kernel: vmcoreinfo: allocate vmcoreinfo_data based on VMCOREINFO_BYTES

Patch series "vmcoreinfo: support VMCOREINFO_BYTES larger than PAGE_SIZE".

VMCOREINFO_BYTES is defined as a configurable size, but multiple
code paths implicitly assume it always fits into a single page.

This series removes that assumption by allocating and mapping
vmcoreinfo based on its actual size.

Patch 1 updates vmcoreinfo allocation to use get_order(VMCOREINFO_BYTES).
Patch 2 updates crash kernel handling to correctly allocate and map
multiple pages when copying vmcoreinfo.

This makes vmcoreinfo size consistent across the kernel and avoids
future breakage if VMCOREINFO_BYTES grows.

(No functional change when VMCOREINFO_BYTES == PAGE_SIZE.)


This patch (of 2):

VMCOREINFO_BYTES defines the size of vmcoreinfo data, but the current
implementation assumes a single page allocation.

Allocate vmcoreinfo_data using get_order(VMCOREINFO_BYTES) so that
vmcoreinfo can safely grow beyond PAGE_SIZE.

This avoids hidden assumptions and keeps vmcoreinfo size consistent across
the kernel.

Link: https://lkml.kernel.org/r/[email protected]
Link: https://lkml.kernel.org/r/[email protected]
Signed-off-by: Pnina Feder <[email protected]>
Reviewed-by: Andrew Morton <[email protected]>
Cc: Baoquan He <[email protected]>
Cc: Dave Young <[email protected]>
Cc: Vivek Goyal <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>

Author: Pnina Feder

kernel/vmcore_info.c

@@ -137,7 +137,9 @@ EXPORT_SYMBOL_GPL(hwerr_log_error_type);
 
 static int __init crash_save_vmcoreinfo_init(void)
 {
-	vmcoreinfo_data = (unsigned char *)get_zeroed_page(GFP_KERNEL);
+	int order;
+	order = get_order(VMCOREINFO_BYTES);
+	vmcoreinfo_data = (unsigned char *)__get_free_pages(GFP_KERNEL | __GFP_ZERO, order);
 	if (!vmcoreinfo_data) {
 		pr_warn("Memory allocation for vmcoreinfo_data failed\n");
 		return -ENOMEM;
@@ -146,7 +148,7 @@ static int __init crash_save_vmcoreinfo_init(void)
 	vmcoreinfo_note = alloc_pages_exact(VMCOREINFO_NOTE_SIZE,
 						GFP_KERNEL | __GFP_ZERO);
 	if (!vmcoreinfo_note) {
-		free_page((unsigned long)vmcoreinfo_data);
+		free_pages((unsigned long)vmcoreinfo_data, order);
 		vmcoreinfo_data = NULL;
 		pr_warn("Memory allocation for vmcoreinfo_note failed\n");
 		return -ENOMEM;