[PATCH v3 1/3] kexec_load: Use new kexec flag for hotplug support

[PATCH v3 1/3] kexec_load: Use new kexec flag for hotplug support

[Sourabh Jain]

Kernel commit 79365026f869 (crash: add a new kexec flag for hotplug
support) has introduced a new kexec flag to generalize hotplug support.
The newly introduced kexec flags for hotplug allow architectures to
exclude all the required kexec segments from SHA calculation so that
the kernel can update them on hotplug events. This was not possible
earlier with the KEXEC_UPDATE_ELFCOREHDR kexec flags since it was added
only for the elfcorehdr segment.

To enable architectures to control the list of kexec segments to exclude
when hotplug support is enabled, add a new architecture-specific
function named arch_do_exclude_segment. During the SHA calculation, this
function gets called to let the architecture decide whether a specific
kexec segment should be considered for SHA calculation or not.

Given that the KEXEC_UPDATE_ELFCOREHDR is no longer required and was
colliding with the KEXEC_LIVE_UPDATE update flag, it is removed.

Cc: Aditya Gupta <adityag@linux.ibm.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Coiby Xu <coxu@redhat.com>
Cc: Mahesh Salgaonkar <mahesh@linux.ibm.com>
Cc: Simon Horman <horms@kernel.org>
Acked-by: Hari Bathini <hbathini@linux.ibm.com>
Signed-off-by: Sourabh Jain <sourabhjain@linux.ibm.com>
---

* No changes in v2.

---
 kexec/arch/arm/kexec-arm.c             |  5 +++++
 kexec/arch/arm64/kexec-arm64.c         |  4 ++++
 kexec/arch/cris/kexec-cris.c           |  4 ++++
 kexec/arch/hppa/kexec-hppa.c           |  5 +++++
 kexec/arch/i386/kexec-x86.c            |  8 ++++++++
 kexec/arch/ia64/kexec-ia64.c           |  4 ++++
 kexec/arch/loongarch/kexec-loongarch.c |  5 +++++
 kexec/arch/m68k/kexec-m68k.c           |  5 +++++
 kexec/arch/mips/kexec-mips.c           |  4 ++++
 kexec/arch/ppc/kexec-ppc.c             |  4 ++++
 kexec/arch/ppc64/kexec-ppc64.c         |  5 +++++
 kexec/arch/s390/kexec-s390.c           |  5 +++++
 kexec/arch/sh/kexec-sh.c               |  5 +++++
 kexec/arch/x86_64/kexec-x86_64.c       |  5 +++++
 kexec/kexec-syscall.h                  |  2 +-
 kexec/kexec.c                          | 14 ++++++++------
 kexec/kexec.h                          |  2 ++
 17 files changed, 79 insertions(+), 7 deletions(-)

diff --git a/kexec/arch/arm/kexec-arm.c b/kexec/arch/arm/kexec-arm.c
index 49f35b1..34531f9 100644
--- a/kexec/arch/arm/kexec-arm.c
+++ b/kexec/arch/arm/kexec-arm.c
@@ -148,3 +148,8 @@ int have_sysfs_fdt(void)
 {
 	return !access(SYSFS_FDT, F_OK);
 }
+
+int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
+{
+	return 0;
+}
diff --git a/kexec/arch/arm64/kexec-arm64.c b/kexec/arch/arm64/kexec-arm64.c
index 4a67b0d..9d052b0 100644
--- a/kexec/arch/arm64/kexec-arm64.c
+++ b/kexec/arch/arm64/kexec-arm64.c
@@ -1363,3 +1363,7 @@ void arch_reuse_initrd(void)
 void arch_update_purgatory(struct kexec_info *UNUSED(info))
 {
 }
+int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
+{
+	return 0;
+}
diff --git a/kexec/arch/cris/kexec-cris.c b/kexec/arch/cris/kexec-cris.c
index 3b69709..7f09121 100644
--- a/kexec/arch/cris/kexec-cris.c
+++ b/kexec/arch/cris/kexec-cris.c
@@ -109,3 +109,7 @@ unsigned long add_buffer(struct kexec_info *info, const void *buf,
                                     buf_min, buf_max, buf_end, 1);
 }
 
+int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
+{
+	return 0;
+}
diff --git a/kexec/arch/hppa/kexec-hppa.c b/kexec/arch/hppa/kexec-hppa.c
index 77c9739..a64dc3d 100644
--- a/kexec/arch/hppa/kexec-hppa.c
+++ b/kexec/arch/hppa/kexec-hppa.c
@@ -146,3 +146,8 @@ unsigned long virt_to_phys(unsigned long addr)
 {
 	return addr - phys_offset;
 }
+
+int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
+{
+	return 0;
+}
diff --git a/kexec/arch/i386/kexec-x86.c b/kexec/arch/i386/kexec-x86.c
index 444cb69..b4947a0 100644
--- a/kexec/arch/i386/kexec-x86.c
+++ b/kexec/arch/i386/kexec-x86.c
@@ -208,3 +208,11 @@ void arch_update_purgatory(struct kexec_info *info)
 	elf_rel_set_symbol(&info->rhdr, "panic_kernel",
 		&panic_kernel, sizeof(panic_kernel));
 }
+
+int arch_do_exclude_segment(struct kexec_segment *seg_ptr, struct kexec_info *info)
+{
+	if (info->elfcorehdr == (unsigned long) seg_ptr->mem)
+		return 1;
+
+	return 0;
+}
diff --git a/kexec/arch/ia64/kexec-ia64.c b/kexec/arch/ia64/kexec-ia64.c
index 418d997..8d9c1f3 100644
--- a/kexec/arch/ia64/kexec-ia64.c
+++ b/kexec/arch/ia64/kexec-ia64.c
@@ -245,3 +245,7 @@ void arch_update_purgatory(struct kexec_info *UNUSED(info))
 {
 }
 
+int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
+{
+	return 0;
+}
diff --git a/kexec/arch/loongarch/kexec-loongarch.c b/kexec/arch/loongarch/kexec-loongarch.c
index ac75030..ee7b9f1 100644
--- a/kexec/arch/loongarch/kexec-loongarch.c
+++ b/kexec/arch/loongarch/kexec-loongarch.c
@@ -381,3 +381,8 @@ unsigned long add_buffer(struct kexec_info *info, const void *buf,
 	return add_buffer_phys_virt(info, buf, bufsz, memsz, buf_align,
 				    buf_min, buf_max, buf_end, 1);
 }
+
+int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
+{
+	return 0;
+}
diff --git a/kexec/arch/m68k/kexec-m68k.c b/kexec/arch/m68k/kexec-m68k.c
index cb54927..0c7dbaf 100644
--- a/kexec/arch/m68k/kexec-m68k.c
+++ b/kexec/arch/m68k/kexec-m68k.c
@@ -108,3 +108,8 @@ void add_segment(struct kexec_info *info, const void *buf, size_t bufsz,
 {
 	add_segment_phys_virt(info, buf, bufsz, base, memsz, 1);
 }
+
+int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
+{
+	return 0;
+}
diff --git a/kexec/arch/mips/kexec-mips.c b/kexec/arch/mips/kexec-mips.c
index d8cbea8..94224ee 100644
--- a/kexec/arch/mips/kexec-mips.c
+++ b/kexec/arch/mips/kexec-mips.c
@@ -189,3 +189,7 @@ unsigned long add_buffer(struct kexec_info *info, const void *buf,
 				    buf_min, buf_max, buf_end, 1);
 }
 
+int arch_do_exclude_segment(const void *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
+{
+	return 0;
+}
diff --git a/kexec/arch/ppc/kexec-ppc.c b/kexec/arch/ppc/kexec-ppc.c
index 03bec36..c8af870 100644
--- a/kexec/arch/ppc/kexec-ppc.c
+++ b/kexec/arch/ppc/kexec-ppc.c
@@ -966,3 +966,7 @@ void arch_update_purgatory(struct kexec_info *UNUSED(info))
 {
 }
 
