https://gitlab.com/tezos/tezos
Tip revision: 6c5dc6f05cb36c574fc2fd65fbde164605072fa3 authored by Joel Bjornson on 08 November 2022, 16:55:40 UTC
WIP: extend minter to also take content
WIP: extend minter to also take content
Tip revision: 6c5dc6f
dal_slot_repr.ml
(*****************************************************************************)
(* *)
(* Open Source License *)
(* Copyright (c) 2022 Nomadic Labs <contact@nomadic-labs.com> *)
(* *)
(* Permission is hereby granted, free of charge, to any person obtaining a *)
(* copy of this software and associated documentation files (the "Software"),*)
(* to deal in the Software without restriction, including without limitation *)
(* the rights to use, copy, modify, merge, publish, distribute, sublicense, *)
(* and/or sell copies of the Software, and to permit persons to whom the *)
(* Software is furnished to do so, subject to the following conditions: *)
(* *)
(* The above copyright notice and this permission notice shall be included *)
(* in all copies or substantial portions of the Software. *)
(* *)
(* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR*)
(* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, *)
(* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL *)
(* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER*)
(* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING *)
(* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER *)
(* DEALINGS IN THE SOFTWARE. *)
(* *)
(*****************************************************************************)
type parameters = Dal.parameters = {
redundancy_factor : int;
page_size : int;
slot_size : int;
number_of_shards : int;
}
let parameters_encoding = Dal.parameters_encoding
module Commitment = struct
(* DAL/FIXME https://gitlab.com/tezos/tezos/-/issues/3389
It is not clear whether the size of the slot associated to the
commitment should be given here. *)
type t = Dal.commitment
let equal = Dal.Commitment.equal
let encoding = Dal.Commitment.encoding
let pp ppf commitment =
Format.fprintf ppf "%s" (Dal.Commitment.to_b58check commitment)
let zero = Dal.Commitment.zero
end
module Index = struct
type t = int
let max_value = 255
let encoding = Data_encoding.uint8
let pp = Format.pp_print_int
let zero = 0
let of_int slot_index =
if Compare.Int.(slot_index <= max_value && slot_index >= zero) then
Some slot_index
else None
let to_int slot_index = slot_index [@@ocaml.inline always]
let compare = Compare.Int.compare
let equal = Compare.Int.equal
end
module Header = struct
type id = {published_level : Raw_level_repr.t; index : Index.t}
type t = {id : id; commitment : Commitment.t}
let slot_id_equal ({published_level; index} : id) s2 =
Raw_level_repr.equal published_level s2.published_level
&& Index.equal index s2.index
let equal {id; commitment} s2 =
slot_id_equal id s2.id && Commitment.equal commitment s2.commitment
let compare_slot_id {published_level; index} s2 =
let c = Raw_level_repr.compare published_level s2.published_level in
if Compare.Int.(c <> 0) then c else Index.compare index s2.index
let zero_id =
{
(* We don't expect to have any published slot at level
Raw_level_repr.root. *)
published_level = Raw_level_repr.root;
index = Index.zero;
}
let zero = {id = zero_id; commitment = Commitment.zero}
let id_encoding =
let open Data_encoding in
conv
(fun {published_level; index} -> (published_level, index))
(fun (published_level, index) -> {published_level; index})
(obj2
(req "level" Raw_level_repr.encoding)
(req "index" Data_encoding.uint8))
let encoding =
let open Data_encoding in
conv
(fun {id; commitment} -> (id, commitment))
(fun (id, commitment) -> {id; commitment})
(merge_objs id_encoding (obj1 (req "commitment" Commitment.encoding)))
let pp_id fmt {published_level; index} =
Format.fprintf
fmt
"published_level: %a, index: %a"
Raw_level_repr.pp
published_level
Format.pp_print_int
index
