Antonio,

Yes...I concur.  That is a better solution as it is agnostic to bus width
and lets each platform compiler deal with the issue more cleanly.  The
solution I proposed admittedly would still be a problem on some
architectures.

I seem to have a faint recollection that a few non-gnu compilers don't
always support "%p" as a formatting code.  This is probably a non-issue for
OpenOCD, however, since most -- if not all -- people who build the software
are probably doing so with gnu gcc and its standard libraries anyway.

Gary


On 8/21/11 4:43 PM, "Antonio Borneo" <[email protected]> wrote:

> On Mon, Aug 22, 2011 at 7:21 AM, Tomek CEDRO <[email protected]> wrote:
>> Hello Carlson and thank you for your support! :-)
>> 
>> On Sun, Aug 21, 2011 at 9:09 PM, Carlson Gary
>> <[email protected]> wrote:
>>> I didn't realize there was a fork for SWD support, so my original attempt
>>> was based on the master branch.  Nevertheless I cloned the fork that you
>>> indicated and immediately ran into a number of compiler issues trying to
>>> compile it on a Mac OS X platform.  The patch attachment covers most of the
>>> changes necessary to address those problems.  These should be neutral for
>>> other platforms, but you may want to make sure.
>>> (..)
> 
> Tomek code prints pointers by casting to int and using format string "0x%08X".
> -    LOG_INFO("New SWD context initialized at 0x%08X",
> (int)jtag_interface->transport->ctx);
> 
> Carlson patch changes the cast to type intptr_t, and seams also using
> 64 bits since changes format string to "0x08lX"
> +    LOG_INFO("New SWD context initialized at 0x%08lX",
> (intptr_t)jtag_interface->transport->ctx);
> 
> A more appropriate way to print pointers is through the format string
> "%p", that does not require casting and is portable across 32 and 64
> bits.
> Above line should be replaced with
> +    LOG_INFO("New SWD context initialized at %p",
> jtag_interface->transport->ctx);
> Tested across Linux x86 and Mac OS X x86-64
> 
> Best Regards,
> Antonio Borneo


_______________________________________________
Openocd-development mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/openocd-development

Reply via email to