| Age | Commit message (Collapse) | Author |
|---|
|
Dialyzer issued two new warnings when the 'catch' was removed in
the previous commit.
|
|
The last clause in asn1ct_gen:type/1 does a catched call to type2/1.
If the type2/1 fails {notype,X} is returned.
Since the body of type2/1 essentially is:
case lists:member(X, [...]) of
true ->
{primitive,bif};
false ->
case lists:member(X, [...]) of
true ->
{constructed,bif};
false ->
{undefined,user}
end
end
there is no way that type2/1 can fail. Therefore, we can eliminate
the catch and put the body of type2/1 into the last clause of
type/1. We can also eliminate the code in the callers of type/1
that match {notype,X}.
|
|
|
|
Stop export functions that are not called from outside their
module. If the functions are not used at all, remove the functions
too.
The unused exports were found by running:
xref:start(s).
xref:add_application(s, code:lib_dir(asn1)).
io:format("~p\n", [xref:analyze(s, exports_not_used)]).
|
|
ObjectDescriptor, UTCTime, and GeneralizedTime are not special and
can be handled in the same way as all other restricted string types.
|
|
This slight optimization will also eliminate some Dialyzer warnings.
|
|
This change brings down the execution time on my computer for the
entire asn1 test suite from about 340 seconds to 310 seconds.
|
|
asn1ct_gen:emit/1 used to make one call to io:put_chars/2 for each
part of the term passed emit/1. By collecting all output into one
iolist for each call emit/1 the time for running the entire asn1
test suite is reduced from about 460 seconds to 340 seconds on my
computer.
|
|
Almost always, encode_open_type/1 is called with the return value
from complete/1, which always is a binary. In the rare situation
that encode_open_type/1 is called directly with data from the
user application, call iolist_to_binary/1 before calling
encode_open_type/1.
|
|
Hiding the details of decoding an external type will facilitate
changing the calling convention in a future commit.
|
|
The code does record operations one step at the time and checks
for conditions that cannot happen.
|
|
asn1ct_check has translated all occurrences of 'ANY' to 'ASN1_OPEN_TYPE'.
|
|
|
|
The encoder wrongly assumed that a known multiplier string (such as
IA5String) encoded as exactly 16 bits did not need to be aligned to an
octet boundary. X.691 (07/2002) 27.5.7 says that it does. Since an
OCTET STRING encoded to 16 bits (two octets) should not be aligned to
an octet boundary, that means that asnct_imm:dec_string() needs an
additional parameter to determine whether a string of a given length
needs to be aligned.
Furthermore, there is another subtle rule difference: An OCTET STRING
which does not have fixed length is always aligned (in PER), but
a known multiplier string is aligned if its upper bound is greater
than or equal to 16.
In encoding, make sure that short known multiplier strings and
OCTET STRINGs with extensible sizes are not aligned when they are
below the appropriate limit.
|
|
Given the type:
S ::= IA5String (SIZE (5, ...))
attempting to encode (to PER/UPER) a string shorter than 5 characters
would fail. Similarly, attempting to decode such string in the BER
format would fail.
In the case of BER, we can do no range checks if the size constraint
is extensible.
|
|
Consider a type with a size constraint with an extension marker
such as:
S ::= OCTET STRING (SIZE (0..10, ...))
For a length outside the root range (e.g. 42), the PER/UPER encoder
will encode the length field in the same way as it would the type
INTEGER (0..MAX) (i.e., as semi-constrained whole number), while the
decoder would decode the length in the same way as length field
without any constraint.
Clearly, either the encoder or the decoder is wrong. But which one?
Dubuisson's [1] book (page 442) says that the length should be encoded
as a semi-constrained whole number if the length is outside the root
range.
The X.691 standard document [2] also says (e.g. in 15.11) that length
fields should be a semi-constrained number, but gives a reference
to section gives a reference to section 10.9, "General rules for encoding
a length determinant", and not to to 10.7, "Encoding of a
semi-constrained whole number".
Reading the standard that way should imply that a length outside the
root range should be encoded in the same way as an unconstrained
length, and that the decoder does the right thing.
Further support for that interpretation:
- Larmouth's book [3], page 303.
- The ASN.1 playground. [4]
References:
[1] http://www.oss.com/asn1/resources/books-whitepapers-pubs/dubuisson-asn1-book.PDF
[2] http://www.itu.int/ITU-T/studygroups/com17/languages/X.691-0207.pdf
[3] http://www.oss.com/asn1/resources/books-whitepapers-pubs/larmouth-asn1-book.pdf
[4] http://asn1-playground.oss.com
|
|
For what seems to be historical reasons,
asn1_rtt_per:encode_octet_string/3 has an ExtensionMarker argument
that is no longer used. The extension mark (if any) is included in the
constraints argument.
|
|
The compiler would crash when given code such as the following:
Type ::= INTEGER (lower<..<upper)
lower INTEGER ::= 0
lower INTEGER ::= 42
|
|
The record #constraint{} is almost unused outside of the parser
except for two places in asn1ct_check.
The only correct usage of the record is in instance_of_constraints/2.
Eliminate that usage by updating the parser to pass that constraint
in the same way as all other constraints.