let pp fmt {id; commitment = c} =
Format.fprintf fmt "id:(%a), commitment: %a" pp_id id Commitment.pp c
end
module Slot_index = Index
module Page = struct
type content = Bytes.t
type slot_index = Index.t
let pages_per_slot = Dal.pages_per_slot
module Index = struct
type t = int
let zero = 0
let encoding = Data_encoding.int16
let pp = Format.pp_print_int
let compare = Compare.Int.compare
let equal = Compare.Int.equal
end
type t = {slot_id : Header.id; page_index : Index.t}
type proof = Dal.page_proof
let encoding =
let open Data_encoding in
conv
(fun {slot_id = {published_level; index}; page_index} ->
(published_level, index, page_index))
(fun (published_level, index, page_index) ->
{slot_id = {published_level; index}; page_index})
(obj3
(req "published_level" Raw_level_repr.encoding)
(req "slot_index" Slot_index.encoding)
(req "page_index" Index.encoding))
let equal {slot_id; page_index} p =
Header.slot_id_equal slot_id p.slot_id
&& Index.equal page_index p.page_index
let proof_encoding = Dal.page_proof_encoding
let content_encoding = Data_encoding.bytes
let pp fmt {slot_id = {published_level; index}; page_index} =
Format.fprintf
fmt
"(published_level: %a, slot_index: %a, page_index: %a)"
Raw_level_repr.pp
published_level
Slot_index.pp
index
Index.pp
page_index
let pp_proof fmt proof =
Data_encoding.Json.pp
fmt
(Data_encoding.Json.construct proof_encoding proof)
end
module Slot_market = struct
(* DAL/FIXME https://gitlab.com/tezos/tezos/-/issues/3108
Think harder about this data structure and whether it can be
optimized. *)
module Slot_index_map = Map.Make (Index)
type t = {length : int; slot_headers : Header.t Slot_index_map.t}
let init ~length =
if Compare.Int.(length < 0) then
invalid_arg "Dal_slot_repr.Slot_market.init: length cannot be negative" ;
let slot_headers = Slot_index_map.empty in
{length; slot_headers}
let length {length; _} = length
let register t new_slot_header =
let open Header in
if
not
Compare.Int.(
0 <= new_slot_header.id.index && new_slot_header.id.index < t.length)
then None
else
let has_changed = ref false in
let update = function
| None ->
has_changed := true ;
Some new_slot_header
| Some x -> Some x
in
let slot_headers =
Slot_index_map.update new_slot_header.id.index update t.slot_headers
in
let t = {t with slot_headers} in
Some (t, !has_changed)
let candidates t =
t.slot_headers |> Slot_index_map.to_seq |> Seq.map snd |> List.of_seq
end
module History = struct
(* History is represented via a skip list. The content of the cell
is the hash of a merkle proof. *)
(* A leaf of the merkle tree is a slot. *)
module Leaf = struct
type t = Header.t
let to_bytes = Data_encoding.Binary.to_bytes_exn Header.encoding
end
module Content_prefix = struct
let (_prefix : string) = "dash1"
(* 32 *)
let b58check_prefix = "\002\224\072\094\219" (* dash1(55) *)
let size = Some 32
let name = "dal_skip_list_content"
let title = "A hash to represent the content of a cell in the skip list"
end
module Content_hash = Blake2B.Make (Base58) (Content_prefix)
module Merkle_list = Merkle_list.Make (Leaf) (Content_hash)
(* Pointers of the skip lists are used to encode the content and the
backpointers. *)
module Pointer_prefix = struct
let (_prefix : string) = "dask1"
(* 32 *)
let b58check_prefix = "\002\224\072\115\035" (* dask1(55) *)
let size = Some 32
let name = "dal_skip_list_pointer"
let title = "A hash that represents the skip list pointers"
end
module Pointer_hash = Blake2B.Make (Base58) (Pointer_prefix)
module Skip_list_parameters = struct
let basis = 2
end
type error += Add_element_in_slots_skip_list_violates_ordering
let () =
register_error_kind
`Temporary
~id:"Dal_slot_repr.add_element_in_slots_skip_list_violates_ordering"
~title:"Add an element in slots skip list that violates ordering"
~description:
"Attempting to add an element on top of the Dal confirmed slots skip \
list that violates the ordering."
