| Age | Commit message (Collapse) | Author |
|---|
|
|
|
When an invalid value is encountered when trying to decode an
ENUMERATED, the exception can be quite verbose. Here is an actual
example from NBAP-PDU-Contents:
_ -> exit({error,{asn1,{decode_enumerated,{V4@V3,[chCode1div1,chCode2div1,chCode2div2,chCode4div1,chCode4div2,chCode4div3,chCode4div4,chCode8div1,chCode8div2,chCode8div3,chCode8div4,chCode8div5,chCode8div6,chCode8div7,chCode8div8,chCode16div1,chCode16div2,chCode16div3,chCode16div4,chCode16div5,chCode16div6,chCode16div7,chCode16div8,chCode16div9,chCode16div10,chCode16div11,chCode16div12,chCode16div13,chCode16div14,chCode16div15,chCode16div16]}}}})
Listing the possible values for an ENUMERATED when decoding fails is
not helpful and increases the code size (it would have made somewhat
more sense to list the possible values if *encoding* failed).
|
|
To make it easier to find your way in the generated code,
add a comment about each attribute in the generated code
for SEQUENCE and SET.
|
|
Generate slightly better error reasons when encoding of INTEGER,
ENUMERATED, or BOOLEAN fails.
|
|
When the 'maps' option is given, the SEQUENCE and SET types are
represented as maps instead of as records. Optional and default values
must be not be given as asn1_NOVALUE or asn1_DEFAULT in a map passed
to the M:encode/2 function; they must be omitted from the
map. Similarly, when decoding missing values will be omitted from the
map.
No .hrl files will be generated when the 'maps' options is used.
That means values in an ASN.1 module must be retrieved by calling the
appropriate function in generated module.
Since we one day hope to get rid of the options 'compact_bit_string',
'legacy_bit_string', and 'legacy_erlang_types', we will not allow them
to be combined with the 'maps' option.
|
|
As a preparation for supporting maps in a future commit, refactor
the functions for encoding optional values.
|
|
|
|
When a constrained INTEGER has more than 16536 values and named
values, the compiler would crash when compiling to the PER
format. Example:
Longitude ::= INTEGER {
oneMicrodegreeEast(10),
oneMicrodegreeWest(-10),
unavailable(1800000001)
} (-1799999999..1800000001)
Reported-by: Ingars
|
|
|
|
|
|
The BEAM compiler now warns more aggressively for expressions whose
values are not used. For example, the generated Erlang code for
the following ASN.1 code will now cause a warning:
Seq ::= SEQUENCE {
...,
os OCTET STRING (SIZE (17000..30000))
}
The generated Erlang code looks similar to the following:
Enc9@bin = iolist_to_binary(Enc9@output),
Enc9@len = byte_size(Enc9@bin),
[align|encode_fragmented(Enc9@bin, 8)]
The variable Enc9@len is not used and the BEAM compiler will
complain that the value returned from the byte_size/1 call is
not used.
Improve the optimization of the intermediate code to eliminate
the assignment to the unused variable.
|
|
* oliv3/math_log2/OTP-12411:
Add math:log2/1
|
|
|
|
The internal representation for constraints (and object sets)
as produced by the parser was awkward, making further processing
convoluted. Here follows some examples of the old representation
for INTEGER constraints.
The constraint 1..2 is represented as:
{'ValueRange',{1,2}}
If we extend the constraint like this:
1..2, ...,
or like this:
1..2, ..., 3
the representation would be:
{{'ValueRange',{1,2}},[]}
and
{{'ValueRange',{1,2}},{'SingleValue',3}}
respectively. Note that the pattern {A,B} will match all these
constraints.
When combining constraints using set operators:
1..2 | 3..4 ^ 5..6
the representation will no longer be a tuple but a list:
[{'ValueRange',{1..2}} union
{'ValueRange',{3..4}} intersection
{'ValueRange',{5..6}}]
The parse has full knowledge of the operator precedence; unfortunately,
the following pass (asn1ct_check) must also have the same knowledge
in order to correctly evaluate the constraints.
