1 files changed, 205 insertions, 156 deletions

diff --git a/arch/powerpc/kernel/fadump.c b/arch/powerpc/kernel/fadump.c
index 4e768d93c6d4..a020597c065a 100644
--- a/arch/powerpc/kernel/fadump.c
+++ b/arch/powerpc/kernel/fadump.c
@@ -53,8 +53,6 @@ static struct kobject *fadump_kobj;
 static atomic_t cpus_in_fadump;
 static DEFINE_MUTEX(fadump_mutex);
 
-static struct fadump_mrange_info crash_mrange_info = { "crash", NULL, 0, 0, 0, false };
-
 #define RESERVED_RNGS_SZ	16384 /* 16K - 128 entries */
 #define RESERVED_RNGS_CNT	(RESERVED_RNGS_SZ / \
 				 sizeof(struct fadump_memory_range))
@@ -373,12 +371,6 @@ static unsigned long __init get_fadump_area_size(void)
 	size = PAGE_ALIGN(size);
 	size += fw_dump.boot_memory_size;
 	size += sizeof(struct fadump_crash_info_header);
-	size += sizeof(struct elfhdr); /* ELF core header.*/
-	size += sizeof(struct elf_phdr); /* place holder for cpu notes */
-	/* Program headers for crash memory regions. */
-	size += sizeof(struct elf_phdr) * (memblock_num_regions(memory) + 2);
-
-	size = PAGE_ALIGN(size);
 
 	/* This is to hold kernel metadata on platforms that support it */
 	size += (fw_dump.ops->fadump_get_metadata_size ?
@@ -915,36 +907,6 @@ static inline int fadump_add_mem_range(struct fadump_mrange_info *mrange_info,
 	return 0;
 }
 
-static int fadump_exclude_reserved_area(u64 start, u64 end)
-{
-	u64 ra_start, ra_end;
-	int ret = 0;
-
-	ra_start = fw_dump.reserve_dump_area_start;
-	ra_end = ra_start + fw_dump.reserve_dump_area_size;
-
-	if ((ra_start < end) && (ra_end > start)) {
-		if ((start < ra_start) && (end > ra_end)) {
-			ret = fadump_add_mem_range(&crash_mrange_info,
-						   start, ra_start);
-			if (ret)
-				return ret;
-
-			ret = fadump_add_mem_range(&crash_mrange_info,
-						   ra_end, end);
-		} else if (start < ra_start) {
-			ret = fadump_add_mem_range(&crash_mrange_info,
-						   start, ra_start);
-		} else if (ra_end < end) {
-			ret = fadump_add_mem_range(&crash_mrange_info,
-						   ra_end, end);
-		}
-	} else
-		ret = fadump_add_mem_range(&crash_mrange_info, start, end);
-
-	return ret;
-}
-
 static int fadump_init_elfcore_header(char *bufp)
 {
 	struct elfhdr *elf;
@@ -982,52 +944,6 @@ static int fadump_init_elfcore_header(char *bufp)
 }
 
 /*
- * Traverse through memblock structure and setup crash memory ranges. These
- * ranges will be used create PT_LOAD program headers in elfcore header.
- */
-static int fadump_setup_crash_memory_ranges(void)
-{
-	u64 i, start, end;
-	int ret;
-
-	pr_debug("Setup crash memory ranges.\n");
-	crash_mrange_info.mem_range_cnt = 0;
-
-	/*
-	 * Boot memory region(s) registered with firmware are moved to
-	 * different location at the time of crash. Create separate program
-	 * header(s) for this memory chunk(s) with the correct offset.
-	 */
-	for (i = 0; i < fw_dump.boot_mem_regs_cnt; i++) {
-		start = fw_dump.boot_mem_addr[i];
-		end = start + fw_dump.boot_mem_sz[i];
-		ret = fadump_add_mem_range(&crash_mrange_info, start, end);
-		if (ret)
-			return ret;
-	}
-
-	for_each_mem_range(i, &start, &end) {
-		/*
-		 * skip the memory chunk that is already added
-		 * (0 through boot_memory_top).
-		 */
-		if (start < fw_dump.boot_mem_top) {
-			if (end > fw_dump.boot_mem_top)
-				start = fw_dump.boot_mem_top;
-			else
-				continue;
-		}
-
-		/* add this range excluding the reserved dump area. */
-		ret = fadump_exclude_reserved_area(start, end);
-		if (ret)
-			return ret;
-	}
-
-	return 0;
-}
-
-/*
  * If the given physical address falls within the boot memory region then
  * return the relocated address that points to the dump region reserved
  * for saving initial boot memory contents.
@@ -1057,36 +973,50 @@ static inline unsigned long fadump_relocate(unsigned long paddr)
 	return raddr;
 }
 