Data_encoding.unit
(function
| Add_element_in_slots_skip_list_violates_ordering -> Some ()
| _ -> None)
(fun () -> Add_element_in_slots_skip_list_violates_ordering)
module Skip_list = struct
include Skip_list_repr.Make (Skip_list_parameters)
(** All confirmed DAL slots will be stored in a skip list, where only the
last cell is remembered in the L1 context. The skip list is used in
the proof phase of a refutation game to verify whether a given slot
exists (i.e., confirmed) or not in the skip list. The skip list is
supposed to be sorted, as its 'search' function explicitly uses a given
`compare` function during the list traversal to quickly (in log(size))
reach the target if any.
In our case, we will store one slot per cell in the skip list and
maintain that the list is well sorted (and without redundancy) w.r.t.
the [compare_slot_id] function.
Below, we redefine the [next] function (that allows adding elements
on top of the list) to enforce that the constructed skip list is
well-sorted. We also define a wrapper around the search function to
guarantee that it can only be called with the adequate compare function.
*)
let next ~prev_cell ~prev_cell_ptr elt =
let open Tzresult_syntax in
let* () =
error_when
(Compare.Int.( <= )
(Header.compare_slot_id
elt.Header.id
(content prev_cell).Header.id)
0)
Add_element_in_slots_skip_list_violates_ordering
in
return @@ next ~prev_cell ~prev_cell_ptr elt
let search ~deref ~cell ~target_id =
search ~deref ~cell ~compare:(fun slot ->
Header.compare_slot_id slot.Header.id target_id)
end
module V1 = struct
(* The content of a cell is the hash of all the slot commitments
represented as a merkle list. *)
(* TODO/DAL: https://gitlab.com/tezos/tezos/-/issues/3765
Decide how to store attested slots in the skip list's content. *)
type content = Header.t
(* A pointer to a cell is the hash of its content and all the back
pointers. *)
type ptr = Pointer_hash.t
type history = (content, ptr) Skip_list.cell
type t = history
let history_encoding =
Skip_list.encoding Pointer_hash.encoding Header.encoding
let equal_history : history -> history -> bool =
Skip_list.equal Pointer_hash.equal Header.equal
let encoding = history_encoding
let equal : t -> t -> bool = equal_history
let genesis : t = Skip_list.genesis Header.zero
let hash_skip_list_cell cell =
let current_slot = Skip_list.content cell in
let back_pointers_hashes = Skip_list.back_pointers cell in
Data_encoding.Binary.to_bytes_exn Header.encoding current_slot
:: List.map Pointer_hash.to_bytes back_pointers_hashes
|> Pointer_hash.hash_bytes
let pp_history fmt (history : history) =
let history_hash = hash_skip_list_cell history in
Format.fprintf
fmt
"@[hash : %a@;%a@]"
Pointer_hash.pp
history_hash
(Skip_list.pp ~pp_content:Header.pp ~pp_ptr:Pointer_hash.pp)
history
module History_cache =
Bounded_history_repr.Make
(struct
let name = "dal_slots_cache"
end)
(Pointer_hash)
(struct
type t = history
let encoding = history_encoding
let pp = pp_history
let equal = equal_history
end)
let add_confirmed_slot_header (t, cache) slot_header =
let open Tzresult_syntax in
let prev_cell_ptr = hash_skip_list_cell t in
let* cache = History_cache.remember prev_cell_ptr t cache in
let* new_cell = Skip_list.next ~prev_cell:t ~prev_cell_ptr slot_header in
return (new_cell, cache)
let add_confirmed_slot_headers (t : t) cache slot_headers =
List.fold_left_e add_confirmed_slot_header (t, cache) slot_headers
let add_confirmed_slot_headers_no_cache =
let no_cache = History_cache.empty ~capacity:0L in
fun t slots ->
List.fold_left_e add_confirmed_slot_header (t, no_cache) slots >|? fst
(* Dal proofs section *)
(** An inclusion proof, for a page ID, is a list of the slots' history
skip list's cells that encodes a minimal path:
- from a starting cell, which serves as a reference. It is usually called
'snapshot' below,
- to a final cell, that is either the exact target cell in case the slot
of the page is confirmed, or a cell whose slot ID is the smallest
that directly follows the page's slot id, in case the target slot
is not confirmed.