If we would change the order of the evaulation with round brackets:
(1..2 | 3..4) ^ 5..6
there would be a nested listed in the representation:
[[{'ValueRange',{1..2}} union {'ValueRange',{3..4}}]
intersection {'ValueRange',{5..6}}]
We will change the representation to make it more explicit.
At the outer level, a constraint is always represented as
{element_set,Root,Extension}
Extension will be 'none' if there is no extension, and 'empty' if
there is an empty extension. Root may also be 'empty' in an object set
if there are no objects in the root. Thus the constraints:
1..2
1..2, ...
1..2, ..., 3
will be represented as:
{element_set,{'ValueRange',{1,2}},none}
{element_set,{'ValueRange',{1,2}},empty}
{element_set,{'ValueRange',{1,2}},{'SingleValue',3}}
We will change the set operators too. This constraint:
1..2 | 3..4 ^ 5..6
will be represented as:
{element_set,
{union,
{'ValueRange',{1,2}},
{intersection,
{'ValueRange',{3,4}},
{'ValueRange',{5,6}}},
none}}
which is trivial to understand and evaluate. Similarly:
(1..2 | 3..4) ^ 5..6
will be represented as:
{element_set,
{intersection,
{union,{'ValueRange',{1,2}},{'ValueRange',{3,4}}},
{'ValueRange',{5,6}}},
none}
|
|
|
|
The code generator would crash.
|
|
If a named BIT STRING has a lower size bound of 0, treat it the same
way as if there was no constraint for the purposes of trailing zero
bits.
That change will eliminate a dialyzer warning.
|
|
The encoder for the following type would generate a dialyzer warning:
Ns ::= NumericString (FROM ("0"|"1"|"2"|"3"|"4"|"5"|"6"|"7"|"8"|"9"))
Optimize this case to make it slightly faster as well as eliminate
the dialyzer warning.
|
|
|
|
|
|
complete/1 is used when encoding open types (as well as in the
encode/2 function in a generated module).
The use of complete/1 for encoding open types used to be optimized
in two different places.
One place was in the alignment optimization pass, where we attempted
to replace the call to complete/1 with a call to iolist_to_binary/1.
That optimization was taken out in a previous commit that introduced
the {list,_,_} intermediate instruction.
The other place was when creating the intermediate representation
for the encoding of the open type. When attempting to wrap primitive
types in an open type, we would attempt to optimize the encoding of
the length decscriptor. We will remove that optimization in this
commit.
Since the previous two optimizations did not optimize encoding of
open types as much as we would want, we will introduce a new
optimization in a separate pass that will go further than the
previous optimizations.
|
|
The {list,List,Dst} instruction gives us as general way to
capture the building of something into a variable. That will make
inlining of intermediate code much easier.
It also allows us to eliminate the versions of the apply, call_gen,
and cond instructions that takes a target variable.
Also remove the optimization in the alignment optimization pass
that attempts to replace calls to complete/1 with calls to
iolist_to_binary/1. That optimization will not work anymore without
rewriting, so we will remove it in this commit and introcude a
more powerful optimization in a future commit.
|
|
Instead of generating:
{assign,Dst,"element(2, Val)"}
generate:
{call,erlang,element,[2,{var,"Val"}],Dst}
The latter expression is easier to understand since there is no
need to parse a string which may contain an arbitrary expression.
While at it, also discontinue the practice to treat "naked"
atoms as variables. A variable must always be given as {var,String}.
|
|
The {assign,Dst,Src} instruction is difficult to cope with when
doing advanced optimizations, since its source argument is a string
which may contain any expression.
Instead of changing how {assign,_,_} works, we will introduce new
instructions that can be used instead of {assign,_}, and remove
{assign,_,_} in a later commit when it is no longer used.
The first new instruction we will introduce is:
{set,{var,Src},{var,Dst}}
It is useful for common sub-expression elemination among other things.
