[mercury-users] unsafe_set_char

[Julien Fischer]

Hi,

On Sun, 30 May 2010 jc at cs.york.ac.uk wrote:

> Thanks for the info. I had already tried my own dirty method which is the
> same as Julien's except it had fewer properties declared in the square
> brackets. Weirdly, I also called it very_unsafe_set_char (I guess it's the
> obvious choice of name!). Unfortunately both it and Julien's code have the
> same problem. Compilation is fine, but we get a runtime error
>
> cscipd001 ~/godot/research/srl/prism/abc abc
>
> *** Mercury runtime: caught segmentation violation ***
> PC at signal: -1222211540 (b726882c)
> address involved: 0x80516a8
> This may have been caused by a stack overflow, due to unbounded recursion.
> exiting from signal handler

I suspect you are attempting to modify a string stored as static data,
i.e. you are encountering the issue that caused the unique mode to be
disabled in the first place.

The following definition does a check for whether something a string
is stored on the heap.  If the above is indeed the problem, then it
should fix it.

:- pragma foreign_proc("C",
    very_unsafe_set_char(Ch::in, Index::in, Str0::di, Str::uo),
    [will_not_call_mercury, promise_pure, thread_safe, does_not_affect_liveness],
"
    if (MR_in_heap_range(Str0)) {
        Str = Str0;
        MR_set_char(Str, Index, Ch);
    } else {
        size_t len = strlen(Str0);
        MR_allocate_aligned_string_msg(Str, len, MR_PROC_LABEL);
        strcpy(Str, Str0);
        MR_set_char(Str, Index, Ch);
    }
").

Julien.
--------------------------------------------------------------------------
mercury-users mailing list
Post messages to:       mercury-users at csse.unimelb.edu.au
Administrative Queries: owner-mercury-users at csse.unimelb.edu.au
Subscriptions:          mercury-users-request at csse.unimelb.edu.au
--------------------------------------------------------------------------