+int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
+{
+	return 0;
+}
diff --git a/kexec/arch/ppc64/kexec-ppc64.c b/kexec/arch/ppc64/kexec-ppc64.c
index bd5274c..fb27b6b 100644
--- a/kexec/arch/ppc64/kexec-ppc64.c
+++ b/kexec/arch/ppc64/kexec-ppc64.c
@@ -967,3 +967,8 @@ int arch_compat_trampoline(struct kexec_info *UNUSED(info))
 void arch_update_purgatory(struct kexec_info *UNUSED(info))
 {
 }
+
+int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
+{
+	return 0;
+}
diff --git a/kexec/arch/s390/kexec-s390.c b/kexec/arch/s390/kexec-s390.c
index 33ba6b9..0561ee7 100644
--- a/kexec/arch/s390/kexec-s390.c
+++ b/kexec/arch/s390/kexec-s390.c
@@ -267,3 +267,8 @@ int get_crash_kernel_load_range(uint64_t *start, uint64_t *end)
 {
 	return parse_iomem_single("Crash kernel\n", start, end);
 }
+
+int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
+{
+	return 0;
+}
diff --git a/kexec/arch/sh/kexec-sh.c b/kexec/arch/sh/kexec-sh.c
index ce341c8..f84c40c 100644
--- a/kexec/arch/sh/kexec-sh.c
+++ b/kexec/arch/sh/kexec-sh.c
@@ -257,3 +257,8 @@ unsigned long add_buffer(struct kexec_info *info, const void *buf,
 	return add_buffer_phys_virt(info, buf, bufsz, memsz, buf_align,
 				    buf_min, buf_max, buf_end, 1);
 }
+
+int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
+{
+	return 0;
+}
diff --git a/kexec/arch/x86_64/kexec-x86_64.c b/kexec/arch/x86_64/kexec-x86_64.c
index ffd84f0..42af90a 100644
--- a/kexec/arch/x86_64/kexec-x86_64.c
+++ b/kexec/arch/x86_64/kexec-x86_64.c
@@ -188,3 +188,8 @@ void arch_update_purgatory(struct kexec_info *info)
 	elf_rel_set_symbol(&info->rhdr, "panic_kernel",
 				&panic_kernel, sizeof(panic_kernel));
 }
+
+int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
+{
+	return 0;
+}
diff --git a/kexec/kexec-syscall.h b/kexec/kexec-syscall.h
index 73e5254..4675c46 100644
--- a/kexec/kexec-syscall.h
+++ b/kexec/kexec-syscall.h
@@ -112,7 +112,7 @@ static inline long kexec_file_load(int kernel_fd, int initrd_fd,
 
 #define KEXEC_ON_CRASH		0x00000001
 #define KEXEC_PRESERVE_CONTEXT	0x00000002
-#define KEXEC_UPDATE_ELFCOREHDR	0x00000004
+#define KEXEC_CRASH_HOTPLUG_SUPPORT	0x00000008
 #define KEXEC_ARCH_MASK		0xffff0000
 
 /* Flags for kexec file based system call */
diff --git a/kexec/kexec.c b/kexec/kexec.c
index 222f79e..034cea6 100644
--- a/kexec/kexec.c
+++ b/kexec/kexec.c
@@ -701,10 +701,13 @@ static void update_purgatory(struct kexec_info *info)
 			continue;
 		}
 
-		/* Don't include elfcorehdr in the checksum, if hotplug
-		 * support enabled.
+		/*
+		 * Let architecture decide which segments to exclude from checksum
+		 * if hotplug support is enabled.
 		 */
-		if (do_hotplug && (info->segment[i].mem == (void *)info->elfcorehdr)) {
+		if (do_hotplug && arch_do_exclude_segment(&info->segment[i], info)) {
+			dbgprintf("Skipping segment mem: 0x%lx from SHA calculation\n",
+				  (unsigned long)info->segment[i].mem);
 			continue;
 		}
 
@@ -1651,7 +1654,6 @@ int main(int argc, char *argv[])
 		die("--load-live-update can only be used with xen\n");
 	}
 
-	/* NOTE: Xen KEXEC_LIVE_UPDATE and KEXEC_UPDATE_ELFCOREHDR collide */
 	if (do_hotplug) {
 		const char *ces = "/sys/kernel/crash_elfcorehdr_size";
 		char *buf, *endptr = NULL;
@@ -1665,8 +1667,8 @@ int main(int argc, char *argv[])
 		if (!elfcorehdrsz || (endptr && *endptr != '\0'))
 			die("Path %s does not exist, the kernel needs CONFIG_CRASH_HOTPLUG\n", ces);
 		dbgprintf("ELFCOREHDR_SIZE %lu\n", elfcorehdrsz);
-		/* Indicate to the kernel it is ok to modify the elfcorehdr */
-		kexec_flags |= KEXEC_UPDATE_ELFCOREHDR;
+		/* Indicate to the kernel it is ok to modify the relevant kexec segments */
+		kexec_flags |= KEXEC_CRASH_HOTPLUG_SUPPORT;
 	}
 
 	fileind = optind;
diff --git a/kexec/kexec.h b/kexec/kexec.h
index 1004aff..2d758c9 100644
--- a/kexec/kexec.h
+++ b/kexec/kexec.h
@@ -307,6 +307,8 @@ extern int do_hotplug;
 #define BOOTLOADER_VERSION PACKAGE_VERSION
 
 void arch_usage(void);
+/* Return non-zero if segment needs to be excluded from SHA calculation, else 0. */
+int arch_do_exclude_segment(struct kexec_segment *seg_ptr, struct kexec_info *info);
 int arch_process_options(int argc, char **argv);
 int arch_compat_trampoline(struct kexec_info *info);
 void arch_update_purgatory(struct kexec_info *info);
-- 
2.45.1


_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec

[PATCH v3 2/3] powerpc/kexec_load: add hotplug support

[Sourabh Jain]

Kernel commits b741092d5976 ("powerpc/crash: add crash CPU hotplug
support") and 849599b702ef ("powerpc/crash: add crash memory hotplug
support") added crash CPU/Memory hotplug support on PowerPC. This patch
extends that support for the kexec_load syscall.

During CPU/Memory hotplug events on PowerPC, two kexec segments,
elfcorehdr, and FDT, get updated by the kernel. To ensure the kernel
can safely update these two kexec segments for the kdump image loaded
using the kexec_load system call, the following changes are made:

1. Extra size is allocated for both elfcorehdr and FDT to accommodate
   additional resources in the future. For the elfcorehdr, the size hint
   is taken from /sys/kernel/crash_elfcorehdr_size sysfs, while for FDT,
   extra size is allocated to hold possible CPU nodes.

2. Both elfcorehdr and FDT are skipped from SHA calculation.

Cc: Aditya Gupta <adityag@linux.ibm.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Coiby Xu <coxu@redhat.com>
Cc: Mahesh Salgaonkar <mahesh@linux.ibm.com>
Cc: Simon Horman <horms@kernel.org>
Acked-by: Hari Bathini <hbathini@linux.ibm.com>
Signed-off-by: Sourabh Jain <sourabhjain@linux.ibm.com>
---

Changelog:

From v2 -> v3:
  - Added a Acked-by, no functional changes.

From v1 -> v2:
  - Find CPUs in the system using the /sys/devices/system/cpu/present sysfs
    instead of traversing all nodes under /proc/device-tree/cpus.

  - Added a new function to find present CPUs in the system.

  - Removed unnecessary NULL check on seg_ptr from arch_do_exclude_segment().

---
 kexec/arch/ppc64/crashdump-ppc64.c  |  16 +-
 kexec/arch/ppc64/fdt.c              | 236 +++++++++++++++++++++++++++-
 kexec/arch/ppc64/include/arch/fdt.h |   2 +-
 kexec/arch/ppc64/kexec-elf-ppc64.c  |   2 +-
 kexec/arch/ppc64/kexec-ppc64.c      |   9 +-
 5 files changed, 258 insertions(+), 7 deletions(-)

diff --git a/kexec/arch/ppc64/crashdump-ppc64.c b/kexec/arch/ppc64/crashdump-ppc64.c
index 6d47898..98d439a 100644
--- a/kexec/arch/ppc64/crashdump-ppc64.c
+++ b/kexec/arch/ppc64/crashdump-ppc64.c
@@ -476,7 +476,7 @@ int load_crashdump_segments(struct kexec_info *info, char* mod_cmdline,
 				uint64_t max_addr, unsigned long min_base)
 {
 	void *tmp;
-	unsigned long sz;
+	unsigned long sz, memsz;
 	uint64_t elfcorehdr;
 	int nr_ranges, align = 1024, i;
 	unsigned long long end;
@@ -531,8 +531,18 @@ int load_crashdump_segments(struct kexec_info *info, char* mod_cmdline,
 		}
 	}
 
