On 24/03/07 18:16:51, Andrew Haley wrote:
> > If I do eg:
> >
> > Pixel16 *tileData[4]; uint16 *pDestPixel; Pixel16 *dataPtr, *transDataPtr;
> > sint32 startX, endX;
> > __asm__ __volatile__ (
> > "push %%ebx \n\t" // save reg for PIC!
> > //code here using %1 but now reg can be rused:
> >
> > "movl %5, %1 \n\t" //no reg left for %5 when gcc should handel it
> > //but now gcc says:
> > //error: memory input 5 is not directly addressable
> > "movl (%1,%%ebx,4), %1 \n\t" //instead of "movl (%5,%%ebx,4), %1 \n\t"
> > //because m for %5 didn't work here
> >
> > //more code using also ebx and ebp
> > "pop %%ebx \n\t" // restore reg for PIC!
> > :
> > : "D" (pDestPixel), "d" (endX), "c" (startX), "a" (transDataPtr), "S" (dataPtr), "m" (tileData)
>
> > : "cc" //don't tell gcc about what was done to ebx because of PIC!!!
> > );
> >
> > gcc complains about: error: memory input 5 is not directly addressable
>
> > Why is that? How else should I load the tileData pointer from
> > memory into a register?
>
> I wouldn't be using the "m" constraint for this. Pass in the pointer
> as an operand using constraint "r" or "g". Something like "g"(&tileData);
>
> Can you post some small sample code? Make it the shortest possible
> programs that shows your problem. Make sure all the types it needs
> are defined.
This is a problem. My task is to tranlate this inline asm written for MSC to
gcc. It's from a big game (Call to Power II, CTP2). I've attached a file with
the function that contains the code. But it won't compile withou a lot
modifications outside the source tree.
At L131 my translation starts. It has the constrains I got it to compile with
but segfaults then. I noticed that using any other var than tileData works
with "m". Could it be a problem for gcc to use the double pointer here:
Pixel16 **tileData ?
>
> > > > Also I wonder which -f option that comes with -O2 makes gcc stop
reporting:
> > > > error: can't find a register in class 'LEGACY_REGS' while reloading
'asm'
> > >
> > > I guess this is a joke.
> >
> > No, it isn't. I wouldn't have asked else wise...
>
> Reload has run out of registers. It's telling you that it needs
> another one. A -f option is not going to give reload another
> register.
The point is, if I just add -O2 to the g++ cammand, this error disappears
without having changed anything in the code or the number of register the part
uses.
It seems that an option in -O2 or above lets g++ be able to cope with the
registers I have left it (ebx and ebp, esp). I found this solution quite often
on the net.
>
> > Is it the combination of all -f options in -O2? I coudn't find a
> > single obviouse one in the manpage. Is it a rule then to include
> > -O2 or higher if inline asm is used???
>
> No, there's no such rule. You've run out of registers.
>
> gcc needs one or two spare registers to work with. The x86 has eight,
> some of which are used for special purposes. You are using five in
> your asm. This is a very tight situation.
Compiling all this with -fomit-frame-pointer:
Can gcc not push all registers, and give me all seven without esp?
I also tryed to use "R" instead of "m" for tileData. It compiled without
errors even when I had specified -fPIC. -fPIC seems to use ebx so I pushed and
poped that in my asm code. But when looking at the disassembly I had to see
gcc used ebx for tileData and not ebp. But if I specified "b" it complains
about:
error: PIC register %ebx clobbered in asm
Is this a bug in gcc??? Is there a constraint just for ebp so I don't have to
use "R" and hope it gets into ebp?
>
> Andrew.
