Re: swapcontext and free memory

[Jon Mayo <jon.mayo@gmail.com>]

On Wed, Mar 18, 2009 at 12:22 PM, Tiago Maluta
<maluta_tiago@yahoo.com.br> wrote:
> xcomp@arcor.de wrote:
>> Hi all,
>> I have written a small test case which simply creates a new context, swaps to it and returns to the main function. I am allocating memory for the ucontext_t structures and the stack on which the context is executed using malloc.
> Because of this I also want to deallocate the memory again using free.
> The problem is I can't deallocate the memory after swapping back. It
> runs only if all three free() calls are commented out. Otherwise a
> segmentation fault occurs. Does swapcontext handle this on its own? Can
> anyone explain this behavior? I didn't find anything concerning this
> problem in the web.
>> I am running this program on Ubuntu 8.04 with the following system information: Linux
> ubuntu8041 2.6.24-23-generic #1 SMP i686 GNU/Linux
>
> (...)
>
>>       main_context = (ucontext*) malloc(sizeof(ucontext));
>>       thread_context = (ucontext*) malloc(sizeof(ucontext));
>
> changing:
>
> main_context = (ucontext_t*) malloc(sizeof(ucontext_t));
> thread_context = (ucontext_t*) malloc(sizeof(ucontext_t));
>
> and compiling with free() I got:
>
> #./a.out
> start
>
> Allocated memory:
>        main_context:   0x804b008
>        thread_context: 0x804b168
>        thread_stack:   0x804b2c8
> thread_function was called...
>
> swapcontext() returned. Freeing memory starts now...
> thread_stack was deallocated...
> thread_context was deallocated...
> main_context was deallocated...
>
> I'm using glibc version 2.8.
>
> --tm
>
>
>

I think you should be able to free() them fine. Could you post a short
bit of source that exhibits the problem? Also I tend to mmap() some
anonymous memory instead of using malloc() for things like makecontext
and sigaltstack. Because I want to put guard pages on my stacks due to
them generally being very small.

-- 
Jon Mayo
<jon.mayo@gmail.com>
--
To unsubscribe from this list: send the line "unsubscribe linux-c-programming" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html