 @@ -89,27 +89,27 @@ typedef struct {
 	CloseScreenProcPtr CloseScreen;
 	CreateScreenResourcesProcPtr CreateScreenResources;
 	OptionInfoPtr	Options;
 	xf86CursorInfoPtr	CursorInfoRec;
 	/* SX accel stuff */
 	void		*sxreg, *sxio;
 	int		use_accel, use_xrender;
 	uint32_t	last_mask;
 	uint32_t	last_rop;
 	uint32_t	srcoff, srcpitch, mskoff, mskpitch;
 	uint32_t	srcformat, dstformat, mskformat;
 	uint32_t	fillcolour;
 	int		op;
-	Bool		source_is_solid;
+	Bool		source_is_solid, no_source_pixmap;
 	int		xdir, ydir;
 	ExaDriverPtr 	pExa;
 } Cg14Rec, *Cg14Ptr;
 /* SX stuff */
 /* write an SX register */
 static inline void
 write_sx_reg(Cg14Ptr p, int reg, uint32_t val)
+{
 	*(volatile uint32_t *)(p->sxreg + reg) = val;
+}
 /* read an SX register */

 @@ -1,14 +1,14 @@
-/* $NetBSD: cg14_accel.c,v 1.4 2013/07/03 02:05:52 macallan Exp $ */
+/* $NetBSD: cg14_accel.c,v 1.5 2013/07/03 15:29:34 macallan Exp $ */
 /*
  * Copyright (c) 2013 Michael Lorenz
  * All rights reserved.
+ *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions
  * are met:
+ *
  *    - Redistributions of source code must retain the above copyright
  *      notice, this list of conditions and the following disclaimer.
  *    - Redistributions in binary form must reproduce the above
  *      copyright notice, this list of conditions and the following
  *      disclaimer in the documentation and/or other materials provided
 @@ -573,65 +573,101 @@ CG14CheckComposite(int op, PicturePtr pS
 Bool
 CG14PrepareComposite(int op, PicturePtr pSrcPicture,
                              PicturePtr pMaskPicture,
                              PicturePtr pDstPicture,
                              PixmapPtr  pSrc,
                              PixmapPtr  pMask,
                              PixmapPtr  pDst)
+{
 	ScrnInfoPtr pScrn = xf86Screens[pDst->drawable.pScreen->myNum];
 	Cg14Ptr p = GET_CG14_FROM_SCRN(pScrn);
 	ENTER;
 	p->no_source_pixmap = FALSE;
 	p->source_is_solid = FALSE;
 	if (pSrcPicture->format == PICT_a1) {
 		xf86Msg(X_ERROR, "src mono, dst %x, op %d\n",
 		    pDstPicture->format, op);
 		if (pMaskPicture != NULL) {
 			xf86Msg(X_ERROR, "msk %x\n", pMaskPicture->format);
+		}
+	}
 	if (pSrcPicture->pSourcePict != NULL) {
 		if (pSrcPicture->pSourcePict->type == SourcePictTypeSolidFill) {
 			p->fillcolour =
 			    pSrcPicture->pSourcePict->solidFill.color;
-			xf86Msg(X_ERROR, "%s: solid src %08x\n",
+			DPRINTF(X_ERROR, "%s: solid src %08x\n",
 			    __func__, p->fillcolour);
 			p->no_source_pixmap = TRUE;
 			p->source_is_solid = TRUE;
+		}
+	}
 	if ((pMaskPicture != NULL) && (pMaskPicture->pSourcePict != NULL)) {
 		if (pMaskPicture->pSourcePict->type ==
 		    SourcePictTypeSolidFill) {
 			p->fillcolour =
 			   pMaskPicture->pSourcePict->solidFill.color;
 			xf86Msg(X_ERROR, "%s: solid mask %08x\n",
 			    __func__, p->fillcolour);
+		}
+	}
 	if (pMaskPicture != NULL) {
 		p->mskoff = exaGetPixmapOffset(pMask);
 		p->mskpitch = exaGetPixmapPitch(pMask);
 		p->mskformat = pMaskPicture->format;
 	} else {
 		p->mskoff = 0;
 		p->mskpitch = 0;
 		p->mskformat = 0;
+	}
-	p->source_is_solid =
+	if (pSrc != NULL) {
-	   ((pSrc->drawable.width == 1) && (pSrc->drawable.height == 1));
+		p->source_is_solid =
-	p->srcoff = exaGetPixmapOffset(pSrc);
+		   ((pSrc->drawable.width == 1) && (pSrc->drawable.height == 1));
-	p->srcpitch = exaGetPixmapPitch(pSrc);
+		p->srcoff = exaGetPixmapOffset(pSrc);
 		p->srcpitch = exaGetPixmapPitch(pSrc);
 		if (p->source_is_solid) {
 			p->fillcolour = *(uint32_t *)(p->fb + p->srcoff);
+		}
+	}
 	p->srcformat = pSrcPicture->format;
 	p->dstformat = pDstPicture->format;
 	if (p->source_is_solid) {
 		uint32_t temp;
 		/* stuff source colour into SX registers, swap as needed */
 		temp = p->fillcolour;
 		switch (p->srcformat) {
 			case PICT_a8r8g8b8:
 			case PICT_x8r8g8b8:
 				write_sx_reg(p, SX_QUEUED(9), temp & 0xff);
 				temp = temp >> 8;
 				write_sx_reg(p, SX_QUEUED(10), temp & 0xff);
 				temp = temp >> 8;
 				write_sx_reg(p, SX_QUEUED(11), temp & 0xff);
 				break;
 			case PICT_a8b8g8r8:
 			case PICT_x8b8g8r8:
 				write_sx_reg(p, SX_QUEUED(11), temp & 0xff);
 				temp = temp >> 8;
 				write_sx_reg(p, SX_QUEUED(10), temp & 0xff);
 				temp = temp >> 8;
 				write_sx_reg(p, SX_QUEUED(9), temp & 0xff);
 				break;
+		}
 		write_sx_reg(p, SX_QUEUED(8), 0xff);
+	}
 	p->op = op;
 	if (op == PictOpSrc) {
 		CG14PrepareCopy(pSrc, pDst, 1, 1, GXcopy, 0xffffffff);
+	}
 #ifdef SX_DEBUG
 	DPRINTF(X_ERROR, "%x %x -> %x\n", p->srcoff, p->mskoff,
 	    *(uint32_t *)(p->fb + p->srcoff));
 #endif
 	return TRUE;
+}
 void
 CG14Composite(PixmapPtr pDst, int srcX, int srcY,

 @@ -1,14 +1,14 @@
-/* $NetBSD: cg14_render.c,v 1.3 2013/07/03 02:05:52 macallan Exp $ */
+/* $NetBSD: cg14_render.c,v 1.4 2013/07/03 15:29:34 macallan Exp $ */
 /*
  * Copyright (c) 2013 Michael Lorenz
  * All rights reserved.