In check_integer_range/2, the record is used incorrectly. A
constraint for an integer will never be a list of #constraint{}
records. Therefore, the list comprehension will always produce
an empty list, and check_constr/2 will not actually check anything
(which is kind of lucky, since the 'ValueRange' range constraint
is incorrectly written - the lower and upper bounds should be in
a tuple).
For now, we will not attempt to actually start validating integer
ranges. Firstly (obviously) we will need to be sure that we
correctly handles all forms of constraints, and secondly we will
need to consider whether we need to produce a warning rather than an
error for compatibility reasons.
|
|
|
|
|
|
The record #typereference{} is only used internally within
the asn1ct_parser2 module (the parser translates it to
an #'Externaltypereference{} record).
|
|
The only code that is really different between the PER
and UPER backends is encoding of primitive types.
|
|
|
|
|
|
The wrong backend was used.
|
|
|
|
|
|
Improve readability and maintainability.
|
|
Optimize the decoding of CHOICE. Most important is to inline decoding
of the extension bit (if present) and decoding of the choice index
to give the BEAM compiler more opportunities for optimization.
We will also change the structure of the generated code. The current
code uses a flattened case for both the root and extension alternatives:
case Choice + NumRootChoices * Ext of
%% Root alternatives.
0 - ...;
:
LastRootAlternative -> ...;
%% Extension alternatives.
LastRootAlternative+1 -> ...;
:
%% Unknown extension.
_ -> ...;
end
We will instead generate nested cases:
case Ext of
0 ->
case Choice of
%% Root alternatives.
0 - ...;
:
LastRootAlternative -> ...
end;
1 ->
%% Extension alternatives.
<Decode the open type here>
case Choice of
0 -> ...;
:
LastExtensionAlternative -> ...;
%% Unknown extension.
_ -> ...;
end
end
Nested cases should be slightly faster. For decoding of the extensions,
it also makes it possible to hoist the decoding of the open type up
from each case to before the case switching on the extension index,
thus reducing the size of the generated code.
We will also do another change to the structure. Currently, the
big flat clase is wrapped in code that repackages the return values:
{Alt,{Value,RemainingEncodedData}} =
case Choice + NumRootChoices * Ext of
:
end,
{{Value,Alt},RemainingEncodedData}.
We still need to do the repackaging, but we can push it down to
the case arm for decoding each alternative. In many cases, that
will give the BEAM compiler the opportunity to avoid building the
temporary tuples.
|
|
We will need more explicit control of decoding of open types for
CHOICEs, so refactor the gen_dec_line() and gen_dec_line_imm()
to break out the decoding of the open types.
Note that gen_dec_line_special() will not generate correct code
if Ext =/= noext; thus, we can eliminate the Ext parameter.
|
|
gen_dec_choice2() is unnecessarily complicated. Code is duplicated
merely to avoid outputting a ";" separator after the last item,
and there is code to skip extensionmarks that are never actually
present.
While at it, eliminate the the wrap_gen_dec_line() function which is
not needed. It puts a {component_type,C} key in the process
dictionary, but that is actually never used. Besides that, it only
shuffles the argument.
|
|
gen_enc_choice2() duplicates code merely to avoid outputting a ";"
separator after the last item.
|
|
Ensure that asn1ct_name:start/0 will call asn1ct_name:clear/0 if the
name server process is already running so that we can be sure that
the variables are cleared.
|
|
active/1 always returned 'true' (curr(Var) can never return 'nil',
except if Var =:= 'nil'). Therefore, we can eliminate its use and
remove it.
|
|
clean/1 is used quite lot; while not a bottleneck it does seem
unnecessary to stop and start the name server just to clear the
variables.
Since the change to clean/1 makes stop/1 unused, we can remove it.
|
|
Since new/1 does not return any useful value and cannot fail, there
is no reason to wait for a response from the name server process.
|
|
|
|
Since we no longer need to support push/1 and pop/1, we can simplify
the data structures used for keeping track of the variables.
Each entry in the list can be simply {Var,integer()} instead of
{Var,[integer()]}.
While at it, eliminate calls to the obsolete lists:keysearch/3
function, don't use integer_to_list/0 within a call to lists:concat/1,
and the catch all clause in get_prev/2 and get_next/2.
|
|
asn1ct_name:delete/1 is used from asn1ct_gen_ber_bin_v2:add_removed_bytes/1,
but that function is not used.
|
|
|
|
* bjorn/asn1/fix-lost-extension-mark/OTP-10995:
Prevent loss of objects after the extension marker
Don't lose the extension mark for object set parameters
|
|
In an object set with a single root object, objects after
the extension marker would be lost.
|
|
When an object set is an actual parameter, the extension marker
for the object set could get lost.
|
|
If the encoding is empty (i.e. if a top-level type is single-valued
and therefore not encoding), the result should be a single zero
byte.
|
|
An ENUMERATED with as single value is not encoded. The decoder
incorrectly assumed that it was encoded in one bit.
|
|
|
|
Also teach otp_SUITE:call_to_size/1 that calls to size/1 is no
longer allowed within the asn1 application.
|
|
It is recommended to use byte_size/1 or tuple_size/1 instead of
size/1.
|
|
|