Using the starting cell as a trustable starting point (i.e. maintained
and provided by L1), and combined with the extra information stored in
the {!proof} type below, one can verify if a slot (and then a page of
that slot) is confirmed on L1 or not. *)
type inclusion_proof = history list
(** (See the documentation in the mli file to understand what we want to
prove in game refutation involving Dal and why.)
A Dal proof is an algebraic datatype with two cases, where we basically
prove that a Dal page is confirmed on L1 or not. Being 'not confirmed'
here includes the case where the slot's header is not published and the
case where the slot's header is published, but the endorsers didn't
confirm the availability of its data.
To produce a proof representation for a page (see function {!produce_proof_repr}
below), we assume given:
- [page_id], identifies the page;
- [slots_history], a current/recent cell of the slots history skip list.
Typically, it should be the skip list cell snapshotted when starting the
refutation game;
- [history_cache], a sufficiently large slots history cache, to navigate
back through the successive cells of the skip list. Typically,
the cache should at least contain the cell whose slot ID is [page_id.slot_id]
in case the page is confirmed, or the cell whose slot ID is immediately
after [page_id.slot_id] in case of an unconfirmed page. Indeed,
inclusion proofs encode paths through skip lists' cells where the head
is the reference/snapshot cell and the last element is the target slot
in or the nearest upper slot (w.r.t [page_id]'s slot id and to
skip list elements ordering) ;
- [page_info], that provides the page's information (the content and
the slot membership proof) for page_id. In case the page is supposed
to be confirmed, this argument should contain the page's content and
the proof that the page is part of the (confirmed) slot whose ID is
given in [page_id]. In case we want to show that the page is not confirmed,
the value [page_info] should be [None].
[dal_parameters] is used when verifying that/if the page is part of
the candidate slot (if any).
*)
type proof_repr =
| Page_confirmed of {
target_cell : history;
(** [target_cell] is a cell whose content contains the slot to
which the page belongs to. *)
inc_proof : inclusion_proof;
(** [inc_proof] is a (minimal) path in the skip list that proves
cells inclusion. The head of the list is the [slots_history]
provided to produce the proof. The last cell's content is
the slot containing the page identified by [page_id],
that is: [target_cell]. *)
page_data : Page.content;
(** [page_data] is the content of the page. *)
page_proof : Page.proof;
(** [page_proof] is the proof that the page whose content is
[page_data] is actually the [page_id.page_index]th page of
the slot stored in [target_cell] and identified by
page_id.slot_id. *)
} (** The case where the slot's page is confirmed/attested on L1. *)
| Page_unconfirmed of {
prev_cell : history;
(** [prev_cell] is the cell of the skip list containing a
(confirmed) slot, and whose ID is the biggest (w.r.t. to skip
list elements ordering), but smaller than [page_id.slot_id]. *)
next_cell_opt : history option;
(** [next_cell_opt] is the cell that immediately follows [prev_cell]
in the skip list, if [prev_cell] is not the latest element in
the list. Otherwise, it's set to [None]. *)
next_inc_proof : inclusion_proof;
(** [inc_proof] is a (minimal) path in the skip list that proves
cells inclusion. In case, [next_cell_opt] contains some cell
'next_cell', the head of the list is the [slots_history]
provided to produce the proof, and the last cell is
'next_cell'. In case [next_cell_opt] is [None], the list is
empty.
