compile wine with icc

[Tomas Carnecky]

anyone already tried that?

I'm getting some errors when compiling wine which I was able to fix, but 
  World of Warcraft won't start... Will icc be supported at some time in 
the furure?

tom

[Eric Pouech]

Tomas Carnecky wrote:
> anyone already tried that?
>
> I'm getting some errors when compiling wine which I was able to fix, but 
>  World of Warcraft won't start... Will icc be supported at some time in 
> the furure?
>
> tom

last time I checked, icc didn't support stdcall calling convention, 
which is rather important for compiling wine (on x86)
didn't check for its presence lately

A+

[Dan Kegel]

On 10/16/05, Tomas Carnecky  wrote:
> I'm getting some errors when compiling wine which I was able to fix, but
>   World of Warcraft won't start... Will icc be supported at some time in
> the furure?

Maybe, especially if people who want it supported pitch in
by running the Wine regression tests and reporting any problems.

[Tomas Carnecky]

Dan Kegel wrote:
> On 10/16/05, Tomas Carnecky  wrote:
> > I'm getting some errors when compiling wine which I was able to fix, but
> >  World of Warcraft won't start... Will icc be supported at some time in
> > the furure?
>
> Maybe, especially if people who want it supported pitch in
> by running the Wine regression tests and reporting any problems.

as icc has other cmd-line options, the compiler spits out lots of 
warnings, like this one:

icc -c -I. -I. -I../../include -I../../include  -D__WINESRC__ 
-DNO_LIBWINE_PORT -DWINE_UNICODE_API="" -D_REENTRANT -fPIC -Wall -pipe 
-mpreferred-stack-boundary=2 -fno-strict-aliasing -gstabs+ 
-Wdeclaration-after-statement -Wpointer-arith  -g -O2  -o sortkey.o 
sortkey.c
iccbin: Command line warning: ignoring option '-mp'; no argument required
iccbin: Command line warning: ignoring option '-g'; no argument required
iccbin: Command line warning: ignoring option '-W'; no argument required

futhermore, lots of compiler warnings a la:
../../include/wine/unicode.h(106): remark #810: conversion from "int" to 
"WCHAR={unsigned short}" may lose significant bits
       return ch + wine_casemap_lower[wine_casemap_lower[ch >> 8] + (ch 
& 0xff)];
relay.c(178): remark #1684: conversion from pointer to same-sized 
integral type (potential portability problem)
       fprintf( outfile, "\t.byte 0x64\n\tmovw %%sp, (%d)\n", STACKOFFSET );

and much more...

and finally fails on file dlls/kernel/thread.c with:
/tmp/iccbinN9FwHFas_: Assembler messages:
/tmp/iccbinN9FwHFas_:2812: Error: symbol `SetLastError' is already defined
/tmp/iccbinN9FwHFas_:2840: Error: symbol `GetCurrentThreadId' is already 
defined

tom

[Dan Kegel]

That's expected.  To support icc, one would have to adjust the
commandline options a bit depending on the compiler being used.
In particular, one would want to suppress many compiler warnings
like the ones you mentioned.  icc provides a nice way of suppressing
any desired set of warnings.

[Eric Pouech]

Dan Kegel wrote:
> That's expected.  To support icc, one would have to adjust the
> commandline options a bit depending on the compiler being used.
> In particular, one would want to suppress many compiler warnings
> like the ones you mentioned.  icc provides a nice way of suppressing
> any desired set of warnings.

