #define pr_fmt(fmt) "crash hp: " fmt |
| 431 | 431 | ||
| 432 | /* These functions provide the value for the sysfs crash_hotplug nodes */ | ||
| 433 | #ifdef CONFIG_HOTPLUG_CPU | ||
| 434 | int arch_crash_hotplug_cpu_support(void) | ||
| 435 | { | ||
| 436 | return crash_check_update_elfcorehdr(); | ||
| 437 | } | ||
| 438 | #endif | ||
| 439 | |||
| 440 | #ifdef CONFIG_MEMORY_HOTPLUG | ||
| 441 | int arch_crash_hotplug_memory_support(void) | ||
| 442 | { | ||
| 443 | return crash_check_update_elfcorehdr(); | ||
| 444 | } | ||
| 445 | #endif | ||
| 446 | |||
| 447 | unsigned int arch_crash_get_elfcorehdr_size(void) | ||
| 448 | { | ||
| 449 | unsigned int sz; | ||
| 450 | |||
| 451 | /* kernel_map, VMCOREINFO and maximum CPUs */ | ||
| 452 | sz = 2 + CONFIG_NR_CPUS_DEFAULT; | ||
| 453 | if (IS_ENABLED(CONFIG_MEMORY_HOTPLUG)) | ||
| 454 | sz += CONFIG_CRASH_MAX_MEMORY_RANGES; | ||
| 455 | sz *= sizeof(Elf64_Phdr); | ||
| 456 | return sz; | ||
| 457 | } | ||
| 458 | |||
| 432 | /** | 459 | /** |
| 433 | * arch_crash_handle_hotplug_event() - Handle hotplug elfcorehdr changes | 460 | * arch_crash_handle_hotplug_event() - Handle hotplug elfcorehdr changes |
| 434 | * @image: a pointer to kexec_crash_image | 461 | * @image: a pointer to kexec_crash_image |
| @@ -320,6 +320,10 @@ struct kimage { · 319 unmodified lines | |||
| 320 | unsigned int preserve_context : 1; | 320 | unsigned int preserve_context : 1; |
| 321 | /* If set, we are using file mode kexec syscall */ | 321 | /* If set, we are using file mode kexec syscall */ |
| 322 | unsigned int file_mode:1; | 322 | unsigned int file_mode:1; |
| 323 | #ifdef CONFIG_CRASH_HOTPLUG | ||
| 324 | /* If set, allow changes to elfcorehdr of kexec_load'd image */ | ||
| 325 | unsigned int update_elfcorehdr:1; | ||
| 326 | #endif | ||
| 323 | 327 | ||
| 324 | #ifdef ARCH_HAS_KIMAGE_ARCH | 328 | #ifdef ARCH_HAS_KIMAGE_ARCH |
| 325 | struct kimage_arch arch; | 329 | struct kimage_arch arch; |
| @@ -396,9 +400,9 @@ bool kexec_load_permitted(int kexec_image_type); · 70 unmodified lines | |||
| 396 | 400 | ||
| 397 | /* List of defined/legal kexec flags */ | 401 | /* List of defined/legal kexec flags */ |
| 398 | #ifndef CONFIG_KEXEC_JUMP | 402 | #ifndef CONFIG_KEXEC_JUMP |
| 399 | #define KEXEC_FLAGS KEXEC_ON_CRASH | 403 | #define KEXEC_FLAGS (KEXEC_ON_CRASH | KEXEC_UPDATE_ELFCOREHDR) |
| 400 | #else | 404 | #else |
| 401 | #define KEXEC_FLAGS (KEXEC_ON_CRASH | KEXEC_PRESERVE_CONTEXT) | 405 | #define KEXEC_FLAGS (KEXEC_ON_CRASH | KEXEC_PRESERVE_CONTEXT | KEXEC_UPDATE_ELFCOREHDR) |
| 402 | #endif | 406 | #endif |
| 403 | 407 | ||
| 404 | /* List of defined/legal kexec file flags */ | 408 | /* List of defined/legal kexec file flags */ |
| @@ -486,6 +490,8 @@ static inline void arch_kexec_pre_free_pages(void *vaddr, unsigned int pages) { · 81 unmodified lines | |||
| 486 | static inline void arch_crash_handle_hotplug_event(struct kimage *image) { } | 490 | static inline void arch_crash_handle_hotplug_event(struct kimage *image) { } |
| 487 | #endif | 491 | #endif |
| 488 | 492 | ||
| 493 | int crash_check_update_elfcorehdr(void); | ||
| 494 | |||
| 489 | #ifndef crash_hotplug_cpu_support | 495 | #ifndef crash_hotplug_cpu_support |
