We can't infer the right type for voidp for old png_jmpbuf()

In libpng 1.2 it's just a macro that's ->jmpbuf,
so there's nothing forcing the conversion to png_structp.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2268953005

Review-Url: https://codereview.chromium.org/2268953005
This commit is contained in:
mtklein 2016-08-24 14:46:35 -07:00 committed by Commit bot
parent e59cb1c1f1
commit f84960187a

View File

@ -420,7 +420,7 @@ public:
// Assume that an error in libpng indicates an incomplete input.
int y = 0;
if (setjmp(png_jmpbuf(fPng_ptr))) {
if (setjmp(png_jmpbuf((png_struct*)fPng_ptr))) {
SkCodecPrintf("Failed to read row.\n");
return y;
}
@ -458,7 +458,7 @@ public:
}
bool onSkipScanlines(int count) override {
if (setjmp(png_jmpbuf(fPng_ptr))) {
if (setjmp(png_jmpbuf((png_struct*)fPng_ptr))) {
SkCodecPrintf("Failed to skip row.\n");
return false;
}
@ -499,7 +499,7 @@ public:
int readRows(const SkImageInfo& dstInfo, void* dst, size_t rowBytes, int count, int startRow)
override {
if (setjmp(png_jmpbuf(fPng_ptr))) {
if (setjmp(png_jmpbuf((png_struct*)fPng_ptr))) {
SkCodecPrintf("Failed to get scanlines.\n");
// FIXME (msarett): Returning 0 is pessimistic. If we can complete a single pass,
// we may be able to report that all of the memory has been initialized. Even if we
@ -810,7 +810,7 @@ void SkPngCodec::destroyReadStruct() {
bool SkPngCodec::initializeXforms(const SkImageInfo& dstInfo, const Options& options,
SkPMColor ctable[], int* ctableCount) {
if (setjmp(png_jmpbuf(fPng_ptr))) {
if (setjmp(png_jmpbuf((png_struct*)fPng_ptr))) {
SkCodecPrintf("Failed on png_read_update_info.\n");
return false;
}