-	elfcorehdr = add_buffer(info, tmp, sz, sz, align, min_base,
-				max_addr, 1);
+	memsz = sz;
+	/* To support --hotplug, replace the calculated memsz with the value
+	 * from /sys/kernel/crash_elfcorehdr_size and align it correctly.
+	 */
+	if (do_hotplug) {
+		if (elfcorehdrsz > sz)
+			memsz = _ALIGN(elfcorehdrsz, align);
+	}
+
+	/* Record the location of the elfcorehdr for hotplug handling */
+	info->elfcorehdr = elfcorehdr = add_buffer(info, tmp, sz, memsz, align,
+						   min_base, max_addr, 1);
 	reserve(elfcorehdr, sz);
 	/* modify and store the cmdline in a global array. This is later
 	 * read by flatten_device_tree and modified if required
diff --git a/kexec/arch/ppc64/fdt.c b/kexec/arch/ppc64/fdt.c
index 8bc6d2d..879240f 100644
--- a/kexec/arch/ppc64/fdt.c
+++ b/kexec/arch/ppc64/fdt.c
@@ -17,6 +17,13 @@
 #include <libfdt.h>
 #include <stdio.h>
 #include <stdlib.h>
+#include <limits.h>
+#include <stdbool.h>
+#include <dirent.h>
+#include <sys/stat.h>
+
+#include "../../kexec.h"
+#include "../../kexec-syscall.h"
 
 /*
  * Let the kernel know it booted from kexec, as some things (e.g.
@@ -46,17 +53,244 @@ static int fixup_kexec_prop(void *fdt)
 	return 0;
 }
 
+static inline bool is_dot_dir(char * d_path)
+{
+	return d_path[0] == '.';
+}
+
+/*
+ * get_cpu_node_size - Returns size of files including file name size under
+ *                     the given @cpu_node_path.
+ */
+static int get_cpu_node_size(char *cpu_node_path)
+{
+	DIR *d;
+	struct dirent *de;
+	struct stat statbuf;
+	int cpu_node_size = 0;
+	char cpu_prop_path[2 * PATH_MAX];
+
+	d = opendir(cpu_node_path);
+	if (!d)
+		return 0;
+
+	while ((de = readdir(d)) != NULL) {
+		if (de->d_type != DT_REG)
+			continue;
+
+		memset(cpu_prop_path, '\0', PATH_MAX);
+		snprintf(cpu_prop_path, 2 * PATH_MAX, "%s/%s", cpu_node_path,
+			 de->d_name);
+
+		if (stat(cpu_prop_path, &statbuf))
+			continue;
+
+		cpu_node_size += statbuf.st_size;
+		cpu_node_size += strlen(de->d_name);
+	}
+
+	return cpu_node_size;
+}
+
+/*
+ * is_cpu_node - Checks if the node specified by the given @path
+ *               represents a CPU node.
+ *
+ * Returns true if the @path has a "device_type" file containing "cpu";
+ * otherwise, returns false.
+ */
+static bool is_cpu_node(char *path)
+{
+	FILE *file;
+	bool ret = false;
+	char device_type[4];
+
+	file = fopen(path, "r");
+	if (!file)
+		return false;
+
+	memset(device_type, '\0', 4);
+	if (fread(device_type, 1, 3, file) < 3)
+		goto out;
+
+	if (strcmp(device_type, "cpu"))
+		goto out;
+
+	ret = true;
+out:
+	fclose(file);
+	return ret;
+}
+
+static int get_threads_per_cpu(char *path)
+{
+	struct stat statbuf;
+	if (stat(path, &statbuf))
+		return 0;
+
+	return statbuf.st_size / 4;
+}
+
+/**
+ * get_present_cpus - finds the present CPUs in the system
+ *
+ * This function opens the file `/sys/devices/system/cpu/present` to read
+ * the range of present CPUs. It parses the range and calculates the
+ * total number of present CPUs in the system.
+ *
+ * Returns total number of present CPUs on success, -1 on failure.
+ */
+static int get_present_cpus()
+{
+	char *range;
+	char buf[1024];
+	int start, end;
+	int cpu_count = 0;
+	FILE *file = fopen("/sys/devices/system/cpu/present", "r");
+
+	if (!file)
+		return -1;
+
+	if (!fgets(buf, sizeof(buf), file))
+		return -1;
+
+	fclose(file);
+
+	range = strtok(buf, ",");
+	while (range != NULL) {
+		if (sscanf(range, "%d-%d", &start, &end) == 2) {
+			for (int i = start; i <= end; i++)
+				cpu_count++;
+		} else if (sscanf(range, "%d", &start) == 1) {
+			cpu_count++;
+		} else {
+			return -1;
+		}
+		range = strtok(NULL, ",");
+	}
+
+	return cpu_count;
+}
+
+/*
+ * get_cpu_info - Finds the following CPU attributes:
+ *
+ * threads_per_cpu: Number of threads per CPU, based on the device tree entry
+ *                  /proc/device-tree/cpus/<cpu_node>/ibm,ppc-interrupt-server#s.
+ * cpu_node_size: Size of files including file name size under a CPU node.
+ *
+ * Returns 0 on success, else -1.
+ */
+static int get_cpu_info(int *_present_cpus, int *_threads_per_cpu, int *_cpu_node_size)
+{
+	DIR *d;
+	struct dirent *de;
+	char path[PATH_MAX];
+	int present_cpus = 0, threads_per_cpu = 0, cpu_node_size = 0;
+	char *cpus_node_path = "/proc/device-tree/cpus";
+
+	present_cpus = get_present_cpus();
+	if (present_cpus < 0)
+		return -1;
+
+	d = opendir(cpus_node_path);
+	if (!d)
+		return -1;
+
+	while ((de = readdir(d)) != NULL) {
+		if ((de->d_type != DT_DIR) || is_dot_dir(de->d_name))
+			continue;
+
+		memset(path, '\0', PATH_MAX);
+		snprintf(path, PATH_MAX, "%s/%s/%s", cpus_node_path,
+			 de->d_name, "device_type");
+
+		/* Skip nodes with device_type != "cpu" */
+		if (!is_cpu_node(path))
+			continue;
+
+		/*
+		 * Found the first node under /proc/device-tree/cpus with
+		 * device_type == "cpu"
+		 */
+		memset(path, '\0', PATH_MAX);
+		snprintf(path, PATH_MAX, "%s/%s", cpus_node_path, de->d_name);
+		cpu_node_size = get_cpu_node_size(path);
+
+		memset(path, '\0', PATH_MAX);
+		snprintf(path, PATH_MAX, "%s/%s/%s", cpus_node_path,
+		de->d_name, "ibm,ppc-interrupt-server#s");
+		threads_per_cpu = get_threads_per_cpu(path);
+		break;
+	}
+
+	closedir(d);
+
+	if (!(threads_per_cpu && cpu_node_size))
+		return -1;
+
+	*_present_cpus = present_cpus;
+	*_cpu_node_size = cpu_node_size;
+	*_threads_per_cpu = threads_per_cpu;
+
+	dbgprintf("present_cpus: %d, threads_per_cpu: %d, cpu_node_size: %d\n",
+		  present_cpus, threads_per_cpu, cpu_node_size);
+
+	return 0;
+}
+
+/*
+ * kdump_fdt_extra_size - Calculates the extra size needed for the Flattened
+ *                        Device Tree (FDT) based on the possible and present
+ *                        CPUs in the system.
+ */
+static unsigned int kdump_fdt_extra_size(void)
+{
+	int cpus_in_system;
+	unsigned int extra_size = 0;
+	int present_cpus = 0, threads_per_cpu = 0, cpu_node_size = 0;
+	int possible_cpus;
+
+	/* ALL possible CPUs are present in FDT so no extra size required */
+	if (sysconf(_SC_NPROCESSORS_ONLN) == sysconf(_SC_NPROCESSORS_CONF))
+		return 0;
+
+	if (get_cpu_info(&present_cpus, &threads_per_cpu, &cpu_node_size)) {
+		die("Failed to get cpu info\n");
+	}
+
+	cpus_in_system = present_cpus / threads_per_cpu;
+	possible_cpus = sysconf(_SC_NPROCESSORS_CONF) / threads_per_cpu;
+	dbgprintf("cpus_in_system: %d, possible_cpus: %d\n", cpus_in_system,
+		  possible_cpus);
+
+	if (cpus_in_system > possible_cpus)
+		die("Possible CPU nodes can't be less than active CPU nodes\n");
+
+	extra_size = (possible_cpus - cpus_in_system) * cpu_node_size;
+	dbgprintf("kdump fdt extra size: %u\n", extra_size);
+
+	return extra_size;
+}
 
 /*
  * For now, assume that the added content fits in the file.
  * This should be the case when flattening from /proc/device-tree,
  * and when passing in a dtb, dtc can be told to add padding.
  */
-int fixup_dt(char **fdt, off_t *size)
+int fixup_dt(char **fdt, off_t *size, unsigned long kexec_flags)
 {
 	int ret;
 