| 490 | static inline int crash_hotplug_cpu_support(void) { return 0; } | 496 | static inline int crash_hotplug_cpu_support(void) { return 0; } |
| 491 | #endif | 497 | #endif |
| @@ -494,6 +500,10 @@ static inline int crash_hotplug_cpu_support(void) { return 0; } · 2 unmodified lines | |||
| 494 | static inline int crash_hotplug_memory_support(void) { return 0; } | 500 | static inline int crash_hotplug_memory_support(void) { return 0; } |
| 495 | #endif | 501 | #endif |
| 496 | 502 | ||
| 503 | #ifndef crash_get_elfcorehdr_size | ||
| 504 | static inline unsigned int crash_get_elfcorehdr_size(void) { return 0; } | ||
| 505 | #endif | ||
| 506 | |||
| 497 | #else /* !CONFIG_KEXEC_CORE */ | 507 | #else /* !CONFIG_KEXEC_CORE */ |
| 498 | struct pt_regs; | 508 | struct pt_regs; |
| 499 | struct task_struct; | 509 | struct task_struct; |
| @@ -12,6 +12,7 @@ · 11 unmodified lines | |||
| 12 | /* kexec flags for different usage scenarios */ | 12 | /* kexec flags for different usage scenarios */ |
| 13 | #define KEXEC_ON_CRASH 0x00000001 | 13 | #define KEXEC_ON_CRASH 0x00000001 |
| 14 | #define KEXEC_PRESERVE_CONTEXT 0x00000002 | 14 | #define KEXEC_PRESERVE_CONTEXT 0x00000002 |
| 15 | #define KEXEC_UPDATE_ELFCOREHDR 0x00000004 | ||
| 15 | #define KEXEC_ARCH_MASK 0xffff0000 | 16 | #define KEXEC_ARCH_MASK 0xffff0000 |
| 16 | 17 | ||
| 17 | /* | 18 | /* |
| @@ -143,4 +143,8 @@ config CRASH_MAX_MEMORY_RANGES · 142 unmodified lines | |||
| 143 | memory buffer/segment size under 1MiB. This represents a sane choice | 143 | memory buffer/segment size under 1MiB. This represents a sane choice |
| 144 | to accommodate both baremetal and virtual machine configurations. | 144 | to accommodate both baremetal and virtual machine configurations. |
| 145 | 145 | ||
| 146 | For the kexec_load() syscall path, CRASH_MAX_MEMORY_RANGES is part of | ||
| 147 | the computation behind the value provided through the | ||
| 148 | /sys/kernel/crash_elfcorehdr_size attribute. | ||
| 149 | |||
| 146 | endmenu | 150 | endmenu |
| @@ -740,6 +740,33 @@ subsys_initcall(crash_notes_memory_init); · 739 unmodified lines | |||
| 740 | #ifdef CONFIG_CRASH_HOTPLUG | 740 | #ifdef CONFIG_CRASH_HOTPLUG |
| 741 | #undef pr_fmt | 741 | #undef pr_fmt |
| 742 | #define pr_fmt(fmt) "crash hp: " fmt | 742 | #define pr_fmt(fmt) "crash hp: " fmt |
| 743 | |||
| 744 | /* | ||
| 745 | * This routine utilized when the crash_hotplug sysfs node is read. | ||
| 746 | * It reflects the kernel's ability/permission to update the crash | ||
| 747 | * elfcorehdr directly. | ||
| 748 | */ | ||
| 749 | int crash_check_update_elfcorehdr(void) | ||
| 750 | { | ||
| 751 | int rc = 0; | ||
| 752 | |||
| 753 | /* Obtain lock while reading crash information */ | ||
| 754 | if (!kexec_trylock()) { | ||
| 755 | pr_info("kexec_trylock() failed, elfcorehdr may be inaccurate\n"); | ||
| 756 | return 0; | ||
| 757 | } | ||
| 758 | if (kexec_crash_image) { | ||
| 759 | if (kexec_crash_image->file_mode) | ||
| 760 | rc = 1; | ||
| 761 | else | ||
| 762 | rc = kexec_crash_image->update_elfcorehdr; | ||
| 763 | } | ||
| 764 | /* Release lock now that update complete */ | ||