We maintain the following invariant in case the inclusion
proof is not empty:
```
(content next_cell).id > page_id.slot_id > (content prev_cell).id AND
hash prev_cell = back_pointer next_cell 0 AND
Some next_cell = next_cell_opt AND
head next_inc_proof = slots_history
```
Said differently, `next_cell` and `prev_cell` are two consecutive
cells of the skip list whose contents' IDs surround the page's
slot ID. Moreover, the head of the list should be equal to
the initial (snapshotted) slots_history skip list.
The case of an empty inclusion proof happens when the inputs
are such that: `page_id.slot_id > (content slots_history).id`.
The returned proof statement implies the following property in this case:
```
next_cell_opt = None AND prev_cell = slots_history
```
*)
}
(** The case where the slot's page doesn't exist or is not
confirmed on L1. *)
let proof_repr_encoding =
let open Data_encoding in
let case_page_confirmed =
case
~title:"confirmed dal page proof representation"
(Tag 0)
(obj5
(req "kind" (constant "confirmed"))
(req "target_cell" history_encoding)
(req "inc_proof" (list history_encoding))
(req "page_data" bytes)
(req "page_proof" Page.proof_encoding))
(function
| Page_confirmed {target_cell; inc_proof; page_data; page_proof} ->
Some ((), target_cell, inc_proof, page_data, page_proof)
| _ -> None)
(fun ((), target_cell, inc_proof, page_data, page_proof) ->
Page_confirmed {target_cell; inc_proof; page_data; page_proof})
and case_page_unconfirmed =
case
~title:"unconfirmed dal page proof representation"
(Tag 1)
(obj4
(req "kind" (constant "unconfirmed"))
(req "prev_cell" history_encoding)
(req "next_cell_opt" (option history_encoding))
(req "next_inc_proof" (list history_encoding)))
(function
| Page_unconfirmed {prev_cell; next_cell_opt; next_inc_proof} ->
Some ((), prev_cell, next_cell_opt, next_inc_proof)
| _ -> None)
(fun ((), prev_cell, next_cell_opt, next_inc_proof) ->
Page_unconfirmed {prev_cell; next_cell_opt; next_inc_proof})
in
union [case_page_confirmed; case_page_unconfirmed]
(** Proof's type is set to bytes and not a structural datatype because
when a proof appears in a tezos operation or in an rpc, a user can not
reasonably understand the proof, thus it eases the work of people decoding
the proof by only supporting bytes and not the whole structured proof. *)
type proof = bytes
(** DAL/FIXME: https://gitlab.com/tezos/tezos/-/issues/4084
DAL proof's encoding should be bounded *)
let proof_encoding = Data_encoding.bytes
type error += Dal_invalid_proof_serialization
let () =
register_error_kind
`Permanent
~id:"Dal_slot_repr.invalid_proof_serialization"
~title:"Dal invalid proof serialization"
~description:"Error occured during dal proof serialization"
Data_encoding.unit
(function Dal_invalid_proof_serialization -> Some () | _ -> None)
(fun () -> Dal_invalid_proof_serialization)
let serialize_proof proof =
match Data_encoding.Binary.to_bytes_opt proof_repr_encoding proof with
| None -> error Dal_invalid_proof_serialization
| Some serialized_proof -> ok serialized_proof
type error += Dal_invalid_proof_deserialization
let () =
register_error_kind
`Permanent
~id:"Dal_slot_repr.invalid_proof_deserialization"
~title:"Dal invalid proof deserialization"
~description:"Error occured during dal proof deserialization"
Data_encoding.unit
(function Dal_invalid_proof_deserialization -> Some () | _ -> None)
(fun () -> Dal_invalid_proof_deserialization)
let deserialize_proof proof =
match Data_encoding.Binary.of_bytes_opt proof_repr_encoding proof with
