Attachment 37900 Details for Bug 27139 – Avoid CPU reads of VRAM

[patch] Avoid CPU reads of VRAM

0.6.13.1-avoid-vram-read.patch (text/plain), 9.90 KB, created by Karl Tomlinson on 2010-08-16 07:34:49 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Karl Tomlinson

Created: 2010-08-16 07:34:49 UTC

Size: 9.90 KB

patch

obsolete

>--- xf86-video-ati-6.13.1/src/radeon_exa.c.6.13.1	2010-08-14 22:59:28.000000000 +1200
>+++ xf86-video-ati-6.13.1/src/radeon_exa.c	2010-08-16 17:07:10.000000000 +1200
>@@ -310,31 +310,56 @@ static void RADEONFinishAccess_BE(Pixmap
> 
> #ifdef XF86DRM_MODE
> Bool RADEONPrepareAccess_CS(PixmapPtr pPix, int index)
> {
>     ScreenPtr pScreen = pPix->drawable.pScreen;
>     ScrnInfoPtr pScrn = xf86Screens[pScreen->myNum];
>     RADEONInfoPtr info = RADEONPTR(pScrn);
>     struct radeon_exa_pixmap_priv *driver_priv;
>+    uint32_t possible_domains = ~0;
>+    uint32_t current_domain = 0;
>+    Bool can_fail = pPix->drawable.bitsPerPixel >= 8 &&
>+	pPix != pScreen->GetScreenPixmap(pScreen);
>+    Bool flush = FALSE;
>     int ret;
> 
> #if X_BYTE_ORDER == X_BIG_ENDIAN
>     /* May need to handle byte swapping in DownloadFrom/UploadToScreen */
>-    if (pPix->drawable.bitsPerPixel > 8 &&
>-	pPix != pScreen->GetScreenPixmap(pScreen))
>+    if (can_fail && pPix->drawable.bitsPerPixel > 8)
> 	return FALSE;
> #endif
> 
>     driver_priv = exaGetPixmapDriverPrivate(pPix);
>     if (!driver_priv)
>       return FALSE;
> 
>     /* if we have more refs than just the BO then flush */
>-    if (radeon_bo_is_referenced_by_cs(driver_priv->bo, info->cs))
>+    if (radeon_bo_is_referenced_by_cs(driver_priv->bo, info->cs)) {
>+	flush = TRUE;
>+
>+	/* if the BO might end up in VRAM, prefer DownloadFromScreen */
>+	if (can_fail) {
>+	    possible_domains = radeon_bo_get_src_domain(driver_priv->bo);
>+	    if (possible_domains == RADEON_GEM_DOMAIN_VRAM)
>+		return FALSE;
>+	}
>+    }
>+
>+    if (can_fail) {
>+	radeon_bo_is_busy(driver_priv->bo, &current_domain);
>+
>+	if (current_domain & possible_domains) {
>+	    if (current_domain == RADEON_GEM_DOMAIN_VRAM)
>+		return FALSE;
>+	} else if (possible_domains & RADEON_GEM_DOMAIN_VRAM)
>+	    return FALSE;
>+    }
>+
>+    if (flush)
>         radeon_cs_flush_indirect(pScrn);
>     
>     /* flush IB */
>     ret = radeon_bo_map(driver_priv->bo, 1);
>     if (ret) {
>       FatalError("failed to map pixmap %d\n", ret);
>       return FALSE;
>     }
>--- xf86-video-ati-6.13.1/src/radeon_exa_funcs.c.6.13.1	2010-08-14 22:59:40.000000000 +1200
>+++ xf86-video-ati-6.13.1/src/radeon_exa_funcs.c	2010-08-16 15:25:39.000000000 +1200
>@@ -454,25 +454,28 @@ RADEONBlitChunk(ScrnInfoPtr pScrn, struc
> #if defined(XF86DRM_MODE)
> static Bool