 	*size += 4096;
+
+	/* To support --hotplug option for the kexec_load syscall, consider
+	 * adding extra buffer to FDT so that the kernel can add CPU nodes
+	 * of hot-added CPUs.
+	 */
+	if (do_hotplug && (kexec_flags & KEXEC_ON_CRASH))
+		*size += kdump_fdt_extra_size();
+
 	*fdt = realloc(*fdt, *size);
 	if (!*fdt) {
 		fprintf(stderr, "%s: out of memory\n", __func__);
diff --git a/kexec/arch/ppc64/include/arch/fdt.h b/kexec/arch/ppc64/include/arch/fdt.h
index b19f185..5f340b0 100644
--- a/kexec/arch/ppc64/include/arch/fdt.h
+++ b/kexec/arch/ppc64/include/arch/fdt.h
@@ -3,6 +3,6 @@
 
 #include <sys/types.h>
 
-int fixup_dt(char **fdt, off_t *size);
+int fixup_dt(char **fdt, off_t *size, unsigned long kexec_flags);
 
 #endif
diff --git a/kexec/arch/ppc64/kexec-elf-ppc64.c b/kexec/arch/ppc64/kexec-elf-ppc64.c
index bdcfd20..858c994 100644
--- a/kexec/arch/ppc64/kexec-elf-ppc64.c
+++ b/kexec/arch/ppc64/kexec-elf-ppc64.c
@@ -345,7 +345,7 @@ int elf_ppc64_load(int argc, char **argv, const char *buf, off_t len,
 		create_flatten_tree(&seg_buf, &seg_size, cmdline);
 	}
 
-	result = fixup_dt(&seg_buf, &seg_size);
+	result = fixup_dt(&seg_buf, &seg_size, info->kexec_flags);
 	if (result < 0)
 		return result;
 
diff --git a/kexec/arch/ppc64/kexec-ppc64.c b/kexec/arch/ppc64/kexec-ppc64.c
index fb27b6b..13c3ce3 100644
--- a/kexec/arch/ppc64/kexec-ppc64.c
+++ b/kexec/arch/ppc64/kexec-ppc64.c
@@ -24,6 +24,7 @@
 #include <errno.h>
 #include <stdint.h>
 #include <string.h>
+#include <libfdt.h>
 #include <sys/stat.h>
 #include <sys/types.h>
 #include <dirent.h>
@@ -968,7 +969,13 @@ void arch_update_purgatory(struct kexec_info *UNUSED(info))
 {
 }
 
-int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
+int arch_do_exclude_segment(struct kexec_segment *seg_ptr, struct kexec_info *info)
 {
+	if (info->elfcorehdr == (unsigned long) seg_ptr->mem)
+		return 1;
+
+	if (seg_ptr->buf && fdt_magic(seg_ptr->buf) == FDT_MAGIC)
+		return 1;
+
 	return 0;
 }
-- 
2.45.1


_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec

[PATCH v3 3/3] doc/hotplug: update man and --help

[Sourabh Jain]

Update the man page and --help option to make the description of the
--hotplug option easier to understand.

Cc: Aditya Gupta <adityag@linux.ibm.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Coiby Xu <coxu@redhat.com>
Cc: Mahesh Salgaonkar <mahesh@linux.ibm.com>
Cc: Simon Horman <horms@kernel.org>
Acked-by: Hari Bathini <hbathini@linux.ibm.com>
Signed-off-by: Sourabh Jain <sourabhjain@linux.ibm.com>
---
Changelog:

From v2 -> v3
  - Updated --hotplug option description

---
 kexec/kexec.8 | 8 ++++----
 kexec/kexec.c | 4 +++-
 2 files changed, 7 insertions(+), 5 deletions(-)

diff --git a/kexec/kexec.8 b/kexec/kexec.8
index 9e995fe..793e876 100644
--- a/kexec/kexec.8
+++ b/kexec/kexec.8
@@ -140,10 +140,10 @@ Open a help file for
 .BR kexec .
 .TP
 .B \-\-hotplug
-Setup for kernel modification of the elfcorehdr. This option performs
-the steps needed to support kernel updates to the elfcorehdr in the
-presence of hot un/plug and/or on/offline events. This option only
-useful for KEXEC_LOAD syscall.
+Setup kexec segments such that kernel can safely update them on CPU/Memory
+hot add/remove events. If this option is enabled, kernel does in-kernel
+update of kexec segments on CPU/Memory hot add/remove events, thus avoiding
+the need to reload kdump kernel.
 .TP
 .B \-i\ (\-\-no-checks)
 Fast reboot, no memory integrity checks.
diff --git a/kexec/kexec.c b/kexec/kexec.c
index 034cea6..9b7c34c 100644
--- a/kexec/kexec.c
+++ b/kexec/kexec.c
@@ -1093,7 +1093,9 @@ void usage(void)
 	       "                      back to the compatibility syscall when file based\n"
 	       "                      syscall is not supported or the kernel did not\n"
 	       "                      understand the image (default)\n"
-	       " --hotplug            Setup for kernel modification of elfcorehdr.\n"
+	       " --hotplug            Do in-kernel update of kexec segments on CPU/Memory\n"
+	       "                      hot add/remove events, avoiding the need to reload\n"
+	       "                      kdump kernel on online/offline events.\n"
 	       " -d, --debug          Enable debugging to help spot a failure.\n"
 	       " -S, --status         Return 1 if the type (by default crash) is loaded,\n"
 	       "                      0 if not.\n"
-- 
2.45.1


_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec

Re: [PATCH v3 1/3] kexec_load: Use new kexec flag for hotplug support

[Baoquan He]

On 07/02/24 at 10:00am, Sourabh Jain wrote:
......
> diff --git a/kexec/arch/i386/kexec-x86.c b/kexec/arch/i386/kexec-x86.c
> index 444cb69..b4947a0 100644
> --- a/kexec/arch/i386/kexec-x86.c
> +++ b/kexec/arch/i386/kexec-x86.c
> @@ -208,3 +208,11 @@ void arch_update_purgatory(struct kexec_info *info)
>  	elf_rel_set_symbol(&info->rhdr, "panic_kernel",
>  		&panic_kernel, sizeof(panic_kernel));
>  }
> +
> +int arch_do_exclude_segment(struct kexec_segment *seg_ptr, struct kexec_info *info)
> +{
> +	if (info->elfcorehdr == (unsigned long) seg_ptr->mem)
> +		return 1;
> +
> +	return 0;
> +}

I know the similar question has been asked in earlier version, I may not
get it. still raise concern here to ask why x86_64 returns 0 directly,
while i386 will have the different cases.

> diff --git a/kexec/arch/ia64/kexec-ia64.c b/kexec/arch/ia64/kexec-ia64.c
> index 418d997..8d9c1f3 100644
> --- a/kexec/arch/ia64/kexec-ia64.c
> +++ b/kexec/arch/ia64/kexec-ia64.c
> @@ -245,3 +245,7 @@ void arch_update_purgatory(struct kexec_info *UNUSED(info))
>  {
>  }
>  
> +int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
                                                            ~~~~~~~
A tiny nit about the parameter naming, since the existing is taking
'struct kexec_segment *segment', can we also take the same way? Not
strong opinion.

=====
kexec-tools/kexec/kexec.c:
static int valid_memory_segment(struct kexec_info *info,
                                struct kexec_segment *segment)
{
        unsigned long sstart, send;
        sstart = (unsigned long)segment->mem;
        send   = sstart + segment->memsz - 1;