| None -> error Dal_invalid_proof_deserialization
| Some deserialized_proof -> ok deserialized_proof
let pp_inclusion_proof = Format.pp_print_list pp_history
let pp_history_opt = Format.pp_print_option pp_history
let pp_proof ~serialized fmt p =
if serialized then Format.pp_print_string fmt (Bytes.to_string p)
else
match deserialize_proof p with
| Error msg -> Error_monad.pp_trace fmt msg
| Ok proof -> (
match proof with
| Page_confirmed {target_cell; inc_proof; page_data; page_proof} ->
Format.fprintf
fmt
"Page_confirmed (target_cell=%a, data=%s,@ \
inc_proof:[size=%d |@ path=%a]@ page_proof:%a)"
pp_history
target_cell
(Bytes.to_string page_data)
(List.length inc_proof)
pp_inclusion_proof
inc_proof
Page.pp_proof
page_proof
| Page_unconfirmed {prev_cell; next_cell_opt; next_inc_proof} ->
Format.fprintf
fmt
"Page_unconfirmed (prev_cell = %a | next_cell = %a | \
prev_inc_proof:[size=%d@ | path=%a])"
pp_history
prev_cell
pp_history_opt
next_cell_opt
(List.length next_inc_proof)
pp_inclusion_proof
next_inc_proof)
type error +=
| Dal_proof_error of string
| Unexpected_page_size of {expected_size : int; page_size : int}
let () =
let open Data_encoding in
register_error_kind
`Permanent
~id:"dal_slot_repr.slots_history.dal_proof_error"
~title:"Dal proof error"
~description:"Error occurred during Dal proof production or validation"
~pp:(fun ppf e -> Format.fprintf ppf "Dal proof error: %s" e)
(obj1 (req "error" string))
(function Dal_proof_error e -> Some e | _ -> None)
(fun e -> Dal_proof_error e)
let () =
let open Data_encoding in
register_error_kind
`Permanent
~id:"dal_slot_repr.slots_history.unexpected_page_size"
~title:"Unexpected page size"
~description:
"The size of the given page content doesn't match the expected one."
~pp:(fun ppf (expected, size) ->
Format.fprintf
ppf
"The size of a Dal page is expected to be %d bytes. The given one \
has %d"
expected
size)
(obj2 (req "expected_size" int16) (req "page_size" int16))
(function
| Unexpected_page_size {expected_size; page_size} ->
Some (expected_size, page_size)
| _ -> None)
(fun (expected_size, page_size) ->
Unexpected_page_size {expected_size; page_size})
let dal_proof_error reason = Dal_proof_error reason
let proof_error reason = error @@ dal_proof_error reason
let check_page_proof dal_params proof data pid commitment =
let open Tzresult_syntax in
let* dal =
match Dal.make dal_params with
| Ok dal -> return dal
| Error (`Fail s) -> proof_error s
in
let page = {Dal.content = data; index = pid.Page.page_index} in
let fail_with_error_msg what =
Format.kasprintf proof_error "%s (page id=%a)." what Page.pp pid
in
match Dal.verify_page dal commitment page proof with
| Ok true -> return ()
| Ok false ->
fail_with_error_msg
"Wrong page content for the given page index and slot commitment"
| Error `Segment_index_out_of_range ->
fail_with_error_msg "Segment_index_out_of_range"
| Error `Page_length_mismatch ->
fail
@@ Unexpected_page_size
{
expected_size = dal_params.page_size;
page_size = Bytes.length data;
}
let produce_proof_repr dal_params page_id ~page_info slots_hist hist_cache =
let open Tzresult_syntax in
let Page.{slot_id; page_index = _} = page_id in
let deref ptr = History_cache.find ptr hist_cache in
(* We search for a slot whose ID is equal to target_id. *)
let search_result =
Skip_list.search ~deref ~target_id:slot_id ~cell:slots_hist
in
match (page_info, search_result.Skip_list.last_cell) with
| _, Deref_returned_none ->
proof_error
"Skip_list.search returned 'Deref_returned_none': Slots history \
cache is ill-formed or has too few entries."