-static int fadump_create_elfcore_headers(char *bufp)
+static void __init populate_elf_pt_load(struct elf_phdr *phdr, u64 start,
+			     u64 size, unsigned long long offset)
 {
-	unsigned long long raddr, offset;
-	struct elf_phdr *phdr;
+	phdr->p_align	= 0;
+	phdr->p_memsz	= size;
+	phdr->p_filesz	= size;
+	phdr->p_paddr	= start;
+	phdr->p_offset	= offset;
+	phdr->p_type	= PT_LOAD;
+	phdr->p_flags	= PF_R|PF_W|PF_X;
+	phdr->p_vaddr	= (unsigned long)__va(start);
+}
+
+static void __init fadump_populate_elfcorehdr(struct fadump_crash_info_header *fdh)
+{
+	char *bufp;
 	struct elfhdr *elf;
-	int i, j;
+	struct elf_phdr *phdr;
+	u64 boot_mem_dest_offset;
+	unsigned long long i, ra_start, ra_end, ra_size, mstart, mend;
 
+	bufp = (char *) fw_dump.elfcorehdr_addr;
 	fadump_init_elfcore_header(bufp);
 	elf = (struct elfhdr *)bufp;
 	bufp += sizeof(struct elfhdr);
 
 	/*
-	 * setup ELF PT_NOTE, place holder for cpu notes info. The notes info
-	 * will be populated during second kernel boot after crash. Hence
-	 * this PT_NOTE will always be the first elf note.
+	 * Set up ELF PT_NOTE, a placeholder for CPU notes information.
+	 * The notes info will be populated later by platform-specific code.
+	 * Hence, this PT_NOTE will always be the first ELF note.
 	 *
 	 * NOTE: Any new ELF note addition should be placed after this note.
 	 */
 	phdr = (struct elf_phdr *)bufp;
 	bufp += sizeof(struct elf_phdr);
 	phdr->p_type = PT_NOTE;
-	phdr->p_flags = 0;
-	phdr->p_vaddr = 0;
-	phdr->p_align = 0;
-
-	phdr->p_offset = 0;
-	phdr->p_paddr = 0;
-	phdr->p_filesz = 0;
-	phdr->p_memsz = 0;
-
+	phdr->p_flags	= 0;
+	phdr->p_vaddr	= 0;
+	phdr->p_align	= 0;
+	phdr->p_offset	= 0;
+	phdr->p_paddr	= 0;
+	phdr->p_filesz	= 0;
+	phdr->p_memsz	= 0;
+	/* Increment number of program headers. */
 	(elf->e_phnum)++;
 
 	/* setup ELF PT_NOTE for vmcoreinfo */
@@ -1096,55 +1026,66 @@ static int fadump_create_elfcore_headers(char *bufp)
 	phdr->p_flags	= 0;
 	phdr->p_vaddr	= 0;
 	phdr->p_align	= 0;
-
-	phdr->p_paddr	= fadump_relocate(paddr_vmcoreinfo_note());
-	phdr->p_offset	= phdr->p_paddr;
-	phdr->p_memsz	= phdr->p_filesz = VMCOREINFO_NOTE_SIZE;
-
+	phdr->p_paddr	= phdr->p_offset = fdh->vmcoreinfo_raddr;
+	phdr->p_memsz	= phdr->p_filesz = fdh->vmcoreinfo_size;
 	/* Increment number of program headers. */
 	(elf->e_phnum)++;
 
