AuroraMiddleware/skia2
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Muck with the right cinfo.

Browse Source 
In SkImageDecoder_libjpeg, modify cinfo *after* destroying it and
creating a new one in its place.

Should fix build breakage.

Committed: https://skia.googlesource.com/skia/+/3629865bac20ae8092177f519594f79f89d09fb0

Review URL: https://codereview.chromium.org/858333002
This commit is contained in:
scroggo 2015-01-22 10:27:25 -08:00 committed by Commit bot
parent 8dd688b756
commit 6d7e47a8e0
1 changed files with 5 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
src/images/SkImageDecoder_libjpeg.cpp
View File

@ -940,10 +940,9 @@ bool SkJPEGImageDecoder::onDecodeYUV8Planes(SkStream* stream, SkISize componentS
bool SkJPEGImageDecoder::onBuildTileIndex(SkStreamRewindable* stream, int *width, int *height) {
SkAutoTDelete<SkJPEGImageIndex> imageIndex(SkNEW_ARGS(SkJPEGImageIndex, (stream, this)));
jpeg_decompress_struct* cinfo = imageIndex->cinfo();
skjpeg_error_mgr sk_err;
set_error_mgr(cinfo, &sk_err);
set_error_mgr(imageIndex->cinfo(), &sk_err);
// All objects need to be instantiated before this setjmp call so that
// they will be cleaned up properly if an error occurs.
@ -968,6 +967,10 @@ bool SkJPEGImageDecoder::onBuildTileIndex(SkStreamRewindable* stream, int *width
return false;
}
jpeg_decompress_struct* cinfo = imageIndex->cinfo();
// We have a new cinfo, so set the error mgr again.
set_error_mgr(cinfo, &sk_err);
// FIXME: This sets cinfo->out_color_space, which we may change later
// based on the config in onDecodeSubset. This should be fine, since
// jpeg_init_read_tile_scanline will check out_color_space again after