I used something like CFLAGS="-g -Wbrief 
-wd108,144,167,181,188,193,279,556,810,869,981,1287,1418,1419,1572"
last time I tried
A+

[Tomas Carnecky]

Eric Pouech wrote:
> Dan Kegel wrote:
> > That's expected.  To support icc, one would have to adjust the
> > commandline options a bit depending on the compiler being used.
> > In particular, one would want to suppress many compiler warnings
> > like the ones you mentioned.  icc provides a nice way of suppressing
> > any desired set of warnings.
>
> I used something like CFLAGS="-g -Wbrief 
> -wd108,144,167,181,188,193,279,556,810,869,981,1287,1418,1419,1572"

problem is, even if I specify CFLAGS on the command-line, wine adds its 
own flags. So the actual compile command looks as follows:

icc -c -I. -I. -I../../include -I../../include  -D__WINESRC__ 
-DNO_LIBWINE_PORT -DWINE_UNICODE_API="" -D_REENTRANT -fPIC -Wall -pipe 
-mpreferred-stack-boundary=2 -fno-strict-aliasing -gstabs+ 
-Wdeclaration-after-statement -Wpointer-arith  -g -Wbrief 
-wd108,144,167,181,188,193,279,556,810,869,981,1287,1418,1419,1572  -o 
c_28604.o c_28604.c

I'd like to get rid of "-Wall -pipe -mpreferred-stack-boundary=2 
-fno-strict-aliasing -gstabs+ -Wdeclaration-after-statement 
-Wpointer-arith"... how can I do that?

thanks
tom

[Tomas Carnecky]

Eric Pouech wrote:
> Dan Kegel wrote:
> > That's expected.  To support icc, one would have to adjust the
> > commandline options a bit depending on the compiler being used.
> > In particular, one would want to suppress many compiler warnings
> > like the ones you mentioned.  icc provides a nice way of suppressing
> > any desired set of warnings.
>
> I used something like CFLAGS="-g -Wbrief 
> -wd108,144,167,181,188,193,279,556,810,869,981,1287,1418,1419,1572"

ok.. I've found what I have to do.. override EXTRACFLAGS...

now I can compile the wine without warnings.. but here comes the first 
compiler error:

tom@eiger ~/cvs/wine/libs/wine $ icc -c -I. -I. -I../../include 
-I../../include  -D__WINESRC__ -DBINDIR=""/usr/bin"" 
-DDLLDIR=""/usr/lib/wine"" -D_REENTRANT -fPIC   -g -Wbrief 
-wd108,144,167,181,188,193,279,556,810,869,981,1287,1418,1419,1572  -o 
ldt.o ldt.c
/tmp/iccbinFR1dsuas_: Assembler messages:
/tmp/iccbinFR1dsuas_:2339: Error: symbol `wine_get_fs' is already defined
/tmp/iccbinFR1dsuas_:2363: Error: symbol `wine_set_fs' is already defined
tom@eiger ~/cvs/wine/libs/wine $

tom

[Dimi Paun]

On Sat, 2005-10-29 at 14:58 +0200, Tomas Carnecky wrote:
> > I used something like CFLAGS="-g -Wbrief 
> > -wd108,144,167,181,188,193,279,556,810,869,981,1287,1418,1419,1572"

I suppose these disables some warnings. Wouldn't they be useful 
for Wine if fixed?

[Eric Pouech]

Dimi Paun wrote:
> On Sat, 2005-10-29 at 14:58 +0200, Tomas Carnecky wrote:
> > > I used something like CFLAGS="-g -Wbrief 
> > > -wd108,144,167,181,188,193,279,556,810,869,981,1287,1418,1419,1572"
>
> I suppose these disables some warnings. Wouldn't they be useful 
> for Wine if fixed?

well, not all of them as you can see
#108	bit field on signed type
#111	statement is unreachable
#144	assign to function pointer a value of type void*
#167	cast from pointer to pointer, but of different type
#181	incompatible pointer type for %p (char*)
#188	enumerated type used as integer
#193	0 used in some pre-processor when ID is not defined
#279	controlling expression is constant (all are debug macros fire this)
#310	old-C style parameter list
#556	cast from function pointer to void*
#593	variable set but never used
#810	conversion to integral type of smaller size
#869	parameter to func not referenced in func
#981	arguments (to func) are evaluated in an unspecified order
#1287	unknown __attribute__ to function definition
#1418	global var/func without prior definition
#1419	extern declaration in .c file
#1572	comparison of floating point values

[Eric Pouech]

>

no, you just need to override CFLAGS when using configure...
CFLAGS="blha...." ./configure
A+