integer types in gdb code

integer types in gdb code

[Kris Warkentin]

I'm having trouble cleaning up our code to compile on various targets.

It seems like most targets have uint32_t, int8_t, etc. defined but not
always (ie. cygwin).  Is there a 'gdb approved' header or set of defines
that works consistently everywhere?  I'd like to be able to use something
consistently in all of our code without having to type in 'unsigned long
long' and such.

cheers,

Kris

Re: integer types in gdb code

[Andrew Cagney]

> I'm having trouble cleaning up our code to compile on various targets.
> 
> It seems like most targets have uint32_t, int8_t, etc. defined but not
> always (ie. cygwin).  Is there a 'gdb approved' header or set of defines
> that works consistently everywhere?  I'd like to be able to use something
> consistently in all of our code without having to type in 'unsigned long
> long' and such.

CORE_ADDR  ~= void *
LONGEST ~= long
ULONGEST ~= unsigned long

Andrew

Re: integer types in gdb code

[Kris Warkentin]

> > I'm having trouble cleaning up our code to compile on various targets.
> >
> > It seems like most targets have uint32_t, int8_t, etc. defined but not
> > always (ie. cygwin).  Is there a 'gdb approved' header or set of defines
> > that works consistently everywhere?  I'd like to be able to use
something
> > consistently in all of our code without having to type in 'unsigned long
> > long' and such.
>
> CORE_ADDR  ~= void *
> LONGEST ~= long
> ULONGEST ~= unsigned long

No 16 and 8 bit constructs?  Also, what if you want to guarantee the number
of bits?  From defs.h it looks to me like LONGEST could be 32 or 64 bits, so
if you're reading structures over the wire, you could have problems.

Kris

Re: integer types in gdb code

[Andrew Cagney]

>
>> CORE_ADDR  ~= void *
>> LONGEST ~= long
>> ULONGEST ~= unsigned long
> 
> 
> No 16 and 8 bit constructs?  Also, what if you want to guarantee the number
> of bits?  From defs.h it looks to me like LONGEST could be 32 or 64 bits, so
> if you're reading structures over the wire, you could have problems.

The above are GDB's host types.

When representing target types (or stuff related to a protocol) GDB can 
either use a function like extract_unsigned_integer() or construct a 
type exactly describing the data using that (but be wary of byte order 
problems).

Trying to do something like:

struct { uint32_t i; ... }

can end up being non-portable (due to struct packing issues and the like).

Andrew