Thanks Andrew
Lynx
compiled with:
g++ -DHAVE_CONFIG_H -I. -I. -I../../../ctp2_code/os/include -I../../../ctp2_code/os/nowin32 -I../../../ctp2_code -I../../.. -I../../../ctp2_code/os/include -I../../../ctp2_code/ctp -I../../../ctp2_code/ctp/ctp2_utils -I../../../ctp2_code/ctp/ctp2_rsrc -I../../../ctp2_code/ctp/debugtools -I../../../ctp2_code/ui/interface -I../../../ctp2_code/ui/netshell -I../../../ctp2_code/robot/utility -I../../../ctp2_code/robot/pathing -I../../../ctp2_code/robot/aibackdoor -I../../../ctp2_code/gfx/spritesys -I../../../ctp2_code/gfx/tilesys -I../../../ctp2_code/gfx/gfx_utils -I../../../ctp2_code/gs/database -I../../../ctp2_code/gs/fileio -I../../../ctp2_code/gs/gameobj -I../../../ctp2_code/gs/utility -I../../../ctp2_code/gs/world -I../../../ctp2_code/net/io -I../../../ctp2_code/net/general -I../../../ctp2_code/ui/aui_utils -I../../../ctp2_code/ui/aui_sdl -I../../../ctp2_code/ui/aui_directx -I../../../ctp2_code/ui/aui_ctp2 -I../../../ctp2_code/ui/aui_common -I../../../ctp2_code/libs/anet/h -I../../../ctp2_code/mm -I../../../ctp2_code/robotcom/backdoor -I../../../ctp2_code/gs/slic -I../../../ctp2_code/gs/slic -I../../../ctp2_code/gfx/layers -I../../../ctp2_code/mapgen -I../../../ctp2_code/ui/freetype -I../../../ctp2_code/sound -I../../../ctp2_code/GameWatch/gamewatch -I../../../ctp2_code/GameWatch/gwciv -I../../../ctp2_code/ctp/fingerprint -I../../../ctp2_code/ui/slic_debug -I../../../ctp2_code/gs/outcom -I../../../ctp2_code/ctp -I../../../ctp2_code/ui/aui_common -I../../../ctp2_code/ui/ldl -I../../../ctp2_code/ui/ldl -I../../../ctp2_code/gs/events -I../../../ctp2_code/gs/newdb -I../../../ctp2_code/gs/newdb -I../../../ctp2_code/ai/diplomacy -I../../../ctp2_code/ai/mapanalysis -I../../../ctp2_code/ai/strategy/scheduler -I../../../ctp2_code/ai/strategy/agents -I../../../ctp2_code/ai/strategy/goals -I../../../ctp2_code/ai/strategy/squads -I../../../ctp2_code/ai -I../../../ctp2_code/ai/CityManagement -I/usr/local/include/SDL -D_GNU_SOURCE=1 -D_REENTRANT -I/usr/X11R6/include -Wall -Wno-unused-variable -fms-extensions -fmessage-length=0 -frtti -fexceptions -g -O2 -fomit-frame-pointer -MT tiledraw.lo -MD -MP -MF .deps/tiledraw.Tpo -c tiledraw.cpp -fPIC -DPIC -fleading-underscore -o .libs/tiledraw.o
void TiledMap::DrawTransitionTileScaled(aui_Surface *surface, const MapPoint &pos, sint32 x, sint32 y, sint32 destWidth, sint32 destHeight) {
Pixel16 *dataPtr;
sint32 x, y;
sint32 startX, endX;
TileInfo *tileInfo;
BaseTile *baseTile, *transitionBuffer;
uint16 index;
Pixel16 *transData, *transDataPtr;
static Pixel16 defaultPixel[4] = {0xf800, 0x07e0, 0x001f, 0xf81f};
if (!surface) surface = m_surface;
ypos+=k_TILE_PIXEL_HEADROOM;
if (xpos < 0)
return;
if (xpos > surface->Width() - k_TILE_PIXEL_WIDTH)
return;
if (ypos < 0)
return;
if (ypos > surface->Height() - k_TILE_PIXEL_HEIGHT)
return;
tileInfo = GetTileInfo(pos);
Assert(tileInfo != NULL);
if (tileInfo == NULL)
return;
index = tileInfo->GetTileNum();
baseTile = m_tileSet->GetBaseTile(index);
if (baseTile == NULL)
return;
Pixel16 *data = baseTile->GetTileData();
Pixel16 *tileData[4];
sint32 tilesetIndex = g_theTerrainDB->Get(tileInfo->GetTerrainType())->GetTilesetIndex();
uint16 tilesetIndex_short = (uint16) tilesetIndex;
#ifdef _DEBUG
Assert(tilesetIndex == ((sint32) tilesetIndex_short));
