[Qemu-devel] Re: [5998] Fix TARGET_LONG_BITS warning in TCG

[Jan Kiszka <jan.kiszka@web.de>]

[-- Attachment #1: Type: text/plain, Size: 3261 bytes --]

Blue Swirl wrote:
> On 12/13/08, Jan Kiszka <jan.kiszka@web.de> wrote:
>> Blue Swirl wrote:
>>  > On 12/13/08, Jan Kiszka <jan.kiszka@web.de> wrote:
>>  >> Blue Swirl wrote:
>>  >>  > On 12/13/08, Jan Kiszka <jan.kiszka@web.de> wrote:
>>  >>  >> Blue Swirl wrote:
>>  >>  >>  > Revision: 5998
>>  >>  >>  >           http://svn.sv.gnu.org/viewvc/?view=rev&root=qemu&revision=5998
>>  >>  >>  > Author:   blueswir1
>>  >>  >>  > Date:     2008-12-13 08:16:43 +0000 (Sat, 13 Dec 2008)
>>  >>  >>  >
>>  >>  >>  > Log Message:
>>  >>  >>  > -----------
>>  >>  >>  > Fix TARGET_LONG_BITS warning in TCG
>>  >>  >>  >
>>  >>  >>  > Looking at tcg/tcg.c:828, the bug that the warning indicated would show up as
>>  >>  >>  > incorrect PC shown in log, only on 32 bit big endian host emulating a 64 bit
>>  >>  >>  > target, -d op flag enabled. Now that dyngen is gone, the patch can be applied.
>>  >>  >>  >
>>  >>  >>  > Modified Paths:
>>  >>  >>  > --------------
>>  >>  >>  >     trunk/tcg/tcg-runtime.c
>>  >>  >>  >
>>  >>  >>  > Modified: trunk/tcg/tcg-runtime.c
>>  >>  >>  > ===================================================================
>>  >>  >>  > --- trunk/tcg/tcg-runtime.c   2008-12-12 20:02:52 UTC (rev 5997)
>>  >>  >>  > +++ trunk/tcg/tcg-runtime.c   2008-12-13 08:16:43 UTC (rev 5998)
>>  >>  >>  > @@ -29,6 +29,7 @@
>>  >>  >>  >
>>  >>  >>  >  #include "config.h"
>>  >>  >>  >  #include "osdep.h"
>>  >>  >>  > +#include "cpu.h" // For TARGET_LONG_BITS
>>  >>  >>  >  #include "tcg.h"
>>  >>  >>  >
>>  >>  >>  >  int64_t tcg_helper_shl_i64(int64_t arg1, int64_t arg2)
>>  >>  >>  >
>>  >>  >>
>>  >>  >>
>>  >>  >> Nice that this warning is finally fixed - but what is the point of
>>  >>  >>  including all those headers at all here? [1] worked well for me while
>>  >>  >>  reducing dependencies.
>>  >>  >
>>  >>  > I think all header files should be checked and even reorganized when
>>  >>  > necessary. One item is the dyngen removal, cpu.h/exec.h distinction is
>>  >>  > not clear for all targets and headers used by both user and system
>>  >>  > emulators could be adjusted. Header files in hw should be OK, but
>>  >>  > worth checking.
>>  >>
>>  >>
>>  >> Header dependencies and usage policies should be rechecked and probably
>>  >>  cleaned up now, no question (see also my dyngen-exec.h problems).
>>  >>
>>  >>  But here I was referring to the need of this _particular_ C file for any
>>  >>  header except inttypes.h. Why including headers _here_ for no use?
>>  >
>>  > I tested the apparently harmless patch, but it isn't OK. osdep.h,
>>  > qemu-log.h and tcg.h use FILE and size_t, and therefore we need at
>>  > least stdio.h. Those headers should be fixed, then the patch would
>>  > work.
>>  >
>>
>>
>> Sorry, confused: Which patch for which file do you mean now? If you
>>  refer to mine: Why do we need osdep.h, qemu-log.h or tcg.h for
>>  tcg-runtime.c?
> 
> This one:
> http://permalink.gmane.org/gmane.comp.emulators.qemu/32076
> 
> They are not needed, but pulled in as a side product.
> 

OK, finally got it: we want prototypes for non-static functions now, and
therefore we need tcg/tcg.h and all what comes with it.

Jan


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 257 bytes --]