APER segfault

[johvik]

When using APER in a CHOICE it segfaults on this line (ct may be NULL):

asn1c/skeletons/constr_CHOICE.c

Line 1169 in a29f447

if(aper_put_nsnnwn(po, ct->range_bits, present - specs->ext_start))

For example when encoding ChoiceExtended1 in this asn definition:

Enum ::= ENUMERATED { one, two, three, four }

ChoiceExtended1 ::= CHOICE {                                                              
    a Enum,                                                                               
    ...,                                                                                  
    b Enum                                                                                
}

[velichkov]

Hi @johvik,

Thanks for reporting this problem, could you also provide also xer or aper files for the example asn1 definition and all commands you need to execute to reproduce this segfault.

Do you have plans to work on a fix?

[mouse07410]

Let me second this request. @johvik please provide a xer file that demonstrates the problem.

[johvik]

Here is a sample that reproduces the issue:

#include <ChoiceExtended1.h>
#include <asn_application.h>

int
main()
{
    ChoiceExtended1_t c;
    memset(&c, 0, sizeof(c));
    c.present = ChoiceExtended1_PR_b;
    c.choice.b = Enum_one;

    uint8_t buffer[500];
    asn_encode_to_buffer(NULL,
                         ATS_ALIGNED_BASIC_PER,
                         &asn_DEF_ChoiceExtended1,
                         &c,
                         &buffer,
                         sizeof(buffer));
}

As a temporary workaround I'm using aper_put_nsnnwn from vlm#125. @velichkov I'm afraid I do not have time to make a permanent fix :/

[mouse07410]

@johvik thank you for the code example. But it would really help if you could provide XER-encoded data file for the ASN.1 that causes this APER problem.

[johvik]

Do you mean this?

<ChoiceExtended1>
    <b><one/></b>
</ChoiceExtended1>

[mouse07410]

Yes! Thank you!

[mouse07410]

I have a preliminary solution, will push later.

[mouse07410]

@johvik thank you - you've provided great info that both unambiguously described the bug, and pointed out a likely solution.