        return valid_memory_range(info, sstart, send);
}

> +{
> +	return 0;
> +}
> diff --git a/kexec/arch/loongarch/kexec-loongarch.c b/kexec/arch/loongarch/kexec-loongarch.c
> index ac75030..ee7b9f1 100644
> --- a/kexec/arch/loongarch/kexec-loongarch.c
> +++ b/kexec/arch/loongarch/kexec-loongarch.c
> @@ -381,3 +381,8 @@ unsigned long add_buffer(struct kexec_info *info, const void *buf,
>  	return add_buffer_phys_virt(info, buf, bufsz, memsz, buf_align,
>  				    buf_min, buf_max, buf_end, 1);
>  }
> +
> +int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
> +{
> +	return 0;
> +}
> diff --git a/kexec/arch/m68k/kexec-m68k.c b/kexec/arch/m68k/kexec-m68k.c
> index cb54927..0c7dbaf 100644
> --- a/kexec/arch/m68k/kexec-m68k.c
> +++ b/kexec/arch/m68k/kexec-m68k.c
> @@ -108,3 +108,8 @@ void add_segment(struct kexec_info *info, const void *buf, size_t bufsz,
>  {
>  	add_segment_phys_virt(info, buf, bufsz, base, memsz, 1);
>  }
> +
> +int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
> +{
> +	return 0;
> +}
> diff --git a/kexec/arch/mips/kexec-mips.c b/kexec/arch/mips/kexec-mips.c
> index d8cbea8..94224ee 100644
> --- a/kexec/arch/mips/kexec-mips.c
> +++ b/kexec/arch/mips/kexec-mips.c
> @@ -189,3 +189,7 @@ unsigned long add_buffer(struct kexec_info *info, const void *buf,
>  				    buf_min, buf_max, buf_end, 1);
>  }
>  
> +int arch_do_exclude_segment(const void *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
> +{
> +	return 0;
> +}
> diff --git a/kexec/arch/ppc/kexec-ppc.c b/kexec/arch/ppc/kexec-ppc.c
> index 03bec36..c8af870 100644
> --- a/kexec/arch/ppc/kexec-ppc.c
> +++ b/kexec/arch/ppc/kexec-ppc.c
> @@ -966,3 +966,7 @@ void arch_update_purgatory(struct kexec_info *UNUSED(info))
>  {
>  }
>  
> +int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
> +{
> +	return 0;
> +}
> diff --git a/kexec/arch/ppc64/kexec-ppc64.c b/kexec/arch/ppc64/kexec-ppc64.c
> index bd5274c..fb27b6b 100644
> --- a/kexec/arch/ppc64/kexec-ppc64.c
> +++ b/kexec/arch/ppc64/kexec-ppc64.c
> @@ -967,3 +967,8 @@ int arch_compat_trampoline(struct kexec_info *UNUSED(info))
>  void arch_update_purgatory(struct kexec_info *UNUSED(info))
>  {
>  }
> +
> +int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
> +{
> +	return 0;
> +}
> diff --git a/kexec/arch/s390/kexec-s390.c b/kexec/arch/s390/kexec-s390.c
> index 33ba6b9..0561ee7 100644
> --- a/kexec/arch/s390/kexec-s390.c
> +++ b/kexec/arch/s390/kexec-s390.c
> @@ -267,3 +267,8 @@ int get_crash_kernel_load_range(uint64_t *start, uint64_t *end)
>  {
>  	return parse_iomem_single("Crash kernel\n", start, end);
>  }
> +
> +int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
> +{
> +	return 0;
> +}
> diff --git a/kexec/arch/sh/kexec-sh.c b/kexec/arch/sh/kexec-sh.c
> index ce341c8..f84c40c 100644
> --- a/kexec/arch/sh/kexec-sh.c
> +++ b/kexec/arch/sh/kexec-sh.c
> @@ -257,3 +257,8 @@ unsigned long add_buffer(struct kexec_info *info, const void *buf,
>  	return add_buffer_phys_virt(info, buf, bufsz, memsz, buf_align,
>  				    buf_min, buf_max, buf_end, 1);
>  }
> +
> +int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
> +{
> +	return 0;
> +}
> diff --git a/kexec/arch/x86_64/kexec-x86_64.c b/kexec/arch/x86_64/kexec-x86_64.c
> index ffd84f0..42af90a 100644
> --- a/kexec/arch/x86_64/kexec-x86_64.c
> +++ b/kexec/arch/x86_64/kexec-x86_64.c
> @@ -188,3 +188,8 @@ void arch_update_purgatory(struct kexec_info *info)
>  	elf_rel_set_symbol(&info->rhdr, "panic_kernel",
>  				&panic_kernel, sizeof(panic_kernel));
>  }
> +
> +int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
> +{
> +	return 0;
> +}
> diff --git a/kexec/kexec-syscall.h b/kexec/kexec-syscall.h
> index 73e5254..4675c46 100644
> --- a/kexec/kexec-syscall.h
> +++ b/kexec/kexec-syscall.h
> @@ -112,7 +112,7 @@ static inline long kexec_file_load(int kernel_fd, int initrd_fd,
>  
>  #define KEXEC_ON_CRASH		0x00000001
>  #define KEXEC_PRESERVE_CONTEXT	0x00000002
> -#define KEXEC_UPDATE_ELFCOREHDR	0x00000004
> +#define KEXEC_CRASH_HOTPLUG_SUPPORT	0x00000008
>  #define KEXEC_ARCH_MASK		0xffff0000

Yeah, removing KEXEC_UPDATE_ELFCOREHDR sounds good to avoid collision
with xen code. 

>  
>  /* Flags for kexec file based system call */
> diff --git a/kexec/kexec.c b/kexec/kexec.c
> index 222f79e..034cea6 100644
> --- a/kexec/kexec.c
> +++ b/kexec/kexec.c
> @@ -701,10 +701,13 @@ static void update_purgatory(struct kexec_info *info)
>  			continue;
>  		}
>  
> -		/* Don't include elfcorehdr in the checksum, if hotplug
> -		 * support enabled.
> +		/*
> +		 * Let architecture decide which segments to exclude from checksum
> +		 * if hotplug support is enabled.
>  		 */
> -		if (do_hotplug && (info->segment[i].mem == (void *)info->elfcorehdr)) {
> +		if (do_hotplug && arch_do_exclude_segment(&info->segment[i], info)) {
> +			dbgprintf("Skipping segment mem: 0x%lx from SHA calculation\n",
> +				  (unsigned long)info->segment[i].mem);
>  			continue;
>  		}
>  
> @@ -1651,7 +1654,6 @@ int main(int argc, char *argv[])
>  		die("--load-live-update can only be used with xen\n");
>  	}
>  
> -	/* NOTE: Xen KEXEC_LIVE_UPDATE and KEXEC_UPDATE_ELFCOREHDR collide */
>  	if (do_hotplug) {
>  		const char *ces = "/sys/kernel/crash_elfcorehdr_size";
>  		char *buf, *endptr = NULL;
> @@ -1665,8 +1667,8 @@ int main(int argc, char *argv[])
>  		if (!elfcorehdrsz || (endptr && *endptr != '\0'))
>  			die("Path %s does not exist, the kernel needs CONFIG_CRASH_HOTPLUG\n", ces);
>  		dbgprintf("ELFCOREHDR_SIZE %lu\n", elfcorehdrsz);
> -		/* Indicate to the kernel it is ok to modify the elfcorehdr */
> -		kexec_flags |= KEXEC_UPDATE_ELFCOREHDR;
> +		/* Indicate to the kernel it is ok to modify the relevant kexec segments */
> +		kexec_flags |= KEXEC_CRASH_HOTPLUG_SUPPORT;
>  	}
>  
>  	fileind = optind;
> diff --git a/kexec/kexec.h b/kexec/kexec.h
> index 1004aff..2d758c9 100644
> --- a/kexec/kexec.h
> +++ b/kexec/kexec.h
> @@ -307,6 +307,8 @@ extern int do_hotplug;
>  #define BOOTLOADER_VERSION PACKAGE_VERSION
>  
>  void arch_usage(void);
> +/* Return non-zero if segment needs to be excluded from SHA calculation, else 0. */
> +int arch_do_exclude_segment(struct kexec_segment *seg_ptr, struct kexec_info *info);
>  int arch_process_options(int argc, char **argv);
>  int arch_compat_trampoline(struct kexec_info *info);
>  void arch_update_purgatory(struct kexec_info *info);
> -- 
> 2.45.1
> 


_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec

Re: [PATCH v3 3/3] doc/hotplug: update man and --help

[Baoquan He]

On 07/02/24 at 10:00am, Sourabh Jain wrote:
> Update the man page and --help option to make the description of the
> --hotplug option easier to understand.
> 
> Cc: Aditya Gupta <adityag@linux.ibm.com>
> Cc: Baoquan He <bhe@redhat.com>
> Cc: Coiby Xu <coxu@redhat.com>
> Cc: Mahesh Salgaonkar <mahesh@linux.ibm.com>
> Cc: Simon Horman <horms@kernel.org>
> Acked-by: Hari Bathini <hbathini@linux.ibm.com>
> Signed-off-by: Sourabh Jain <sourabhjain@linux.ibm.com>
> ---
> Changelog:
> 
> From v2 -> v3
>   - Updated --hotplug option description
> 
> ---
>  kexec/kexec.8 | 8 ++++----
>  kexec/kexec.c | 4 +++-
>  2 files changed, 7 insertions(+), 5 deletions(-)

