cogl: Don't assume that CoglBitmaps are allocated to height*rowstride

Cogl no longer has any code that assumes the buffer in a CoglBitmap is
allocated to the full size of height*rowstride. We should comment that
this is the case so that we remember to keep it that way. This is
important for cogl_texture_new_from_data because the application may
have created the data from a sub-region of a larger image and in that
case it's not safe to read the full rowstride of the last row when the
sub region contains the last row of the larger image.

http://bugzilla.clutter-project.org/show_bug.cgi?id=2491
This commit is contained in:
Neil Roberts 2010-12-17 16:30:23 +00:00
parent 0f43e01cbc
commit 4a7dbc0e4e
2 changed files with 17 additions and 51 deletions

View File

@ -198,13 +198,8 @@ _cogl_bitmap_from_file (const char *filename,
int width;
int height;
int rowstride;
int aligned_rowstride;
int bits_per_sample;
int n_channels;
guint8 *pixels;
guint8 *out_data;
guint8 *out;
int r;
g_return_val_if_fail (error == NULL || *error == NULL, FALSE);
@ -247,55 +242,18 @@ _cogl_bitmap_from_file (const char *filename,
return FALSE;
}
/* Work out what the rowstride would be if it was packing the data
but aligned to 4 bytes */
aligned_rowstride = (width * n_channels + 3) & ~3;
/* The documentation for GdkPixbuf states that is not safe to read
all of the data as height*rowstride because the buffer might not
be allocated to include the full length of the rowstride for the
last row so arguably we should always copy the buffer when
rowstride != width*bpp because some places in Cogl assume that it
can memcpy(height*rowstride). However that rule is probably only
in place so that GdkPixbuf can implement gdk_pixbuf_new_subpixbuf
by just sharing the data and setting a large rowstride. That does
not apply in this case because we are just creating a new buffer
for a file. It seems very unlikely that GdkPixbuf would not
allocate the full rowstride in this case and it is highly
desirable to avoid copying the buffer. This instead just assumes
that whatever buffer pixbuf points into will always be allocated
to a 4-byte aligned buffer so we can avoid copying unless the
rowstride is unusually large */
if (rowstride <= aligned_rowstride)
return _cogl_bitmap_new_from_data (gdk_pixbuf_get_pixels (pixbuf),
pixel_format,
width,
height,
rowstride,
_cogl_bitmap_unref_pixbuf,
pixbuf);
pixels = gdk_pixbuf_get_pixels (pixbuf);
out_data = g_malloc (aligned_rowstride * height);
out = out_data;
for (r = 0; r < height; ++r)
{
memcpy (out, pixels, n_channels * width);
pixels += rowstride;
out += aligned_rowstride;
}
/* Destroy GdkPixbuf object */
g_object_unref (pixbuf);
return _cogl_bitmap_new_from_data (out_data,
/* We just use the data directly from the pixbuf so that we don't
have to copy to a seperate buffer. Note that Cogl is expected not
to read past the end of bpp*width on the last row even if the
rowstride is much larger so we don't need to worry about
GdkPixbuf's semantics that it may under-allocate the buffer. */
return _cogl_bitmap_new_from_data (gdk_pixbuf_get_pixels (pixbuf),
pixel_format,
width,
height,
aligned_rowstride,
(CoglBitmapDestroyNotify) g_free,
NULL);
rowstride,
_cogl_bitmap_unref_pixbuf,
pixbuf);
}
#else

View File

@ -180,6 +180,14 @@ _cogl_bitmap_get_height (CoglBitmap *bitmap);
int
_cogl_bitmap_get_rowstride (CoglBitmap *bitmap);
/* Maps the bitmap so that the pixels can be accessed directly or if
the bitmap is just a memory bitmap then it just returns the pointer
to memory. Note that the bitmap isn't guaranteed to allocated to
the full size of rowstride*height so it is not safe to read up to
the rowstride of the last row. This will be the case if the user
uploads data using gdk_pixbuf_new_subpixbuf with a sub region
containing the last row of the pixbuf because in that case the
rowstride can be much larger than the width of the image */
guint8 *
_cogl_bitmap_map (CoglBitmap *bitmap,
CoglBufferAccess access,