> RADEONUploadToScreenCS(PixmapPtr pDst, int x, int y, int w, int h,
> 		       char *src, int src_pitch)
> {
>     ScreenPtr pScreen = pDst->drawable.pScreen;
>     RINFO_FROM_SCREEN(pScreen);
>     struct radeon_exa_pixmap_priv *driver_priv;
>-    struct radeon_bo *scratch;
>+    struct radeon_bo *scratch = NULL;
>+    struct radeon_bo *copy_dst;
>     unsigned char *dst;
>     unsigned size;
>     uint32_t datatype = 0;
>     uint32_t dst_domain;
>     uint32_t dst_pitch_offset;
>     unsigned bpp = pDst->drawable.bitsPerPixel;
>     uint32_t scratch_pitch = RADEON_ALIGN(w * bpp / 8, 64);
>+    uint32_t copy_pitch = pDst->devKind;
>     uint32_t swap = RADEON_HOST_DATA_SWAP_NONE;
>+    Bool flush = TRUE;
>     Bool r;
>     int i;
> 
>     if (bpp < 8)
> 	return FALSE;
> 
>     driver_priv = exaGetPixmapDriverPrivate(pDst);
>     if (!driver_priv || !driver_priv->bo)
>@@ -484,94 +487,94 @@ RADEONUploadToScreenCS(PixmapPtr pDst, i
> 	swap = RADEON_HOST_DATA_SWAP_32BIT;
> 	break;
>     case 16:
> 	swap = RADEON_HOST_DATA_SWAP_16BIT;
> 	break;
>     }
> #endif
> 
>-    /* If we know the BO won't be busy, don't bother */
>-    if (!radeon_bo_is_referenced_by_cs(driver_priv->bo, info->cs) &&
>-	!radeon_bo_is_busy(driver_priv->bo, &dst_domain)) {
>-#if X_BYTE_ORDER == X_BIG_ENDIAN
>-	/* Can't return FALSE here if we need to swap bytes */
>-	if (swap != RADEON_HOST_DATA_SWAP_NONE &&
>-	    driver_priv->bo != info->front_bo) {
>-	    scratch = driver_priv->bo;
>-	    scratch_pitch = pDst->devKind;
>+    /* If we know the BO won't be busy, don't bother with a scratch*/
>+    copy_dst = driver_priv->bo;
>+    if (!radeon_bo_is_referenced_by_cs(driver_priv->bo, info->cs)) {
>+	flush = FALSE;
>+	if (!radeon_bo_is_busy(driver_priv->bo, &dst_domain))
> 	    goto copy;
>-	}
>-#endif
>-	return FALSE;
>     }
> 
>     size = scratch_pitch * h;
>     scratch = radeon_bo_open(info->bufmgr, 0, size, 0, RADEON_GEM_DOMAIN_GTT, 0);
>     if (scratch == NULL) {
>-	return FALSE;
>+	goto copy;
>     }
>     radeon_cs_space_reset_bos(info->cs);
>     radeon_add_pixmap(info->cs, pDst, 0, RADEON_GEM_DOMAIN_VRAM);
>     radeon_cs_space_add_persistent_bo(info->cs, scratch, RADEON_GEM_DOMAIN_GTT, 0);
>     r = radeon_cs_space_check(info->cs);
>     if (r) {
>-        r = FALSE;
>-        goto out;
>+	goto copy;
>     }
>+    flush = FALSE;
>+    copy_dst = scratch;
>+    copy_pitch = scratch_pitch;
> 
>-#if X_BYTE_ORDER == X_BIG_ENDIAN
> copy:
>-#endif
>-    r = radeon_bo_map(scratch, 0);
>+    if (flush)
>+	radeon_cs_flush_indirect(pScrn);
>+
>+    r = radeon_bo_map(copy_dst, 0);
>     if (r) {
>         r = FALSE;
>         goto out;
>     }
>     r = TRUE;
>     size = w * bpp / 8;
>-    dst = scratch->ptr;
>-    if (scratch == driver_priv->bo)