Acked-by: Baoquan He <bhe@redhat.com>

> 
> diff --git a/kexec/kexec.8 b/kexec/kexec.8
> index 9e995fe..793e876 100644
> --- a/kexec/kexec.8
> +++ b/kexec/kexec.8
> @@ -140,10 +140,10 @@ Open a help file for
>  .BR kexec .
>  .TP
>  .B \-\-hotplug
> -Setup for kernel modification of the elfcorehdr. This option performs
> -the steps needed to support kernel updates to the elfcorehdr in the
> -presence of hot un/plug and/or on/offline events. This option only
> -useful for KEXEC_LOAD syscall.
> +Setup kexec segments such that kernel can safely update them on CPU/Memory
> +hot add/remove events. If this option is enabled, kernel does in-kernel
> +update of kexec segments on CPU/Memory hot add/remove events, thus avoiding
> +the need to reload kdump kernel.
>  .TP
>  .B \-i\ (\-\-no-checks)
>  Fast reboot, no memory integrity checks.
> diff --git a/kexec/kexec.c b/kexec/kexec.c
> index 034cea6..9b7c34c 100644
> --- a/kexec/kexec.c
> +++ b/kexec/kexec.c
> @@ -1093,7 +1093,9 @@ void usage(void)
>  	       "                      back to the compatibility syscall when file based\n"
>  	       "                      syscall is not supported or the kernel did not\n"
>  	       "                      understand the image (default)\n"
> -	       " --hotplug            Setup for kernel modification of elfcorehdr.\n"
> +	       " --hotplug            Do in-kernel update of kexec segments on CPU/Memory\n"
> +	       "                      hot add/remove events, avoiding the need to reload\n"
> +	       "                      kdump kernel on online/offline events.\n"
>  	       " -d, --debug          Enable debugging to help spot a failure.\n"
>  	       " -S, --status         Return 1 if the type (by default crash) is loaded,\n"
>  	       "                      0 if not.\n"
> -- 
> 2.45.1
> 


_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec

Re: [PATCH v3 1/3] kexec_load: Use new kexec flag for hotplug support

[Sourabh Jain]

Hello Baoquan,

On 03/07/24 10:01, Baoquan He wrote:
> On 07/02/24 at 10:00am, Sourabh Jain wrote:
> ......
>> diff --git a/kexec/arch/i386/kexec-x86.c b/kexec/arch/i386/kexec-x86.c
>> index 444cb69..b4947a0 100644
>> --- a/kexec/arch/i386/kexec-x86.c
>> +++ b/kexec/arch/i386/kexec-x86.c
>> @@ -208,3 +208,11 @@ void arch_update_purgatory(struct kexec_info *info)
>>   	elf_rel_set_symbol(&info->rhdr, "panic_kernel",
>>   		&panic_kernel, sizeof(panic_kernel));
>>   }
>> +
>> +int arch_do_exclude_segment(struct kexec_segment *seg_ptr, struct kexec_info *info)
>> +{
>> +	if (info->elfcorehdr == (unsigned long) seg_ptr->mem)
>> +		return 1;
>> +
>> +	return 0;
>> +}
> I know the similar question has been asked in earlier version, I may not
> get it. still raise concern here to ask why x86_64 returns 0 directly,
> while i386 will have the different cases.

Currently, info->elfcorehdr is only getting initialized in 
load_crashdump_segments() in
kexec/arch/i386/crashdump-x86.c. This led me think that elfcorehdr is 
only skipped
for i386 and NOT x86_64.

However, info->elfcorehdr is actually getting initialized for x86_64 
too. Because
load_crashdump_segments() defined under kexec/arch/i386/crashdump-x86.c 
is called for
both i386 and x86_64.

You and Hari are both right; the arch_do_exclude_segment() 
implementation should be
the same for both i386 and x86_64.

I will update the patch.

>
>> diff --git a/kexec/arch/ia64/kexec-ia64.c b/kexec/arch/ia64/kexec-ia64.c
>> index 418d997..8d9c1f3 100644
>> --- a/kexec/arch/ia64/kexec-ia64.c
>> +++ b/kexec/arch/ia64/kexec-ia64.c
>> @@ -245,3 +245,7 @@ void arch_update_purgatory(struct kexec_info *UNUSED(info))
>>   {
>>   }
>>   
>> +int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
>                                                              ~~~~~~~
> A tiny nit about the parameter naming, since the existing is taking
> 'struct kexec_segment *segment', can we also take the same way? Not
> strong opinion.
>
> =====
> kexec-tools/kexec/kexec.c:
> static int valid_memory_segment(struct kexec_info *info,
>                                  struct kexec_segment *segment)
> {
>          unsigned long sstart, send;
>          sstart = (unsigned long)segment->mem;
>          send   = sstart + segment->memsz - 1;
>
>          return valid_memory_range(info, sstart, send);
> }

Sure, I will rename the parameter and reorder them just like in the 
above function.

>
>> +{
>> +	return 0;
>> +}
>> diff --git a/kexec/arch/loongarch/kexec-loongarch.c b/kexec/arch/loongarch/kexec-loongarch.c
>> index ac75030..ee7b9f1 100644
>> --- a/kexec/arch/loongarch/kexec-loongarch.c
>> +++ b/kexec/arch/loongarch/kexec-loongarch.c
>> @@ -381,3 +381,8 @@ unsigned long add_buffer(struct kexec_info *info, const void *buf,
>>   	return add_buffer_phys_virt(info, buf, bufsz, memsz, buf_align,
>>   				    buf_min, buf_max, buf_end, 1);
>>   }
>> +
>> +int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
>> +{
>> +	return 0;
>> +}
>> diff --git a/kexec/arch/m68k/kexec-m68k.c b/kexec/arch/m68k/kexec-m68k.c
>> index cb54927..0c7dbaf 100644
>> --- a/kexec/arch/m68k/kexec-m68k.c
>> +++ b/kexec/arch/m68k/kexec-m68k.c
>> @@ -108,3 +108,8 @@ void add_segment(struct kexec_info *info, const void *buf, size_t bufsz,
>>   {
>>   	add_segment_phys_virt(info, buf, bufsz, base, memsz, 1);
>>   }
>> +
>> +int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
>> +{
>> +	return 0;
>> +}
>> diff --git a/kexec/arch/mips/kexec-mips.c b/kexec/arch/mips/kexec-mips.c
>> index d8cbea8..94224ee 100644
>> --- a/kexec/arch/mips/kexec-mips.c
>> +++ b/kexec/arch/mips/kexec-mips.c
>> @@ -189,3 +189,7 @@ unsigned long add_buffer(struct kexec_info *info, const void *buf,
>>   				    buf_min, buf_max, buf_end, 1);
>>   }
>>   
>> +int arch_do_exclude_segment(const void *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
>> +{
>> +	return 0;
>> +}
>> diff --git a/kexec/arch/ppc/kexec-ppc.c b/kexec/arch/ppc/kexec-ppc.c
>> index 03bec36..c8af870 100644
>> --- a/kexec/arch/ppc/kexec-ppc.c
>> +++ b/kexec/arch/ppc/kexec-ppc.c
>> @@ -966,3 +966,7 @@ void arch_update_purgatory(struct kexec_info *UNUSED(info))
>>   {
>>   }
>>   
>> +int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
>> +{
>> +	return 0;
>> +}
>> diff --git a/kexec/arch/ppc64/kexec-ppc64.c b/kexec/arch/ppc64/kexec-ppc64.c
>> index bd5274c..fb27b6b 100644
>> --- a/kexec/arch/ppc64/kexec-ppc64.c
>> +++ b/kexec/arch/ppc64/kexec-ppc64.c
>> @@ -967,3 +967,8 @@ int arch_compat_trampoline(struct kexec_info *UNUSED(info))
>>   void arch_update_purgatory(struct kexec_info *UNUSED(info))
>>   {
>>   }
>> +
>> +int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
>> +{
>> +	return 0;
>> +}
>> diff --git a/kexec/arch/s390/kexec-s390.c b/kexec/arch/s390/kexec-s390.c
>> index 33ba6b9..0561ee7 100644
>> --- a/kexec/arch/s390/kexec-s390.c
>> +++ b/kexec/arch/s390/kexec-s390.c
>> @@ -267,3 +267,8 @@ int get_crash_kernel_load_range(uint64_t *start, uint64_t *end)
>>   {
>>   	return parse_iomem_single("Crash kernel\n", start, end);
>>   }
>> +
>> +int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
>> +{
>> +	return 0;
>> +}
>> diff --git a/kexec/arch/sh/kexec-sh.c b/kexec/arch/sh/kexec-sh.c
>> index ce341c8..f84c40c 100644
>> --- a/kexec/arch/sh/kexec-sh.c
>> +++ b/kexec/arch/sh/kexec-sh.c
>> @@ -257,3 +257,8 @@ unsigned long add_buffer(struct kexec_info *info, const void *buf,
>>   	return add_buffer_phys_virt(info, buf, bufsz, memsz, buf_align,
>>   				    buf_min, buf_max, buf_end, 1);
>>   }
>> +
>> +int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
>> +{
>> +	return 0;
>> +}
>> diff --git a/kexec/arch/x86_64/kexec-x86_64.c b/kexec/arch/x86_64/kexec-x86_64.c
>> index ffd84f0..42af90a 100644
>> --- a/kexec/arch/x86_64/kexec-x86_64.c
>> +++ b/kexec/arch/x86_64/kexec-x86_64.c
>> @@ -188,3 +188,8 @@ void arch_update_purgatory(struct kexec_info *info)
>>   	elf_rel_set_symbol(&info->rhdr, "panic_kernel",
>>   				&panic_kernel, sizeof(panic_kernel));
>>   }
>> +
>> +int arch_do_exclude_segment(struct kexec_segment *UNUSED(seg_ptr), struct kexec_info *UNUSED(info))
>> +{
>> +	return 0;
>> +}
>> diff --git a/kexec/kexec-syscall.h b/kexec/kexec-syscall.h
>> index 73e5254..4675c46 100644
>> --- a/kexec/kexec-syscall.h
>> +++ b/kexec/kexec-syscall.h
>> @@ -112,7 +112,7 @@ static inline long kexec_file_load(int kernel_fd, int initrd_fd,
>>   
>>   #define KEXEC_ON_CRASH		0x00000001
>>   #define KEXEC_PRESERVE_CONTEXT	0x00000002
>> -#define KEXEC_UPDATE_ELFCOREHDR	0x00000004
>> +#define KEXEC_CRASH_HOTPLUG_SUPPORT	0x00000008
>>   #define KEXEC_ARCH_MASK		0xffff0000
> Yeah, removing KEXEC_UPDATE_ELFCOREHDR sounds good to avoid collision
> with xen code.