| 765 | kexec_unlock(); | ||
| 766 | |||
| 767 | return rc; | ||
| 768 | } | ||
| 769 | |||
| 743 | /* | 770 | /* |
| 744 | * To accurately reflect hot un/plug changes of cpu and memory resources | 771 | * To accurately reflect hot un/plug changes of cpu and memory resources |
| 745 | * (including onling and offlining of those resources), the elfcorehdr | 772 | * (including onling and offlining of those resources), the elfcorehdr |
| @@ -770,6 +797,10 @@ static void crash_handle_hotplug_event(unsigned int hp_action, unsigned int cpu) · 24 unmodified lines | |||
| 770 | 797 | ||
| 771 | image = kexec_crash_image; | 798 | image = kexec_crash_image; |
| 772 | 799 | ||
| 800 | /* Check that updating elfcorehdr is permitted */ | ||
| 801 | if (!(image->file_mode || image->update_elfcorehdr)) | ||
| 802 | goto out; | ||
| 803 | |||
| 773 | if (hp_action == KEXEC_CRASH_HP_ADD_CPU || | 804 | if (hp_action == KEXEC_CRASH_HP_ADD_CPU || |
| 774 | hp_action == KEXEC_CRASH_HP_REMOVE_CPU) | 805 | hp_action == KEXEC_CRASH_HP_REMOVE_CPU) |
| 775 | pr_debug("hp_action %u, cpu %u\n", hp_action, cpu); | 806 | pr_debug("hp_action %u, cpu %u\n", hp_action, cpu); |
| @@ -129,6 +129,11 @@ static int do_kexec_load(unsigned long entry, unsigned long nr_segments, · 128 unmodified lines | |||
| 129 | if (flags & KEXEC_PRESERVE_CONTEXT) | 129 | if (flags & KEXEC_PRESERVE_CONTEXT) |
| 130 | image->preserve_context = 1; | 130 | image->preserve_context = 1; |
| 131 | 131 | ||
| 132 | #ifdef CONFIG_CRASH_HOTPLUG | ||
| 133 | if (flags & KEXEC_UPDATE_ELFCOREHDR) | ||
| 134 | image->update_elfcorehdr = 1; | ||
| 135 | #endif | ||
| 136 | |||
| 132 | ret = machine_kexec_prepare(image); | 137 | ret = machine_kexec_prepare(image); |
| 133 | if (ret) | 138 | if (ret) |
| 134 | goto out; | 139 | goto out; |
| @@ -165,6 +165,18 @@ static ssize_t vmcoreinfo_show(struct kobject *kobj, · 164 unmodified lines | |||
| 165 | } | 165 | } |
| 166 | KERNEL_ATTR_RO(vmcoreinfo); | 166 | KERNEL_ATTR_RO(vmcoreinfo); |
| 167 | 167 | ||
| 168 | #ifdef CONFIG_CRASH_HOTPLUG | ||
| 169 | static ssize_t crash_elfcorehdr_size_show(struct kobject *kobj, | ||
| 170 | struct kobj_attribute *attr, char *buf) | ||
| 171 | { | ||
| 172 | unsigned int sz = crash_get_elfcorehdr_size(); | ||
| 173 | |||
| 174 | return sysfs_emit(buf, "%u\n", sz); | ||
| 175 | } | ||
| 176 | KERNEL_ATTR_RO(crash_elfcorehdr_size); | ||
| 177 | |||
| 178 | #endif | ||
| 179 | |||
| 168 | #endif /* CONFIG_CRASH_CORE */ | 180 | #endif /* CONFIG_CRASH_CORE */ |
| 169 | 181 | ||
| 170 | /* whether file capabilities are enabled */ | 182 | /* whether file capabilities are enabled */ |
| @@ -255,6 +267,9 @@ static struct attribute * kernel_attrs[] = { · 84 unmodified lines | |||
| 255 | #endif | 267 | #endif |
| 256 | #ifdef CONFIG_CRASH_CORE | 268 | #ifdef CONFIG_CRASH_CORE |
| 257 | &vmcoreinfo_attr.attr, | 269 | &vmcoreinfo_attr.attr, |
| 270 | #ifdef CONFIG_CRASH_HOTPLUG | ||
| 271 | &crash_elfcorehdr_size_attr.attr, | ||
| 272 | #endif | ||
| 258 | #endif | 273 | #endif |
| 259 | #ifndef CONFIG_TINY_RCU | 274 | #ifndef CONFIG_TINY_RCU |
| 260 | &rcu_expedited_attr.attr, | 275 | &rcu_expedited_attr.attr, |