#endif
tileData[0] = m_tileSet->GetTransitionData(tilesetIndex_short, tileInfo->GetTransition(0), 0);
tileData[1] = m_tileSet->GetTransitionData(tilesetIndex_short, tileInfo->GetTransition(1), 1);
tileData[2] = m_tileSet->GetTransitionData(tilesetIndex_short, tileInfo->GetTransition(2), 2);
tileData[3] = m_tileSet->GetTransitionData(tilesetIndex_short, tileInfo->GetTransition(3), 3);
transitionBuffer = m_tileSet->GetBaseTile(static_cast<uint16>((tilesetIndex * 100) + 99));
if(transitionBuffer) {
transData = transitionBuffer->GetTileData();
transDataPtr = transData;
} else {
transData = NULL;
transDataPtr = NULL;
}
dataPtr = data;
uint8 *pSurfBase;
pSurfBase = m_surfBase;
sint32 surfWidth = m_surfWidth;
sint32 surfHeight = m_surfHeight;
sint32 surfPitch = m_surfPitch;
Pixel16 srcPixel, transPixel = 0;
uint16 *pDestPixel = (Pixel16 *)(pSurfBase + ypos * surfPitch + 2 * xpos);
{
for (y=0; y<k_TILE_PIXEL_HEIGHT; y++) {
if (y<=23) {
startX = (23-y)*2;
endX = k_TILE_PIXEL_WIDTH - startX;
} else {
startX = (y-24)*2;
endX = k_TILE_PIXEL_WIDTH - startX;
}
if (transDataPtr)
{
#ifdef _MSC_VER //use this if __asm__ is used
_asm {
mov edx, endX
mov edi, pDestPixel
mov esi, dataPtr
mov ecx, startX
lea edi, [edi + 2*edx]
sub ecx, edx
mov ebx, transDataPtr
L0:
mov dx, [esi]
add esi, 2
xor eax, eax
mov ax, dx
cmp eax, 4
jge L1
mov edx, tileData[4*eax]
test edx, edx
jz L2
add edx, 2
mov tileData[4*eax], edx
mov dx, [edx-2]
jmp L1
L2:
mov dx, [ebx]
L1:
add ebx, 2
mov [edi + 2*ecx], dx
inc ecx
jnz L0
mov transDataPtr, ebx
mov dataPtr, esi
}
#else
__asm__ (
// "movl $endX, %edx \n\t" //done by gcc!
// "movl $pDestPixel, %edi \n\t" //done by gcc!
// "movl $dataPtr, %esi \n\t" //done by gcc!
// "movl $startX, %ecx \n\t" //done by gcc!
"leal (%2,%3,2),%2 \n\t"//load value %0 + %1 * s is pointing to
"subl %3,%4 \n\t" //%1 is now reuseable
// "movl $transDataPtr, %ebx \n\t" //done by gcc!
".L0: \n\t"
"movw (%1),%%dx \n\t" //reusing edx (%3)
"addl $2,%1 \n\t"
// "movl %%eax, %4 \n\t"
"movl %0, %%ebp \n\t"
"pushl %%ebx \n\t" // save reg for PIC!
"xorl %%ebx,%%ebx \n\t" //make %eax = 0, see below, we use ebx
"movw %%dx,%%bx \n\t" //because of this ebx has to be static
"cmpl $4,%%ebx \n\t" //because of above line eax may be != 0
"jge .L1 \n\t"
"movl (%%eax,%%ebx,4), %3 \n\t" // ** tileData can't be passed as "m"!
"testl %3,%3 \n\t" //check if %3 is zero, set Z-bit
"jz .L2 \n\t"
"addl $2,%3 \n\t"
"movl %3, (%%eax,%%ebx,4) \n\t" // (%eax + %ebx * 4) = %3
"movw -2(%3),%%dx \n\t"
"jmp .L1 \n\t"
".L2: \n\t"
"movw (%%ebp),%%dx \n\t"//movw (%3),%%dx
".L1: \n\t"
"addl $2,%%ebp \n\t"
"movw %%dx,(%2,%4,2) \n\t"
"incl %4 \n\t"
"jnz .L0 \n\t"
"popl %%ebx \n\t" // restore reg for PIC!
"movl %%ebp, %0 \n\t"
// "movl %esi, $dataPtr \n\t" //does gcc
: "=m" (transDataPtr), "=S" (dataPtr) // "=D" (pDestPixel)
: "D" (pDestPixel), "d" (endX), "c" (startX), "0" (transDataPtr), "1" (dataPtr), "a" (tileData)
: "%ebp", "cc" //don't tell gcc about what was done to ebx!!!
);
#endif //use this if __asm__ is used
}
else
{
for (x = startX; x<endX; x++)
{
srcPixel = *dataPtr++;
if (srcPixel < 4)
{
Pixel16 *tile = tileData[srcPixel];
if (tile != NULL)
{
tileData[srcPixel]++;
srcPixel = *tile;
}
else
{
srcPixel = defaultPixel[srcPixel];
}
}
pDestPixel[x] = srcPixel;
}
}
pDestPixel += (surfPitch>>1);
}
}
}