Thanks for the review.

I will make the necessary changes as mentioned above and send v4.

Thanks,
Sourabh Jain

_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec

Re: [PATCH v3 3/3] doc/hotplug: update man and --help

[Sourabh Jain]

On 03/07/24 11:00, Baoquan He wrote:
> On 07/02/24 at 10:00am, Sourabh Jain wrote:
>> Update the man page and --help option to make the description of the
>> --hotplug option easier to understand.
>>
>> Cc: Aditya Gupta <adityag@linux.ibm.com>
>> Cc: Baoquan He <bhe@redhat.com>
>> Cc: Coiby Xu <coxu@redhat.com>
>> Cc: Mahesh Salgaonkar <mahesh@linux.ibm.com>
>> Cc: Simon Horman <horms@kernel.org>
>> Acked-by: Hari Bathini <hbathini@linux.ibm.com>
>> Signed-off-by: Sourabh Jain <sourabhjain@linux.ibm.com>
>> ---
>> Changelog:
>>
>>  From v2 -> v3
>>    - Updated --hotplug option description
>>
>> ---
>>   kexec/kexec.8 | 8 ++++----
>>   kexec/kexec.c | 4 +++-
>>   2 files changed, 7 insertions(+), 5 deletions(-)
> Acked-by: Baoquan He <bhe@redhat.com>
Thanks for Ack!

- Sourabh Jain

_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec

[PATCH v3 2/3] powerpc/kexec_load: add hotplug support

[Sourabh Jain]

Kernel commits b741092d5976 ("powerpc/crash: add crash CPU hotplug
support") and 849599b702ef ("powerpc/crash: add crash memory hotplug
support") added crash CPU/Memory hotplug support on PowerPC. This patch
extends that support for the kexec_load syscall.

During CPU/Memory hotplug events on PowerPC, two kexec segments,
elfcorehdr, and FDT, get updated by the kernel. To ensure the kernel
can safely update these two kexec segments for the kdump image loaded
using the kexec_load system call, the following changes are made:

1. Extra size is allocated for both elfcorehdr and FDT to accommodate
   additional resources in the future. For the elfcorehdr, the size hint
   is taken from /sys/kernel/crash_elfcorehdr_size sysfs, while for FDT,
   extra size is allocated to hold possible CPU nodes.

2. Both elfcorehdr and FDT are skipped from SHA calculation.

Cc: Aditya Gupta <adityag@linux.ibm.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Coiby Xu <coxu@redhat.com>
Cc: Mahesh Salgaonkar <mahesh@linux.ibm.com>
Cc: Simon Horman <horms@kernel.org>
Acked-by: Hari Bathini <hbathini@linux.ibm.com>
Signed-off-by: Sourabh Jain <sourabhjain@linux.ibm.com>
---
 kexec/arch/ppc64/crashdump-ppc64.c  |  16 +-
 kexec/arch/ppc64/fdt.c              | 236 +++++++++++++++++++++++++++-
 kexec/arch/ppc64/include/arch/fdt.h |   2 +-
 kexec/arch/ppc64/kexec-elf-ppc64.c  |   2 +-
 kexec/arch/ppc64/kexec-ppc64.c      |   9 +-
 5 files changed, 258 insertions(+), 7 deletions(-)

diff --git a/kexec/arch/ppc64/crashdump-ppc64.c b/kexec/arch/ppc64/crashdump-ppc64.c
index 6d47898..98d439a 100644
--- a/kexec/arch/ppc64/crashdump-ppc64.c
+++ b/kexec/arch/ppc64/crashdump-ppc64.c
@@ -476,7 +476,7 @@ int load_crashdump_segments(struct kexec_info *info, char* mod_cmdline,
 				uint64_t max_addr, unsigned long min_base)
 {
 	void *tmp;
-	unsigned long sz;
+	unsigned long sz, memsz;
 	uint64_t elfcorehdr;
 	int nr_ranges, align = 1024, i;
 	unsigned long long end;
@@ -531,8 +531,18 @@ int load_crashdump_segments(struct kexec_info *info, char* mod_cmdline,
 		}
 	}
 
-	elfcorehdr = add_buffer(info, tmp, sz, sz, align, min_base,
-				max_addr, 1);
+	memsz = sz;
+	/* To support --hotplug, replace the calculated memsz with the value
+	 * from /sys/kernel/crash_elfcorehdr_size and align it correctly.
+	 */
+	if (do_hotplug) {
+		if (elfcorehdrsz > sz)
+			memsz = _ALIGN(elfcorehdrsz, align);
+	}
+
+	/* Record the location of the elfcorehdr for hotplug handling */
+	info->elfcorehdr = elfcorehdr = add_buffer(info, tmp, sz, memsz, align,
+						   min_base, max_addr, 1);
 	reserve(elfcorehdr, sz);
 	/* modify and store the cmdline in a global array. This is later
 	 * read by flatten_device_tree and modified if required
diff --git a/kexec/arch/ppc64/fdt.c b/kexec/arch/ppc64/fdt.c
index 8bc6d2d..879240f 100644
--- a/kexec/arch/ppc64/fdt.c
+++ b/kexec/arch/ppc64/fdt.c
@@ -17,6 +17,13 @@
 #include <libfdt.h>
 #include <stdio.h>
 #include <stdlib.h>
+#include <limits.h>
+#include <stdbool.h>
+#include <dirent.h>
+#include <sys/stat.h>
+
+#include "../../kexec.h"
+#include "../../kexec-syscall.h"
 
 /*
  * Let the kernel know it booted from kexec, as some things (e.g.
@@ -46,17 +53,244 @@ static int fixup_kexec_prop(void *fdt)
 	return 0;
 }
 