For the moment, we will only allow a variable as a source argument,
but we could extend it in the future to allow constants as well.
|
|
To facilitate inlining of apply calls in the intermediate format.
|
|
It will greatly facilitate further optimizations if we include the
intermediate code (if available) in the call_gen tuple.
|
|
|
|
Make sure that we don't construct:
{cons,{integer,I},{cons,{binary,B},T}} - OR -
{cons,{binary,B},{cons,{integer,I},T}}
but:
{cons,{binary,[{put_bits,I,8,[1]}|B]},T} - OR -
{cons,{binary,B++[{put_bits,I,8,[1]}]},T}
|
|
* bjorn/asn1/fix-integer-constraint/OTP-11504:
PER/UPER: Handle a range in the extension part of the constraint
|
|
Constraints such as:
INTEGER (1..10, ..., 11..20)
would fail to compile. Make sure it is properly ignored.
|
|
* bjorn/asn1/fix-extensible-single-values/OTP-11415:
PER/UPER: Correct encoding for single-value extensible constraints
asn1ct_value: Handle named INTEGERs with constraints
|
|
For DER/PER/UPER, a value equal to the DEFAULT is not supposed to
be encoded.
BIT STRINGs values can be represented as Erlang terms in four
different ways: as an integer, as a list of zeroes and ones,
as a {Unused,Binary} tuple, or as an Erlang bitstring.
When encoding a BIT STRING, only certain representations of
BIT STRINGs values were recognized. All representations must
be recognized.
When decoding a DEFAULT value for a BIT STRING, the actual value
given in the decoding would be either an integer or a list
of zeroes and one (depending on how the literal was written in
the specification). We expect that the default value should be
in the same representation as any other BIT STRING value (i.e.
by default an Erlang bitstring, or a list if the 'legacy_bitstring'
option has been given, or as compact bitstring if 'compact_bitstring'
has been given).
|
|
An extensible constraint which is a union of single values, such as:
INTEGER (1|17, ...)
would be incorrectly encoded.
|
|
|
|
The generated code for table constraints has several problems:
* For each object set, a function for getting an encoding or decoding
fun is generated, regardless of whether it is actually used. In many
specifications, the object set actually used is the union of several
other object sets. That means that the code can become a lot bulkier
than it would need to be.
* The funs are not necessary. The funs just add to the code bloat
and generate more unnecessary garbage at run-time. Also, one of
the arguments of the fun is the name of the field in the class which
is known at compile-time, and the fun for decoding has unused arguments.
How to fix the problems:
At each call site where an open type should be encoded/decoded, call a
specific generated function specialized for the actual object set and
the name of the field in the class. When generating the specialized
functions, make sure that we re-use a previously generated function if
possible.
|
|
There are some minor incompatibilities for BIT STRING:
{bit,Position} is now only only supported for a named
BIT STRING type.
Values longer than the maximum size for the BIT STRING type
would be truncated silently - they now cause an exception.
|
|
Break out the the rules for determining whether a string should
be in aligned so that it can be reused for encoding.
|
|
|
|
A semi-constrained INTEGER with a non-zero lower bound would be
incorrectly decoded. This bug was introduced in R16.
|
|
|
|
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.
|
|
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.
|
|
An ENUMERATED with as single value is not encoded. The decoder
incorrectly assumed that it was encoded in one bit.
|
|
|
|
|
|
|
|
|
|
Dialyzer will warn for default clauses that cannot possible match.
|
|
Refactor decoding of BIT STRINGs so that the run-time code does
not spend testing conditions that are known already at compile-time
(which wastes time and produces unnecessary Dialyzer warnings).
There are three ways to decode BIT STRINGs:
1) To a list of bit names (and {bit,Position} for unnamed positions)
if the BIT STRING type has any bit names.
2a) To a list of ones and zeros if there are no named bits, and the
compact_bit_string option was NOT given.
2b) To a {Unused,Bin} tuple if there are no named bits compact_bit_string
option WAS given.
Structure the decoding functions in the same way.
|
|
The template modules (asn1rtt_*.erl) are based on the existing
run-time modules, but with some simplifications and improvements,
for example:
The run-time functions for BER encoding took a Constraint argument which
was not used. It has been eliminated, along with the unused StringType
argument for the encode_restricted_string function.
The Range argument for decode_enumerated() has been dropped since it
was not used.
|
