Re: [pdf-devel] Problems Compiling with PDF_FORCE_BIGNUMS (Unexpected Code Generation?)

[Jeffrey Walton]

Hi Aleksander,

Thanks for the QC... It's a new day, and I'm fresh, so hopefully no
dumb mistakes.

Jeff

On Tue, Feb 2, 2010 at 3:13 AM, Aleksander Morgado
<address@hidden> wrote:
> Hi Jeff,
>
>> I believe the problem arises because, in the case of
>> PDF_FORCE_BIGNUMS, a struct is being pushed on the stack and accessed
>> incorrectly in the called function. I was able to duplicate the issue
>> in the attached struct-test.c. In the test file, I create a similar
>> struct initialized to 1 (ie, high=0, low=1). I then push on the stack
>> to Print:
>>
>> $ ./struct-test.exe
>> high=-1079373640, low=134513824
>> high=1, low=0
>>
>
> For the first case, your NEW_MY_ADT64_T() function is not returning
> anything, thus the values printed are undefined. The proper way would be
> including the 'return t' at the end:
> MY_ADT64_T NEW_MY_ADT64_T(int32_t high, uint32_t low)
> {
>     MY_ADT64_T t;
>     t.high = high;
>     t.low = low;
>     return t;
> }
>
> For the second case, you are initializing the variable as follows:
> MY_ADT64_T t2 = {0,1};
> And your struct is defined having 'low' as first variable in the struct:
> struct MY_ADT64_S
> {
>     uint32_t low;
>     int32_t high;
> };
>
> Thus, when initializing t2={0,1}, you are initializing low=0 and high=1,
> which is exactly what you print...
>
> Cheers,
> -Aleksander
>