+ *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions
  * are met:
+ *
  *    - Redistributions of source code must retain the above copyright
  *      notice, this list of conditions and the following disclaimer.
  *    - Redistributions in binary form must reproduce the above
  *      copyright notice, this list of conditions and the following
  *      disclaimer in the documentation and/or other materials provided
 @@ -53,29 +53,27 @@
 char c[8] = " .,:+*oX";
 void CG14Comp_Over32Solid(Cg14Ptr p,
                    uint32_t src, uint32_t srcpitch,
                    uint32_t dst, uint32_t dstpitch,
                    int width, int height)
+{
 	uint32_t msk = src, mskx, dstx, m;
 	int line, x, i;
 	ENTER;
 	/* first get the source colour */
 	write_sx_io(p, p->srcoff, SX_LDUQ0(8, 0, p->srcoff & 7));
 	write_sx_reg(p, SX_QUEUED(8), 0xff);
 	for (line = 0; line < height; line++) {
 		mskx = msk;
 		dstx = dst;
 #ifdef SX_SINGLE
 		for (x = 0; x < width; x++) {
 			m = *(volatile uint32_t *)(p->fb + mskx);
 			m = m >> 24;
 			if (m == 0) {
 				/* nothing to do - all transparent */
 			} else if (m == 0xff) {
 				/* all opaque */
 				write_sx_io(p, dstx, SX_STUQ0(8, 0, dstx & 7));
 @@ -154,29 +152,26 @@ void CG14Comp_Over32Solid(Cg14Ptr p,
 void CG14Comp_Over8Solid(Cg14Ptr p,
                    uint32_t src, uint32_t srcpitch,
                    uint32_t dst, uint32_t dstpitch,
                    int width, int height)
+{
 	uint32_t msk = src, mskx, dstx, m;
 	int line, x, i;
 #ifdef SX_DEBUG
 	char buffer[256];
 #endif
 	ENTER;
 	/* first get the source colour */
 	write_sx_io(p, p->srcoff, SX_LDUQ0(8, 0, p->srcoff & 7));
 	write_sx_reg(p, SX_QUEUED(8), 0xff);
 	DPRINTF(X_ERROR, "src: %d %d %d, %08x\n", read_sx_reg(p, SX_QUEUED(9)),
 	    read_sx_reg(p, SX_QUEUED(10)), read_sx_reg(p, SX_QUEUED(11)),
 	    *(uint32_t *)(p->fb + p->srcoff));
 	for (line = 0; line < height; line++) {
 		mskx = msk;
 		dstx = dst;
 #ifdef SX_SINGLE
 		for (x = 0; x < width; x++) {
 			m = *(volatile uint8_t *)(p->fb + mskx);
 #ifdef SX_DEBUG
 			buffer[x] = c[m >> 5];
 #endif
 @@ -437,30 +432,33 @@ void CG14Comp_Over32Mask(Cg14Ptr p,
 	write_sx_reg(p, SX_QUEUED(8), 0xff);
 	for (line = 0; line < height; line++) {
 		srcx = src;
 		mskx = msk;
 		dstx = dst;
 		for (x = 0; x < width; x++) {
 			/* fetch source pixel */
 			write_sx_io(p, srcx, SX_LDUQ0(12, 0, srcx & 7));
 			/* fetch mask */
 			write_sx_io(p, mskx & (~7), SX_LDB(9, 0, mskx & 7));
 			/* fetch dst pixel */
 			write_sx_io(p, dstx, SX_LDUQ0(20, 0, dstx & 7));
-			/* apply mask */
+			/* stick mask alpha into SCAM */
 			write_sx_reg(p, SX_INSTRUCTIONS,
-			    SX_ANDS(12, 9, 16, 3));
+			    SX_ORS(9, 0, R_SCAM, 0));
 			/* apply mask */
 			/* src is premultiplied with alpha */
 			write_sx_reg(p, SX_INSTRUCTIONS,
 			    SX_SAXP16X16SR8(12, 0, 16, 3));
 			/* write inverted alpha into SCAM */
 			write_sx_reg(p, SX_INSTRUCTIONS,
 			    SX_XORV(16, 8, R_SCAM, 0));
 			/* dst * (1 - alpha) + R[13:15] */
 			write_sx_reg(p, SX_INSTRUCTIONS,
 			    SX_SAXP16X16SR8(21, 17, 25, 2));
 			write_sx_io(p, dstx,
 			    SX_STUQ0C(24, 0, dstx & 7));
 			srcx += 4;
 			mskx += 1;
 			dstx += 4;
+		}
 		src += srcpitch;