| _, No_exact_or_lower_ptr ->
proof_error
"Skip_list.search returned 'No_exact_or_lower_ptr', while it is \
initialized with a min elt (slot zero)."
| Some (page_data, page_proof), Found target_cell ->
(* The slot to which the page is supposed to belong is found. *)
let Header.{id; commitment} = Skip_list.content target_cell in
(* We check that the slot is not the dummy slot. *)
let* () =
error_when
Compare.Int.(Header.compare_slot_id id Header.zero.id = 0)
(dal_proof_error
"Skip_list.search returned 'Found <zero_slot>': No existence \
proof should be constructed with the slot zero.")
in
let* () =
check_page_proof dal_params page_proof page_data page_id commitment
in
let inc_proof = List.rev search_result.Skip_list.rev_path in
let* () =
error_when
(List.is_empty inc_proof)
(dal_proof_error "The inclusion proof cannot be empty")
in
(* All checks succeeded. We return a `Page_confirmed` proof. *)
return
( Page_confirmed {inc_proof; target_cell; page_data; page_proof},
Some page_data )
| None, Nearest {lower = prev_cell; upper = next_cell_opt} ->
(* There is no previously confirmed slot in the skip list whose ID
corresponds to the {published_level; slot_index} information
given in [page_id]. But, `search` returned a skip list [prev_cell]
(and possibly [next_cell_opt]) such that:
- the ID of [prev_cell]'s slot is the biggest immediately smaller than
the page's information {published_level; slot_index}
- if not equal to [None], the ID of [next_cell_opt]'s slot is the smallest
immediately bigger than the page's slot id `slot_id`.
- if [next_cell_opt] is [None] then, [prev_cell] should be equal to
the given history_proof cell. *)
let* next_inc_proof =
match search_result.Skip_list.rev_path with
| [] -> assert false (* Not reachable *)
| prev :: rev_next_inc_proof ->
let* () =
error_unless
(equal_history prev prev_cell)
(dal_proof_error
"Internal error: search's Nearest result is \
inconsistent.")
in
return @@ List.rev rev_next_inc_proof
in
return
(Page_unconfirmed {prev_cell; next_cell_opt; next_inc_proof}, None)
| None, Found _ ->
proof_error
"The page ID's slot is confirmed, but no page content and proof \
are provided."
| Some _, Nearest _ ->
proof_error
"The page ID's slot is not confirmed, but page content and proof \
are provided."
let produce_proof dal_params page_id ~page_info slots_hist hist_cache =
let open Tzresult_syntax in
let* proof_repr, page_data =
produce_proof_repr dal_params page_id ~page_info slots_hist hist_cache
in
let* serialized_proof = serialize_proof proof_repr in
return (serialized_proof, page_data)
(* Given a starting cell [snapshot] and a (final) [target], this function
checks that the provided [inc_proof] encodes a minimal path from
[snapshot] to [target]. *)
let verify_inclusion_proof inc_proof ~src:snapshot ~dest:target =
let assoc = List.map (fun c -> (hash_skip_list_cell c, c)) inc_proof in
let path = List.split assoc |> fst in
let deref =
let open Map.Make (Pointer_hash) in
let map = of_seq (List.to_seq assoc) in
fun ptr -> find_opt ptr map
in
let snapshot_ptr = hash_skip_list_cell snapshot in
let target_ptr = hash_skip_list_cell target in
error_unless
(Skip_list.valid_back_path
~equal_ptr:Pointer_hash.equal
~deref
~cell_ptr:snapshot_ptr
~target_ptr
path)
(dal_proof_error "verify_proof_repr: invalid inclusion Dal proof.")
