[Bf-committers] BlenderCE - the adventure continue...

[Daniel Dunbar]

Don't see how this is related, but you probably
should also have the /J flag to make char
unsigned. (Unless it is unsigned by default on
WinCE, but I doubt it).

--- Salvatore Russo <salvatore.russo at laposte.net> wrote:

> Hallo,
> 
> I reworked on Ghost to have a GostCE working with char and not wchar_t. And the same think with
> BlenderCE. I am
> quite happy now for GhostCE
 now, only a couple of additional files and 2 minor modifications in
> the old ghost tree
> are now necessary to make Ghost work for WinCE.
> For Blender, you were right! So I wrote some additional functions to locally solve this Unicode
> story
 only
> locally. I looks really better like that :=)
> 
> Now Blender runs ok until:
> MEM_mallocN(unsigned int 22992, char * 0x00555504) line 188
> dna_sdna_from_data(void * 0x005593e0 DNAstr, int 22992, int 0) line 479 + 12 bytes
> filedata_new() line 848 + 20 bytes
> blo_openblendermemory(void * 0x0055fe08 datatoc_B_blend, int 93008) line 902 + 4 bytes
> BLO_read_from_memory(void * 0x0055fe08 datatoc_B_blend, int 93008, int * 0x2469f9fc) line 278 +
> 12 bytes
> BKE_read_file_from_memory(char * 0x0055fe08 datatoc_B_blend, int 93008, void * 0x00000000) line
> 465 + 16 bytes
> BIF_read_homefile() line 301 + 20 bytes
> BIF_init() line 610
> main(int 1, char * * 0x006a0180) line 448
> 
> and craches here:
> memh= (MemHead *)malloc(len+sizeof(MemHead)+sizeof(MemTail));
> with the following exception:
> 
> “First chance exception in BlenderCE.exe: 0x80000002 Datatype misalignment.
> 
> My complile option (in debug mode) in EVC4 are:
> /nologo /GX /Zi /Od /Gy /I "..\intern\img\extern” /I... /D "DEBUG" /D "POCKETCONSOLE" /D "ARM"
> /D "_ARM_" /D
> "ARMV4" /D "MS_WINDOWS" /D "__CUSTOM_STREAM" /D "WIN32" /D _WIN32_WCE=$(CEVersion) /D
> "$(CePlatform)" /D
> UNDER_CE=$(CEVersion) /D "UNICODE" /D "_UNICODE" /FR"ARMV4Dbg/" /Fp"ARMV4Dbg/BlenderCE.pch" /YX
> /Fo"ARMV4Dbg/"
> /Fd"ARMV4Dbg/" /M$(CECrtMTDebug) /c
> 
> I was obliged to add /Gy to solve (for some functions) the error:
> cannot adjust code at offset=offset, va=value
> LINK was unable to pad the code as required.
> Certain instructions are not allowed to cross page boundaries on some processors. LINK attempts
> to add pads to
> correct this situation. In this case, LINK could not work around the problem.
> 
> I also tried with /Zp1 and /Zp 16 to have a structure member alignement of 1 or 16. By default
> it is 8 but no
> improvement! I found in genfile.c the following remark:
> IMPORTANT:
> - aligment of variables has to be done in such a way, that any system does
>    not create 'padding' (gaps) in structures. So make sure that:
>    - short: 2 aligned
>    - int: 4 aligned
>    - float: 4 aligned
>    - double: 8 aligned
>    - long: 8 aligned
>    - struct: 8 aligned
> 
> So it should work with an alignmenent of 8 :=( No? But I am not really understanding all what
> this imply...
> 
> I read these last days some documentations on WinCE memory management
 For example, the
> following article is very
> interesting:
> 
> http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dncenet/html/advmemmgmt.asp
> 
> (especially until dll load problem
 I don’t think Windows mobile 2003 is concerned by this
> problem
)
> 
> Has someone any idea where the problem is coming from? Do you think that I missed any important
> requirements to
> take care to compile Blender? 
 especially about the memory management critical on WinCE?
> I have on my pocketpc the wonderful grey color of Blender
 it is already a first thing :=) And
> the console say me
> to continue happily!
> 
> Sorry for this long post...
> 
> Thanks guys for your help,
> Salvatore
> 
> Accédez au courrier électronique de La Poste : www.laposte.net ; 
> 3615 LAPOSTENET (0,34€/mn) ; tél : 08 92 68 13 50 (0,34€/mn)
> 
> 
> 
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at projects.blender.org
> http://projects.blender.org/mailman/listinfo/bf-committers
> 


daniel dunbar