Ημερήσια αρχεία: 3 Ιανουαρίου 2017


asn1c: Generating code using ‘Automatic Tags’ and negative value as default value creates invalid function names

The following post is for the https://lionet.info/asn1c/ (repository: https://github.com/vlm/asn1c/)

When compiling the following ASN.1 data structure

GeographyModule DEFINITIONS AUTOMATIC TAGS ::= BEGIN

    Coordinates ::= SEQUENCE
    {
        -- latitude from -90 till 90 degrees --
        latitude INTEGER(-9000000..9000000) DEFAULT -8000000,
        -- longitude from -180 till 179.99999 degrees, worst precision 1.1132m at equator --
        longitude INTEGER(-18000000..17999999) DEFAULT -12000000
    }

END

the use of both the AUTOMATIC TAGS option and the use of a negative value -8000000 in the position of the default value causes asn1c to create invalid function names in the Coordinates object.

For example, the above ASN.1 syntax will produce the following invalid function name int asn_DFL_2_set_-800000(int set_value, void **sptr).

Compilation command for asn1c

From folder asn1c_gps/asn1 we used the following command:

/home/developer/asn1c/asn1c/asn1c -pdu=auto -S /home/developer/asn1c/skeletons/ -fcompound-names -gen-PER ../geography.asn1

Version of asn1c

'ASN.1 Compiler, v0.9.28'

Example

Full example code demonstrating the bug can be found here ([download id=”2544″]).

If you want to use the code and see that all other operations are fine, replace _- with _minus_ in the file Coordinates.c and the code will become valid and usable.

After you perform the above change, you can use the code in main.cpp to see the our cycle of execution that encodes and decodes an object.


asn1c: Decoding an OCTET STRING with lower bound limit on its size fails for uper_decode()

The following post is for the https://lionet.info/asn1c/ (repository: https://github.com/vlm/asn1c/)

Hello guys,

I’ve noticed that when I set a lower bound limit on the size of an octet string, it fails to decode it.
To reproduce this scenario I created a small but full example that is located here([download id=”2539″]).

The example([download id=”2539″]) is an application that uses the code generated by asn1c and has the following behavior:

  1. It will read a name of a file from the command line
  2. read the file to memory
  3. convert it to an octet string using OCTET_STRING_fromBuf()
  4. encode it to an ASN.1 structure using uper_encode_to_new_buffer(), after asn_check_constraints() succeeds
  5. save the encoded data to a file for debugging (same folder as the original file)
  6. decode the buffer from memory using uper_decode()
  7. save the decoded data to a file (same folder as the original file)

Methodology

To create/view the bug use this ASN1 data structure as input to the asn1c compiler:

ImagesModule DEFINITIONS ::= BEGIN

 Image ::= SEQUENCE
 {
  data OCTET STRING SIZE (40..81920)
 }

END

To hide the bug, use:

ImagesModule DEFINITIONS ::= BEGIN

 Image ::= SEQUENCE
 {
  data OCTET STRING SIZE (0..81920)
 }

END

The only difference between the two versions is the use of a lower limit constraint on the size of the OCTET string.

Compilation command for asn1c

From folder asn1c_image/asn1 we used the following command:

/home/developer/asn1c/asn1c/asn1c -pdu=auto -S /home/developer/asn1c/skeletons/ -fcompound-names -gen-PER ../images.asn1

Version of asn1c

'ASN.1 Compiler, v0.9.28'

Samples

Inside the archive, there are two files [test_01.png, bad_data.bin].

  • test_01.png is larger than 80K so it should always fail.
  • bad_data.bin fails only when there is a lower bound limit on the size

[download id=”2539″]