-	/* setup PT_LOAD sections. */
-	j = 0;
-	offset = 0;
-	raddr = fw_dump.boot_mem_addr[0];
-	for (i = 0; i < crash_mrange_info.mem_range_cnt; i++) {
-		u64 mbase, msize;
-
-		mbase = crash_mrange_info.mem_ranges[i].base;
-		msize = crash_mrange_info.mem_ranges[i].size;
-		if (!msize)
-			continue;
-
+	/*
+	 * Setup PT_LOAD sections. first include boot memory regions
+	 * and then add rest of the memory regions.
+	 */
+	boot_mem_dest_offset = fw_dump.boot_mem_dest_addr;
+	for (i = 0; i < fw_dump.boot_mem_regs_cnt; i++) {
 		phdr = (struct elf_phdr *)bufp;
 		bufp += sizeof(struct elf_phdr);
-		phdr->p_type	= PT_LOAD;
-		phdr->p_flags	= PF_R|PF_W|PF_X;
-		phdr->p_offset	= mbase;
-
-		if (mbase == raddr) {
-			/*
-			 * The entire real memory region will be moved by
-			 * firmware to the specified destination_address.
-			 * Hence set the correct offset.
-			 */
-			phdr->p_offset = fw_dump.boot_mem_dest_addr + offset;
-			if (j < (fw_dump.boot_mem_regs_cnt - 1)) {
-				offset += fw_dump.boot_mem_sz[j];
-				raddr = fw_dump.boot_mem_addr[++j];
-			}
+		populate_elf_pt_load(phdr, fw_dump.boot_mem_addr[i],
+				     fw_dump.boot_mem_sz[i],
+				     boot_mem_dest_offset);
+		/* Increment number of program headers. */
+		(elf->e_phnum)++;
+		boot_mem_dest_offset += fw_dump.boot_mem_sz[i];
+	}
+
+	/* Memory reserved for fadump in first kernel */
+	ra_start = fw_dump.reserve_dump_area_start;
+	ra_size = get_fadump_area_size();
+	ra_end = ra_start + ra_size;
+
+	phdr = (struct elf_phdr *)bufp;
+	for_each_mem_range(i, &mstart, &mend) {
+		/* Boot memory regions already added, skip them now */
+		if (mstart < fw_dump.boot_mem_top) {
+			if (mend > fw_dump.boot_mem_top)
+				mstart = fw_dump.boot_mem_top;
+			else
+				continue;
 		}
 
-		phdr->p_paddr = mbase;
-		phdr->p_vaddr = (unsigned long)__va(mbase);
-		phdr->p_filesz = msize;
-		phdr->p_memsz = msize;
-		phdr->p_align = 0;
+		/* Handle memblock regions overlaps with fadump reserved area */
+		if ((ra_start < mend) && (ra_end > mstart)) {
+			if ((mstart < ra_start) && (mend > ra_end)) {
+				populate_elf_pt_load(phdr, mstart, ra_start - mstart, mstart);
+				/* Increment number of program headers. */
+				(elf->e_phnum)++;
+				bufp += sizeof(struct elf_phdr);
+				phdr = (struct elf_phdr *)bufp;
+				populate_elf_pt_load(phdr, ra_end, mend - ra_end, ra_end);
+			} else if (mstart < ra_start) {
+				populate_elf_pt_load(phdr, mstart, ra_start - mstart, mstart);
+			} else if (ra_end < mend) {
+				populate_elf_pt_load(phdr, ra_end, mend - ra_end, ra_end);
+			}
+		} else {
+		/* No overlap with fadump reserved memory region */
+			populate_elf_pt_load(phdr, mstart, mend - mstart, mstart);
+		}
 