>-	dst += y * scratch_pitch + x * bpp / 8;
>+    dst = copy_dst->ptr;
>+    if (copy_dst == driver_priv->bo)
>+	dst += y * copy_pitch + x * bpp / 8;
>     for (i = 0; i < h; i++) {
>-        RADEONCopySwap(dst + i * scratch_pitch, (uint8_t*)src, size, swap);
>+        RADEONCopySwap(dst + i * copy_pitch, (uint8_t*)src, size, swap);
>         src += src_pitch;
>     }
>-    radeon_bo_unmap(scratch);
>+    radeon_bo_unmap(copy_dst);
> 
>-    if (scratch != driver_priv->bo) {
>+    if (copy_dst == scratch) {
> 	RADEONGetDatatypeBpp(pDst->drawable.bitsPerPixel, &datatype);
> 	RADEONGetPixmapOffsetPitch(pDst, &dst_pitch_offset);
> 	ACCEL_PREAMBLE();
> 	RADEON_SWITCH_TO_2D();
> 	RADEONBlitChunk(pScrn, scratch, driver_priv->bo, datatype, scratch_pitch << 16,
> 			dst_pitch_offset, 0, 0, x, y, w, h,
> 			RADEON_GEM_DOMAIN_GTT, RADEON_GEM_DOMAIN_VRAM);
>     }
> 
> out:
>-    if (scratch != driver_priv->bo)
>+    if (scratch)
> 	radeon_bo_unref(scratch);
>     return r;
> }
> 
> static Bool
> RADEONDownloadFromScreenCS(PixmapPtr pSrc, int x, int y, int w,
>                            int h, char *dst, int dst_pitch)
> {
>     RINFO_FROM_SCREEN(pSrc->drawable.pScreen);
>     struct radeon_exa_pixmap_priv *driver_priv;
>-    struct radeon_bo *scratch;
>+    struct radeon_bo *scratch = NULL;
>+    struct radeon_bo *copy_src;
>     unsigned size;
>     uint32_t datatype = 0;
>     uint32_t src_domain = 0;
>     uint32_t src_pitch_offset;
>     unsigned bpp = pSrc->drawable.bitsPerPixel;
>     uint32_t scratch_pitch = RADEON_ALIGN(w * bpp / 8, 64);
>+    uint32_t copy_pitch = pSrc->devKind;
>     uint32_t swap = RADEON_HOST_DATA_SWAP_NONE;
>+    int ret;
>+    Bool flush = FALSE;
>     Bool r;
> 
>     if (bpp < 8)
> 	return FALSE;
> 
>     driver_priv = exaGetPixmapDriverPrivate(pSrc);
>     if (!driver_priv || !driver_priv->bo)
> 	return FALSE;
>@@ -582,84 +585,81 @@ RADEONDownloadFromScreenCS(PixmapPtr pSr
> 	swap = RADEON_HOST_DATA_SWAP_32BIT;
> 	break;
>     case 16:
> 	swap = RADEON_HOST_DATA_SWAP_16BIT;
> 	break;
>     }
> #endif
> 
>-    /* If we know the BO won't end up in VRAM anyway, don't bother */
>+    /* If we know the BO won't end up in VRAM anyway, don't bother with a scratch */
>+    copy_src = driver_priv->bo;
>     if (radeon_bo_is_referenced_by_cs(driver_priv->bo, info->cs)) {
> 	src_domain = radeon_bo_get_src_domain(driver_priv->bo);
> 	if ((src_domain & (RADEON_GEM_DOMAIN_GTT | RADEON_GEM_DOMAIN_VRAM)) ==
> 	    (RADEON_GEM_DOMAIN_GTT | RADEON_GEM_DOMAIN_VRAM))
> 	    src_domain = 0;
>+	else /* A write may be scheduled */
>+	    flush = TRUE;
>     }
> 
>     if (!src_domain)
> 	radeon_bo_is_busy(driver_priv->bo, &src_domain);
> 
>-    if (src_domain != RADEON_GEM_DOMAIN_VRAM) {
