[PATCH v31 09/12] arm64: kdump: provide /proc/vmcore file

On Wed, Feb 01, 2017 at 09:46:28PM +0900, AKASHI Takahiro wrote:

Add arch-specific functions to provide a dump file, /proc/vmcore.

This file is in ELF format and its ELF header needs to be prepared by
userspace tools, like kexec-tools, in adance. The primary kernel is
responsible to allocate the region with reserve_elfcorehdr() at boot time
and advertize its location to crash dump kernel via a new device-tree
property, "linux,elfcorehdr".

+static int __init early_init_dt_scan_elfcorehdr(unsigned long node,
+		const char *uname, int depth, void *data)
+{
+	const __be32 *reg;
+	int len;
+
+	if (depth != 1 || strcmp(uname, "chosen") != 0)
+		return 0;
+
+	reg = of_get_flat_dt_prop(node, "linux,elfcorehdr", &len);
+	if (!reg || (len < (dt_root_addr_cells + dt_root_size_cells)))
+		return 1;
+
+	elfcorehdr_addr = dt_mem_next_cell(dt_root_addr_cells, &reg);
+	elfcorehdr_size = dt_mem_next_cell(dt_root_size_cells, &reg);
+
+	return 1;
+}
+
+/*
+ * reserve_elfcorehdr() - reserves memory for elf core header
+ *
+ * This function reserves elf core header given in "elfcorehdr=" kernel
+ * command line parameter. This region contains all the information about
+ * primary kernel's core image and is used by a dump capture kernel to
+ * access the system memory on primary kernel.
+ */
+static void __init reserve_elfcorehdr(void)
+{
+	of_scan_flat_dt(early_init_dt_scan_elfcorehdr, NULL);
+
+	if (!elfcorehdr_size)
+		return;
+
+	if (memblock_is_region_reserved(elfcorehdr_addr, elfcorehdr_size)) {
+		pr_warn("elfcorehdr is overlapped\n");
+		return;
+	}
+
+	memblock_reserve(elfcorehdr_addr, elfcorehdr_size);
+
+	pr_info("Reserving %lldKB of memory at 0x%llx for elfcorehdr\n",
+		elfcorehdr_size >> 10, elfcorehdr_addr);
+}

This doesn't seem right to me.  The logic here doesn't match the commit
message, the comment above reserve_elfcorehdr() doesn't match the
implementation, and this doesn't match my understanding of how this was
intended to be used from the DT binding.

I had assumed that we'd treat this in much the same way as the
linux,reserved-memory-region property, with the primary kernel either
dynamically allocating the region or using a command line option, and
the base being exposed to userspace via /sys/ or /proc/ somehow.

Why is that not the case?

Thanks,
Mark.