+static inline bool is_dot_dir(char * d_path)
+{
+	return d_path[0] == '.';
+}
+
+/*
+ * get_cpu_node_size - Returns size of files including file name size under
+ *                     the given @cpu_node_path.
+ */
+static int get_cpu_node_size(char *cpu_node_path)
+{
+	DIR *d;
+	struct dirent *de;
+	struct stat statbuf;
+	int cpu_node_size = 0;
+	char cpu_prop_path[2 * PATH_MAX];
+
+	d = opendir(cpu_node_path);
+	if (!d)
+		return 0;
+
+	while ((de = readdir(d)) != NULL) {
+		if (de->d_type != DT_REG)
+			continue;
+
+		memset(cpu_prop_path, '\0', PATH_MAX);
+		snprintf(cpu_prop_path, 2 * PATH_MAX, "%s/%s", cpu_node_path,
+			 de->d_name);
+
+		if (stat(cpu_prop_path, &statbuf))
+			continue;
+
+		cpu_node_size += statbuf.st_size;
+		cpu_node_size += strlen(de->d_name);
+	}
+
+	return cpu_node_size;
+}
+
+/*
+ * is_cpu_node - Checks if the node specified by the given @path
+ *               represents a CPU node.
+ *
+ * Returns true if the @path has a "device_type" file containing "cpu";
+ * otherwise, returns false.
+ */
+static bool is_cpu_node(char *path)
+{
+	FILE *file;
+	bool ret = false;
+	char device_type[4];
+
+	file = fopen(path, "r");
+	if (!file)
+		return false;
+
+	memset(device_type, '\0', 4);
+	if (fread(device_type, 1, 3, file) < 3)
+		goto out;
+
+	if (strcmp(device_type, "cpu"))
+		goto out;
+
+	ret = true;
+out:
+	fclose(file);
+	return ret;
+}
+
+static int get_threads_per_cpu(char *path)
+{
+	struct stat statbuf;
+	if (stat(path, &statbuf))
+		return 0;
+
+	return statbuf.st_size / 4;
+}
+
+/**
+ * get_present_cpus - finds the present CPUs in the system
+ *
+ * This function opens the file `/sys/devices/system/cpu/present` to read
+ * the range of present CPUs. It parses the range and calculates the
+ * total number of present CPUs in the system.
+ *
+ * Returns total number of present CPUs on success, -1 on failure.
+ */
+static int get_present_cpus()
+{
+	char *range;
+	char buf[1024];
+	int start, end;
+	int cpu_count = 0;
+	FILE *file = fopen("/sys/devices/system/cpu/present", "r");
+
+	if (!file)
+		return -1;
+
+	if (!fgets(buf, sizeof(buf), file))
+		return -1;
+
+	fclose(file);
+
+	range = strtok(buf, ",");
+	while (range != NULL) {
+		if (sscanf(range, "%d-%d", &start, &end) == 2) {
+			for (int i = start; i <= end; i++)
+				cpu_count++;
+		} else if (sscanf(range, "%d", &start) == 1) {
+			cpu_count++;
+		} else {
+			return -1;
+		}
+		range = strtok(NULL, ",");
+	}
+
+	return cpu_count;
+}
+
+/*
+ * get_cpu_info - Finds the following CPU attributes:
+ *
+ * threads_per_cpu: Number of threads per CPU, based on the device tree entry
+ *                  /proc/device-tree/cpus/<cpu_node>/ibm,ppc-interrupt-server#s.
+ * cpu_node_size: Size of files including file name size under a CPU node.
+ *
+ * Returns 0 on success, else -1.
+ */
+static int get_cpu_info(int *_present_cpus, int *_threads_per_cpu, int *_cpu_node_size)
+{
+	DIR *d;
+	struct dirent *de;
+	char path[PATH_MAX];
+	int present_cpus = 0, threads_per_cpu = 0, cpu_node_size = 0;
+	char *cpus_node_path = "/proc/device-tree/cpus";
+
+	present_cpus = get_present_cpus();
+	if (present_cpus < 0)
+		return -1;
+
+	d = opendir(cpus_node_path);
+	if (!d)
+		return -1;
+
+	while ((de = readdir(d)) != NULL) {
+		if ((de->d_type != DT_DIR) || is_dot_dir(de->d_name))
+			continue;
+
+		memset(path, '\0', PATH_MAX);
+		snprintf(path, PATH_MAX, "%s/%s/%s", cpus_node_path,
+			 de->d_name, "device_type");
+
+		/* Skip nodes with device_type != "cpu" */
+		if (!is_cpu_node(path))
+			continue;
+
+		/*
+		 * Found the first node under /proc/device-tree/cpus with
+		 * device_type == "cpu"
+		 */
+		memset(path, '\0', PATH_MAX);
+		snprintf(path, PATH_MAX, "%s/%s", cpus_node_path, de->d_name);
+		cpu_node_size = get_cpu_node_size(path);
+
+		memset(path, '\0', PATH_MAX);
+		snprintf(path, PATH_MAX, "%s/%s/%s", cpus_node_path,
+		de->d_name, "ibm,ppc-interrupt-server#s");
+		threads_per_cpu = get_threads_per_cpu(path);
+		break;
+	}
+
+	closedir(d);
+
+	if (!(threads_per_cpu && cpu_node_size))
+		return -1;
+
+	*_present_cpus = present_cpus;
+	*_cpu_node_size = cpu_node_size;
+	*_threads_per_cpu = threads_per_cpu;
+
+	dbgprintf("present_cpus: %d, threads_per_cpu: %d, cpu_node_size: %d\n",
+		  present_cpus, threads_per_cpu, cpu_node_size);
+
+	return 0;
+}
+
+/*
+ * kdump_fdt_extra_size - Calculates the extra size needed for the Flattened
+ *                        Device Tree (FDT) based on the possible and present
+ *                        CPUs in the system.
+ */
+static unsigned int kdump_fdt_extra_size(void)
+{
+	int cpus_in_system;
+	unsigned int extra_size = 0;
+	int present_cpus = 0, threads_per_cpu = 0, cpu_node_size = 0;
+	int possible_cpus;
+
+	/* ALL possible CPUs are present in FDT so no extra size required */
+	if (sysconf(_SC_NPROCESSORS_ONLN) == sysconf(_SC_NPROCESSORS_CONF))
+		return 0;
+
+	if (get_cpu_info(&present_cpus, &threads_per_cpu, &cpu_node_size)) {
+		die("Failed to get cpu info\n");
+	}
+
+	cpus_in_system = present_cpus / threads_per_cpu;
+	possible_cpus = sysconf(_SC_NPROCESSORS_CONF) / threads_per_cpu;
+	dbgprintf("cpus_in_system: %d, possible_cpus: %d\n", cpus_in_system,
+		  possible_cpus);
+
+	if (cpus_in_system > possible_cpus)
+		die("Possible CPU nodes can't be less than active CPU nodes\n");
+
+	extra_size = (possible_cpus - cpus_in_system) * cpu_node_size;
+	dbgprintf("kdump fdt extra size: %u\n", extra_size);
+
+	return extra_size;
+}
 
 /*
  * For now, assume that the added content fits in the file.
  * This should be the case when flattening from /proc/device-tree,
  * and when passing in a dtb, dtc can be told to add padding.
  */
-int fixup_dt(char **fdt, off_t *size)
+int fixup_dt(char **fdt, off_t *size, unsigned long kexec_flags)
 {
 	int ret;
 
 	*size += 4096;
+
+	/* To support --hotplug option for the kexec_load syscall, consider
+	 * adding extra buffer to FDT so that the kernel can add CPU nodes
+	 * of hot-added CPUs.
+	 */
+	if (do_hotplug && (kexec_flags & KEXEC_ON_CRASH))
+		*size += kdump_fdt_extra_size();
+
 	*fdt = realloc(*fdt, *size);
 	if (!*fdt) {
 		fprintf(stderr, "%s: out of memory\n", __func__);
diff --git a/kexec/arch/ppc64/include/arch/fdt.h b/kexec/arch/ppc64/include/arch/fdt.h
index b19f185..5f340b0 100644
--- a/kexec/arch/ppc64/include/arch/fdt.h
+++ b/kexec/arch/ppc64/include/arch/fdt.h
@@ -3,6 +3,6 @@
 
 #include <sys/types.h>
 
-int fixup_dt(char **fdt, off_t *size);
+int fixup_dt(char **fdt, off_t *size, unsigned long kexec_flags);
 
 #endif
diff --git a/kexec/arch/ppc64/kexec-elf-ppc64.c b/kexec/arch/ppc64/kexec-elf-ppc64.c
index bdcfd20..858c994 100644
--- a/kexec/arch/ppc64/kexec-elf-ppc64.c
+++ b/kexec/arch/ppc64/kexec-elf-ppc64.c
@@ -345,7 +345,7 @@ int elf_ppc64_load(int argc, char **argv, const char *buf, off_t len,
 		create_flatten_tree(&seg_buf, &seg_size, cmdline);
 	}
 
-	result = fixup_dt(&seg_buf, &seg_size);
+	result = fixup_dt(&seg_buf, &seg_size, info->kexec_flags);
 	if (result < 0)
 		return result;
 
diff --git a/kexec/arch/ppc64/kexec-ppc64.c b/kexec/arch/ppc64/kexec-ppc64.c
index 90734fe..6653246 100644
--- a/kexec/arch/ppc64/kexec-ppc64.c
+++ b/kexec/arch/ppc64/kexec-ppc64.c
@@ -24,6 +24,7 @@
 #include <errno.h>
 #include <stdint.h>
 #include <string.h>
+#include <libfdt.h>
 #include <sys/stat.h>
 #include <sys/types.h>
 #include <dirent.h>
@@ -968,7 +969,13 @@ void arch_update_purgatory(struct kexec_info *UNUSED(info))
 {
 }
 
-int arch_do_exclude_segment(struct kexec_info *UNUSED(info), struct kexec_segment *UNUSED(segment))
+int arch_do_exclude_segment(struct kexec_info *info, struct kexec_segment *segment)
 {
+	if (info->elfcorehdr == (unsigned long) segment->mem)
+		return 1;
+
+	if (segment->buf && fdt_magic(segment->buf) == FDT_MAGIC)
+		return 1;
+
 	return 0;
 }
-- 
2.45.1


_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec