On Fri, Oct 24 2025, Mike Rapoport wrote:
On Wed, Oct 22, 2025 at 01:01:08PM +0200, Pratyush Yadav wrote:
Hi Pasha,
On Tue, Oct 21 2025, Pasha Tatashin wrote:
From: "Mike Rapoport (Microsoft)" rppt@kernel.org
The KHO framework uses a notifier chain as the mechanism for clients to participate in the finalization process. While this works for a single, central state machine, it is too restrictive for kernel-internal components like pstore/reserve_mem or IMA. These components need a simpler, direct way to register their state for preservation (e.g., during their initcall) without being part of a complex, shutdown-time notifier sequence. The notifier model forces all participants into a single finalization flow and makes direct preservation from an arbitrary context difficult. This patch refactors the client participation model by removing the notifier chain and introducing a direct API for managing FDT subtrees.
The core kho_finalize() and kho_abort() state machine remains, but clients now register their data with KHO beforehand.
...
@@ -1280,7 +1298,7 @@ static __init int kho_init(void) kho_enable = false; return err; } -late_initcall(kho_init); +fs_initcall(kho_init);
Is this change related to this patch? Also, why fs_initcall?
memblock registers sub-fdt in late_initcall(), so we should have the root fdt ready by then.
I see. Should this be even earlier then? Other components might also depend on KHO being initialized, and those might be at or before fs_initcall.
For example, LUO does its init using early_initcall and uses KHO, even before it is initialized [0]. This works because kho_retrieve_subtree() only uses parts initialized very early in boot (the KHO FDT), but I suppose we want to have a proper initialization order to not rely on things that just happen to work until they don't.
Since kho_init() has a dependency on debugfs, which gets initialized in core_initcall, I guess the earliest it can be is postcore_initcall. Or, we split out the debugfs parts into a separate init function (they have their own file anyway) and initialize "core KHO" in early_initcall? Then LUO can be in core_initcall and all its users in later ones.
Thoughts?
[0] https://lore.kernel.org/lkml/20250929010321.3462457-9-pasha.tatashin@soleen....