 		/* Increment number of program headers. */
 		(elf->e_phnum)++;
+		bufp += sizeof(struct elf_phdr);
+		phdr = (struct elf_phdr *) bufp;
 	}
-	return 0;
 }
 
 static unsigned long init_fadump_header(unsigned long addr)
@@ -1159,14 +1100,25 @@ static unsigned long init_fadump_header(unsigned long addr)
 
 	memset(fdh, 0, sizeof(struct fadump_crash_info_header));
 	fdh->magic_number = FADUMP_CRASH_INFO_MAGIC;
-	fdh->elfcorehdr_addr = addr;
+	fdh->version = FADUMP_HEADER_VERSION;
 	/* We will set the crashing cpu id in crash_fadump() during crash. */
 	fdh->crashing_cpu = FADUMP_CPU_UNKNOWN;
+
+	/*
+	 * The physical address and size of vmcoreinfo are required in the
+	 * second kernel to prepare elfcorehdr.
+	 */
+	fdh->vmcoreinfo_raddr = fadump_relocate(paddr_vmcoreinfo_note());
+	fdh->vmcoreinfo_size = VMCOREINFO_NOTE_SIZE;
+
+
+	fdh->pt_regs_sz = sizeof(struct pt_regs);
 	/*
 	 * When LPAR is terminated by PYHP, ensure all possible CPUs'
 	 * register data is processed while exporting the vmcore.
 	 */
 	fdh->cpu_mask = *cpu_possible_mask;
+	fdh->cpu_mask_sz = sizeof(struct cpumask);
 
 	return addr;
 }
@@ -1174,8 +1126,6 @@ static unsigned long init_fadump_header(unsigned long addr)
 static int register_fadump(void)
 {
 	unsigned long addr;
-	void *vaddr;
-	int ret;
 
 	/*
 	 * If no memory is reserved then we can not register for firmware-
@@ -1184,18 +1134,10 @@ static int register_fadump(void)
 	if (!fw_dump.reserve_dump_area_size)
 		return -ENODEV;
 
-	ret = fadump_setup_crash_memory_ranges();
-	if (ret)
-		return ret;
-
 	addr = fw_dump.fadumphdr_addr;
 
 	/* Initialize fadump crash info header. */
 	addr = init_fadump_header(addr);
-	vaddr = __va(addr);
-
-	pr_debug("Creating ELF core headers at %#016lx\n", addr);
-	fadump_create_elfcore_headers(vaddr);
 
 	/* register the future kernel dump with firmware. */
 	pr_debug("Registering for firmware-assisted kernel dump...\n");
@@ -1214,7 +1156,6 @@ void fadump_cleanup(void)
 	} else if (fw_dump.dump_registered) {
 		/* Un-register Firmware-assisted dump if it was registered. */
 		fw_dump.ops->fadump_unregister(&fw_dump);
-		fadump_free_mem_ranges(&crash_mrange_info);
 	}
 
 	if (fw_dump.ops->fadump_cleanup)
@@ -1400,6 +1341,22 @@ static void fadump_release_memory(u64 begin, u64 end)
 		fadump_release_reserved_area(tstart, end);
 }
 
+static void fadump_free_elfcorehdr_buf(void)
+{
+	if (fw_dump.elfcorehdr_addr == 0 || fw_dump.elfcorehdr_size == 0)
+		return;
+
+	/*
+	 * Before freeing the memory of `elfcorehdr`, reset the global
+	 * `elfcorehdr_addr` to prevent modules like `vmcore` from accessing
+	 * invalid memory.
+	 */
+	elfcorehdr_addr = ELFCORE_ADDR_ERR;
+	fadump_free_buffer(fw_dump.elfcorehdr_addr, fw_dump.elfcorehdr_size);
+	fw_dump.elfcorehdr_addr = 0;
+	fw_dump.elfcorehdr_size = 0;
+}
+
 static void fadump_invalidate_release_mem(void)
 {
 	mutex_lock(&fadump_mutex);
@@ -1411,6 +1368,7 @@ static void fadump_invalidate_release_mem(void)
 	fadump_cleanup();
 	mutex_unlock(&fadump_mutex);
 
+	fadump_free_elfcorehdr_buf();
 	fadump_release_memory(fw_dump.boot_mem_top, memblock_end_of_DRAM());
 	fadump_free_cpu_notes_buf();
 