>-#if X_BYTE_ORDER == X_BIG_ENDIAN
>-	/* Can't return FALSE here if we need to swap bytes */
>-	if (swap != RADEON_HOST_DATA_SWAP_NONE) {
>-	    scratch = driver_priv->bo;
>-	    scratch_pitch = pSrc->devKind;
>-	    goto copy;
>-	}
>-#endif
>-	return FALSE;
>+    if (src_domain & ~RADEON_GEM_DOMAIN_VRAM) {
>+	goto copy;
>     }
> 
>     size = scratch_pitch * h;
>     scratch = radeon_bo_open(info->bufmgr, 0, size, 0, RADEON_GEM_DOMAIN_GTT, 0);
>     if (scratch == NULL) {
>-	return FALSE;
>+	goto copy;
>     }
>     radeon_cs_space_reset_bos(info->cs);
>     radeon_add_pixmap(info->cs, pSrc, RADEON_GEM_DOMAIN_GTT | RADEON_GEM_DOMAIN_VRAM, 0);
>     radeon_cs_space_add_persistent_bo(info->cs, scratch, 0, RADEON_GEM_DOMAIN_GTT);
>-    r = radeon_cs_space_check(info->cs);
>-    if (r) {
>-        r = FALSE;
>-        goto out;
>+    ret = radeon_cs_space_check(info->cs);
>+    if (ret) {
>+        goto copy;
>     }
>     RADEONGetDatatypeBpp(pSrc->drawable.bitsPerPixel, &datatype);
>     RADEONGetPixmapOffsetPitch(pSrc, &src_pitch_offset);
>     ACCEL_PREAMBLE();
>     RADEON_SWITCH_TO_2D();
>     RADEONBlitChunk(pScrn, driver_priv->bo, scratch, datatype, src_pitch_offset,
>                     scratch_pitch << 16, x, y, 0, 0, w, h,
>                     RADEON_GEM_DOMAIN_VRAM | RADEON_GEM_DOMAIN_GTT,
>                     RADEON_GEM_DOMAIN_GTT);
>-    FLUSH_RING();
>-
>-#if X_BYTE_ORDER == X_BIG_ENDIAN
>+    flush = TRUE;
>+    copy_src = scratch;
>+    copy_pitch = scratch_pitch;
> copy:
>-#endif
>-    r = radeon_bo_map(scratch, 0);
>-    if (r) {
>+    if (flush)
>+	FLUSH_RING();
>+
>+    ret = radeon_bo_map(copy_src, 0);
>+    if (ret) {
>+	FatalError("failed to map pixmap %d\n", ret);
>         r = FALSE;
>         goto out;
>     }
>     r = TRUE;
>     w *= bpp / 8;
>-    if (scratch == driver_priv->bo)
>-	size = y * scratch_pitch + x * bpp / 8;
>+    if (copy_src == driver_priv->bo)
>+	size = y * copy_pitch + x * bpp / 8;
>     else
> 	size = 0;
>     while (h--) {
>-        RADEONCopySwap((uint8_t*)dst, scratch->ptr + size, w, swap);
>-        size += scratch_pitch;
>+        RADEONCopySwap((uint8_t*)dst, copy_src->ptr + size, w, swap);
>+        size += copy_pitch;
>         dst += dst_pitch;
>     }
>-    radeon_bo_unmap(scratch);
>+    radeon_bo_unmap(copy_src);
> out:
>-    if (scratch != driver_priv->bo)
>+    if (scratch)
> 	radeon_bo_unref(scratch);
>     return r;
> }
> #endif
> 
> static Bool
> RADEONDownloadFromScreenCP(PixmapPtr pSrc, int x, int y, int w, int h,
> 				    char *dst, int dst_pitch)

Actions: View | Splinter Review

Attachments on bug 27139: 34163 | 37900 | 38202 | 38203 | 38204 | 38205 | 38206 | 38207 | 38208

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.