let verify_proof_repr dal_params page_id snapshot proof =
let open Tzresult_syntax in
let Page.{slot_id; page_index = _} = page_id in
match proof with
| Page_confirmed {target_cell; page_data; page_proof; inc_proof} ->
(* If the page is supposed to be confirmed, the last cell in
[inc_proof] should store the slot of the page. *)
let Header.{id; commitment} = Skip_list.content target_cell in
let* () =
error_when
Compare.Int.(Header.compare_slot_id id Header.zero.id = 0)
(dal_proof_error
"verify_proof_repr: cannot construct a confirmation page \
proof with 'zero' as target slot.")
in
let* () =
verify_inclusion_proof inc_proof ~src:snapshot ~dest:target_cell
in
(* We check that the page indeed belongs to the target slot at the
given page index. *)
let* () =
check_page_proof dal_params page_proof page_data page_id commitment
in
(* If all checks succeed, we return the data/content of the page. *)
return_some page_data
| Page_unconfirmed {prev_cell; next_cell_opt; next_inc_proof} ->
(* The page's slot is supposed to be unconfirmed. *)
let ( < ) a b = Compare.Int.(Header.compare_slot_id a b < 0) in
(* We retrieve the last cell of the inclusion proof to be able to
call {!verify_inclusion_proof}. We also do some well-formedness on
the shape of the inclusion proof (see the case [Page_unconfirmed]
of type {!proof}). *)
let* () =
match next_cell_opt with
| None ->
let* () =
error_unless
(List.is_empty next_inc_proof)
(dal_proof_error
"verify_proof_repr: invalid next_inc_proof")
in
(* In case the inclusion proof has no elements, we check that:
- the prev_cell slot's id is smaller than the unconfirmed slot's ID
- the snapshot is equal to the [prev_cell] skip list.
This way, and since the skip list is sorted wrt.
{!compare_slot_id}, we are sure that the skip list whose head
is [snapshot] = [prev_cell] cannot contain a slot whose ID is
[slot_id]. *)
error_unless
((Skip_list.content prev_cell).id < slot_id
&& equal_history snapshot prev_cell)
(dal_proof_error "verify_proof_repr: invalid next_inc_proof")
| Some next_cell ->
(* In case the inclusion proof has at least one element,
we check that:
- the [prev_cell] slot's id is smaller than [slot_id]
- the [next_cell] slot's id is greater than [slot_id]
- the [next_cell] cell is a direct successor of the
[prev_cell] cell.
- the [next_cell] cell is a predecessor of [snapshot]
Since the skip list is sorted wrt. {!compare_slot_id}, and
if the call to {!verify_inclusion_proof} succeeds, we are
sure that the skip list whose head is [snapshot] cannot
contain a slot whose ID is [slot_id]. *)
let* () =
error_unless
((Skip_list.content prev_cell).id < slot_id
&& slot_id < (Skip_list.content next_cell).id
&&
let prev_cell_pointer =
Skip_list.back_pointer next_cell 0
in
match prev_cell_pointer with
| None -> false
| Some prev_ptr ->
Pointer_hash.equal
prev_ptr
(hash_skip_list_cell prev_cell))
(dal_proof_error
"verify_proof_repr: invalid next_inc_proof")
in
verify_inclusion_proof
next_inc_proof
~src:snapshot
~dest:next_cell
in
return_none
let verify_proof dal_params page_id snapshot serialized_proof =
let open Tzresult_syntax in
let* proof_repr = deserialize_proof serialized_proof in
verify_proof_repr dal_params page_id snapshot proof_repr
module Internal_for_tests = struct
let content = Skip_list.content
let proof_statement_is serialized_proof expected =
match deserialize_proof serialized_proof with
| Error _ -> false
| Ok proof -> (
match (expected, proof) with
| `Confirmed, Page_confirmed _ | `Unconfirmed, Page_unconfirmed _ ->
true
| _ -> false)
end
end
include V1
end