X64: Changed 0x%x formats in log.cc to 0x%p and omitted reinterpretting pointers.

X64: Changed 0x%x formats in log.cc to 0x%p and omitted reinterpretting pointers.

[Lasse Reichstein, 2009-05-07 07:59:15]

Review please.

[Mikhail Naganov, 2009-05-07 08:25:43]

On 2009/05/07 07:59:15, Lasse Reichstein wrote:
> Review please.

Are you sure that "%p" doesn't also print "0x" on Linux? I've just checked that
under gcc 4.0.3 it does.

The inconsitency between Windows and Linux behavior on interpreting %p was the
main reason of using "%x" which works everywhere the same.

[Lasse Reichstein, 2009-05-07 08:47:02]

On 2009/05/07 08:25:43, Mikhail Naganov wrote:
> Are you sure that "%p" doesn't also print "0x" on Linux? I've just checked
that
> under gcc 4.0.3 it does.

Not sure at all. 

> 
> The inconsitency between Windows and Linux behavior on interpreting %p was the
> main reason of using "%x" which works everywhere the same.

You are right. The problem is that "%x" is not sufficient to print a pointer in
64-bit mode.

Since Windows doesn't have inttypes.h, the simpest solution is probably to
create a macro, V8PRIp, to print a pointer our way (which is either "x", "lx" or
"llx" depending on compiler and wordsize).

[Søren Thygesen Gjesse, 2009-05-07 09:14:42]

On 2009/05/07 08:47:02, Lasse Reichstein wrote:
> On 2009/05/07 08:25:43, Mikhail Naganov wrote:
> > Are you sure that "%p" doesn't also print "0x" on Linux? I've just checked
> that
> > under gcc 4.0.3 it does.
> 
> Not sure at all. 
> 
> > 
> > The inconsitency between Windows and Linux behavior on interpreting %p was
the
> > main reason of using "%x" which works everywhere the same.
> 
> You are right. The problem is that "%x" is not sufficient to print a pointer
in
> 64-bit mode.
> 
> Since Windows doesn't have inttypes.h, the simpest solution is probably to
> create a macro, V8PRIp, to print a pointer our way (which is either "x", "lx"
or
> "llx" depending on compiler and wordsize).

LGTM, but where is V8PRIp defined?

[Søren Thygesen Gjesse, 2009-05-07 09:17:12]

On 2009/05/07 09:14:42, Søren Gjesse wrote:
> On 2009/05/07 08:47:02, Lasse Reichstein wrote:
> > On 2009/05/07 08:25:43, Mikhail Naganov wrote:
> > > Are you sure that "%p" doesn't also print "0x" on Linux? I've just checked
> > that
> > > under gcc 4.0.3 it does.
> > 
> > Not sure at all. 
> > 
> > > 
> > > The inconsitency between Windows and Linux behavior on interpreting %p was
> the
> > > main reason of using "%x" which works everywhere the same.
> > 
> > You are right. The problem is that "%x" is not sufficient to print a pointer
> in
> > 64-bit mode.
> > 
> > Since Windows doesn't have inttypes.h, the simpest solution is probably to
> > create a macro, V8PRIp, to print a pointer our way (which is either "x",
"lx"
> or
> > "llx" depending on compiler and wordsize).
> 
> LGTM, but where is V8PRIp defined?

LGTM - found V8PRIp!

[William Hesse, 2009-05-07 09:42:11]

http://codereview.chromium.org/114010/diff/2002/2003
File src/globals.h (right):

http://codereview.chromium.org/114010/diff/2002/2003#newcode89
Line 89: #define V8PRIp V8_PTR_PREFIX "x"
why not #define V8_PTR_FORMAT "0x%llx"
etc. for all the cases, then
printf("My pointer: "V8_PTR_FORMAT
       " it is easy to break lines here too.", ptr);

I guess you need to break out of quotes for the macro to apply, or don't you? 
If not, I wouldn't include the quotes.

[Erik Corry, 2009-05-07 09:58:57]

http://codereview.chromium.org/114010/diff/2002/2003
File src/globals.h (right):

http://codereview.chromium.org/114010/diff/2002/2003#newcode89
Line 89: #define V8PRIp V8_PTR_PREFIX "x"
Certainly prefix is the wrong name; since it goes after the '%' it must be a
suffix.