@@ -1616,6 +1574,102 @@ static void __init fadump_init_files(void)
 	return;
 }
 
+static int __init fadump_setup_elfcorehdr_buf(void)
+{
+	int elf_phdr_cnt;
+	unsigned long elfcorehdr_size;
+
+	/*
+	 * Program header for CPU notes comes first, followed by one for
+	 * vmcoreinfo, and the remaining program headers correspond to
+	 * memory regions.
+	 */
+	elf_phdr_cnt = 2 + fw_dump.boot_mem_regs_cnt + memblock_num_regions(memory);
+	elfcorehdr_size = sizeof(struct elfhdr) + (elf_phdr_cnt * sizeof(struct elf_phdr));
+	elfcorehdr_size = PAGE_ALIGN(elfcorehdr_size);
+
+	fw_dump.elfcorehdr_addr = (u64)fadump_alloc_buffer(elfcorehdr_size);
+	if (!fw_dump.elfcorehdr_addr) {
+		pr_err("Failed to allocate %lu bytes for elfcorehdr\n",
+		       elfcorehdr_size);
+		return -ENOMEM;
+	}
+	fw_dump.elfcorehdr_size = elfcorehdr_size;
+	return 0;
+}
+
+/*
+ * Check if the fadump header of crashed kernel is compatible with fadump kernel.
+ *
+ * It checks the magic number, endianness, and size of non-primitive type
+ * members of fadump header to ensure safe dump collection.
+ */
+static bool __init is_fadump_header_compatible(struct fadump_crash_info_header *fdh)
+{
+	if (fdh->magic_number == FADUMP_CRASH_INFO_MAGIC_OLD) {
+		pr_err("Old magic number, can't process the dump.\n");
+		return false;
+	}
+
+	if (fdh->magic_number != FADUMP_CRASH_INFO_MAGIC) {
+		if (fdh->magic_number == swab64(FADUMP_CRASH_INFO_MAGIC))
+			pr_err("Endianness mismatch between the crashed and fadump kernels.\n");
+		else
+			pr_err("Fadump header is corrupted.\n");
+
+		return false;
+	}
+
+	/*
+	 * Dump collection is not safe if the size of non-primitive type members
+	 * of the fadump header do not match between crashed and fadump kernel.
+	 */
+	if (fdh->pt_regs_sz != sizeof(struct pt_regs) ||
+	    fdh->cpu_mask_sz != sizeof(struct cpumask)) {
+		pr_err("Fadump header size mismatch.\n");
+		return false;
+	}
+
+	return true;
+}
+
+static void __init fadump_process(void)
+{
+	struct fadump_crash_info_header *fdh;
+
+	fdh = (struct fadump_crash_info_header *) __va(fw_dump.fadumphdr_addr);
+	if (!fdh) {
+		pr_err("Crash info header is empty.\n");
+		goto err_out;
+	}
+
+	/* Avoid processing the dump if fadump header isn't compatible */
+	if (!is_fadump_header_compatible(fdh))
+		goto err_out;
+
+	/* Allocate buffer for elfcorehdr */
+	if (fadump_setup_elfcorehdr_buf())
+		goto err_out;
+
+	fadump_populate_elfcorehdr(fdh);
+
+	/* Let platform update the CPU notes in elfcorehdr */
+	if (fw_dump.ops->fadump_process(&fw_dump) < 0)
+		goto err_out;
+
+	/*
+	 * elfcorehdr is now ready to be exported.
+	 *
+	 * set elfcorehdr_addr so that vmcore module will export the
+	 * elfcorehdr through '/proc/vmcore'.
+	 */
+	elfcorehdr_addr = virt_to_phys((void *)fw_dump.elfcorehdr_addr);
+	return;
+
+err_out:
+	fadump_invalidate_release_mem();
+}
+
 /*
  * Prepare for firmware-assisted dump.
  */
@@ -1635,12 +1689,7 @@ int __init setup_fadump(void)
 	 * saving it to the disk.
 	 */
 	if (fw_dump.dump_active) {
-		/*
-		 * if dump process fails then invalidate the registration
-		 * and release memory before proceeding for re-registration.
-		 */
-		if (fw_dump.ops->fadump_process(&fw_dump) < 0)
-			fadump_invalidate_release_mem();
+		fadump_process();
 	}
 	/* Initialize the kernel dump memory structure and register with f/w */
 	else if (fw_dump.reserve_dump_area_size) {