Memory Alignment Problem

[iduniq]

Hi,
Good day! Anyone knows how to extract shorts, long, chars from
a char array considering the memory alignment? It's like this,
there's a char buffer, which is cast into a struct, then cast back to
a char pointer. The first two chars should be a short; when displayed
as char the number (value 0x8600, equal to 134) is correct but when a
short pointer is pointed to this char and the short is displayed, it
becomes a negative number (-31232).

TIA,
Roy

[Ian Collins]

iduniq wrote:

Hi,
Good day! Anyone knows how to extract shorts, long, chars from
a char array considering the memory alignment? It's like this,
there's a char buffer, which is cast into a struct, then cast back to
a char pointer. The first two chars should be a short; when displayed
as char the number (value 0x8600, equal to 134) is correct but when a
short pointer is pointed to this char and the short is displayed, it
becomes a negative number (-31232).

The bit pattern 0x8600 is -31232 when cast to a (16 bit) short. Are you
muddling up signed and unsigned types?

--
Ian Collins.

[Kenneth Brody]

Ian Collins wrote:

>
iduniq wrote:

Hi,
Good day! Anyone knows how to extract shorts, long, chars from
a char array considering the memory alignment? It's like this,
there's a char buffer, which is cast into a struct, then cast back to
a char pointer. The first two chars should be a short; when displayed
as char the number (value 0x8600, equal to 134) is correct but when a
short pointer is pointed to this char and the short is displayed, it
becomes a negative number (-31232).

The bit pattern 0x8600 is -31232 when cast to a (16 bit) short. Are you
muddling up signed and unsigned types?

I think he's mixing up endianness. If you have:

char foo[] = { 0x86, 0x00 };

and read *(short *)foo on a little-endian, 16-bit-short, properly-
aligned system, you will probably (though it's probably UB according
to the standard) read 134 (0x0086). However, on a big-endian,
everything-else-the-same-as-before system, you will probably read
-31232 (0x8600).

--
+-------------------------+--------------------+-----------------------+
| Kenneth J. Brody | www.hvcomputer.com | #include |
| kenbrody/at\spamcop.net | www.fptech.com | <std_disclaimer.h|
+-------------------------+--------------------+-----------------------+
Don't e-mail me at: <mailto:Th*************@gmail.com>

[iduniq]

Hi guys,

I'm very sorry I only realise my fault after I sent SOS. Ian is
correct, there was a prior ntosh before this code.

Thanks,
Roy
On Mar 16, 9:49 pm, Kenneth Brody <[e-mail]:

Ian Collins wrote:

iduniq wrote:

Hi,
Good day! Anyone knows how to extract shorts, long, chars from
a char array considering the memory alignment? It's like this,
there's a char buffer, which is cast into a struct, then cast back to
a char pointer. The first two chars should be a short; when displayed
as char the number (value 0x8600, equal to 134) is correct but when a
short pointer is pointed to this char and the short is displayed, it
becomes a negative number (-31232).

The bit pattern 0x8600 is -31232 when cast to a (16 bit) short. Are you
muddling up signed and unsigned types?

I think he's mixing up endianness. If you have:

char foo[] = { 0x86, 0x00 };

and read *(short *)foo on a little-endian, 16-bit-short, properly-
aligned system, you will probably (though it's probably UB according
to the standard) read 134 (0x0086). However, on a big-endian,
everything-else-the-same-as-before system, you will probably read
-31232 (0x8600).

--
+-------------------------+--------------------+-----------------------+
| Kenneth J. Brody |www.hvcomputer.com| #include |
| kenbrody/at\spamcop.net |www.fptech.com | <std_disclaimer.h|
+-------------------------+--------------------+-----------------------+
Don't e-mail me at: <mailto:[e-mail]>