Re: [REGRESSION] Kernel booted via kexec fails to resume from hibernation

[msizanoen]

Here's an updated version of the patch that better handles pathological 
e820 tables.

On 4/4/25 11:56, msizanoen wrote:
> Also, can you reproduce this issue with a target kernel (the kernel 
> being kexec-ed) that has one of the patches attached (select the 
> correct one according to your kernel version) applied, with either 
> kexec_load or kexec_file_load?
>
> On 4/4/25 09:54, msizanoen wrote:
> > Can you send the dmesg logs for this case (6.13 + mentioned patch 
> > series backported as target kernel, using kexec_load)?
> >
> > On 4/4/25 05:00, Roberto Ricci wrote:
> > > On 2025-04-01 19:59 +0700, msizanoen wrote:
> > > > [snip]
> > > > It seems like `e820__register_nosave_regions` is erroneously 
> > > > marking some
> > > > kernel memory as nosave in the presence of sub-page e820 regions. 
> > > > In theory
> > > > backporting
> > > > https://lore.kernel.org/all/20250214090651.3331663-1-rppt@xxxxxxxxxx/ 
> > > > should
> > > > be sufficient to avoid this but a fix for the actual root cause is
> > > > preferred.
> > > When using kexec_file_load, this patch series fixes the issue not only
> > > in theory but also in practice.
> > > But the issue with kexec_load (see
> > > https://lore.kernel.org/all/Z-hYWc9LtBU1Yhtg@desktop0a/
> > > ), which might be related, is not fixed.
diff --git a/arch/x86/kernel/e820.c b/arch/x86/kernel/e820.c
index 4893d30ce438..5d963df63b7a 100644
--- a/arch/x86/kernel/e820.c
+++ b/arch/x86/kernel/e820.c
@@ -754,22 +754,21 @@ void __init e820__memory_setup_extended(u64 phys_addr, u32 data_len)
 void __init e820__register_nosave_regions(unsigned long limit_pfn)
 {
 	int i;
-	unsigned long pfn = 0;
+	u64 last_addr = 0;
 
 	for (i = 0; i < e820_table->nr_entries; i++) {
 		struct e820_entry *entry = &e820_table->entries[i];
 
-		if (pfn < PFN_UP(entry->addr))
-			register_nosave_region(pfn, PFN_UP(entry->addr));
-
-		pfn = PFN_DOWN(entry->addr + entry->size);
-
 		if (entry->type != E820_TYPE_RAM && entry->type != E820_TYPE_RESERVED_KERN)
-			register_nosave_region(PFN_UP(entry->addr), pfn);
+			continue;
 
-		if (pfn >= limit_pfn)
-			break;
+		if (last_addr < entry->addr)
+			register_nosave_region(PFN_UP(last_addr), PFN_DOWN(entry->addr));
+
+		last_addr = entry->addr + entry->size;
 	}
+
+	register_nosave_region(PFN_UP(last_addr), limit_pfn);
 }
 
 #ifdef CONFIG_ACPI
diff --git a/arch/x86/kernel/e820.c b/arch/x86/kernel/e820.c
index 57120f0749cc..656ed7abd28d 100644
--- a/arch/x86/kernel/e820.c
+++ b/arch/x86/kernel/e820.c
@@ -753,22 +753,21 @@ void __init e820__memory_setup_extended(u64 phys_addr, u32 data_len)
 void __init e820__register_nosave_regions(unsigned long limit_pfn)
 {
 	int i;
-	unsigned long pfn = 0;
+	u64 last_addr = 0;
 
 	for (i = 0; i < e820_table->nr_entries; i++) {
 		struct e820_entry *entry = &e820_table->entries[i];
 
-		if (pfn < PFN_UP(entry->addr))
-			register_nosave_region(pfn, PFN_UP(entry->addr));
-
-		pfn = PFN_DOWN(entry->addr + entry->size);
-
 		if (entry->type != E820_TYPE_RAM)
-			register_nosave_region(PFN_UP(entry->addr), pfn);
+			continue;
 
-		if (pfn >= limit_pfn)
-			break;
+		if (last_addr < entry->addr)
+			register_nosave_region(PFN_UP(last_addr), PFN_DOWN(entry->addr));
+
+		last_addr = entry->addr + entry->size;
 	}
+
+	register_nosave_region(PFN_UP(last_addr), limit_pfn);
 }
 
 #ifdef CONFIG_ACPI