Calculate staging buffer maximum size in Picture_Convert

This commit is contained in:
Eidolon authored and Alug committed 2025-10-13 18:18:05 +02:00
1 parent c3547812de
commit 94fdae6f21
1 file changed
+20 -8
+20 -8
View File
@@ -458,17 +458,23 @@ static UINT8 *imgbuf = NULL;
void *R_PixelsToPatch(UINT8 *raw, INT16 width, INT16 height, INT16 leftoffset, INT16 topoffset, size_t *destsize)
{
INT16 x, y;
UINT8 *img;
UINT8 *img, *imgptr;
UINT8 *colpointers, *startofspan;
size_t size = 0;
if (!raw)
return NULL;
if (!imgbuf)
imgbuf = Z_Malloc(1<<26, PU_STATIC, NULL);
// Allocate a staging buffer with the maximum size needed for a patch of the same size as the input.
UINT8 *imgptr = imgbuf;
// round up to nearest multiple of 254-pixel posts, plus 1 more 254-pixel post for paranoia reasons
size_t maxcolumnsize = (2 + (height - 1) / 256) * 256;
// the patch header, and width columns of the max column size
size_t maxoutsize = maxcolumnsize * width + (8 + 4 * width);
// so, a 512x512 flat should maximally need 393,760 (384.53 KiB) bytes.
// quite a bit smaller than 64 megabytes, and much less annoying to the windows debug allocator!
imgbuf = Z_Malloc(maxoutsize, PU_STATIC, NULL);
imgptr = imgbuf;
// Write image size and offset
WRITEINT16(imgptr, width);
@@ -582,17 +588,23 @@ void *R_PixelsToPatch(UINT8 *raw, INT16 width, INT16 height, INT16 leftoffset, I
void *R_MaskedFlatToPatch(UINT16 *raw, INT16 width, INT16 height, INT16 leftoffset, INT16 topoffset, size_t *destsize)
{
INT16 x, y;
UINT8 *img;
UINT8 *img, *imgptr;
UINT8 *colpointers, *startofspan;
size_t size = 0;
if (!raw)
return NULL;
if (!imgbuf)
imgbuf = Z_Malloc(1<<26, PU_STATIC, NULL);
// Allocate a staging buffer with the maximum size needed for a patch of the same size as the input.
UINT8 *imgptr = imgbuf;
// round up to nearest multiple of 254-pixel posts, plus 1 more 254-pixel post for paranoia reasons
size_t maxcolumnsize = (2 + (height - 1) / 256) * 256;
// the patch header, and width columns of the max column size
size_t maxoutsize = maxcolumnsize * width + (8 + 4 * width);
// so, a 512x512 flat should maximally need 393,760 (384.53 KiB) bytes.
// quite a bit smaller than 64 megabytes, and much less annoying to the windows debug allocator!
imgbuf = Z_Malloc(maxoutsize, PU_STATIC, NULL);
imgptr = imgbuf;
// Write image size and offset
WRITEINT16(imgptr, width);