我有一个很大的jpeg图像,我想在我的opengl引擎中异步加载到tile中。 如果它在主线程上完成但速度很慢,那么一切都很好。
当我尝试将tile加载到NSOperationBlock上时,它总是在尝试访问我之前在主线程上加载的共享图像数据指针时崩溃。
必须有一些我无法通过后台操作得到的东西,因为我假设我可以访问我在主线程上创建的内存部分。
我尝试做的是以下内容:
@interface MyViewer
{
}
@property (atomic, assign) CGImageRef imageRef;
@property (atomic, assign) CGDataProviderRef dataProvider;
@property (atomic, assign) int loadedTextures;
@end
...
- (void) loadAllTiles:(NSData*) imgData
{
queue = [[NSOperationQueue alloc] init];
//Loop for Total Number of Textures
self.dataProvider = CGDataProviderCreateWithData(NULL,[imgData bytes],[imgData length],0);
self.imageRef = CGImageCreateWithJPEGDataProvider(self.dataProvider, NULL, NO, kCGRenderingIntentDefault);
for (int i=0; i<tileCount; i++)
{
// I also tried this but without luck
//CGImageRetain(self.imageRef);
//CGDataProviderRetain(self.dataProvider);
NSBlockOperation *partsLoading = [[NSBlockOperation alloc] init];
__weak NSBlockOperation *weakpartsLoadingOp = partsLoading;
[partsLoading addExecutionBlock: ^ {
TamTexture2D& pTex2D = viewer->getTile(i);
CGImageRef subImgRef = CGImageCreateWithImageInRect(self.imageRef, CGRectMake(pTex2D.left, pTex2D.top, pTex2D.width, pTex2D.height));
//!!!Its crashing here!!!
CFDataRef cgSubImgDataRef = CGDataProviderCopyData(CGImageGetDataProvider(subImgRef));
CGImageRelease(subImgRef);
...
}];
//Adding Parts loading on low priority thread. Is it all right ????
[partsLoading setThreadPriority:0.0];
[queue addOperation:partsLoading];
}
答案 0 :(得分:2)
我终于找到了我的问题......
我已阅读Quartz2D文档,似乎我们不应再使用CGDataProviderCreateWithData和CGImageCreateWithJPEGDataProvider了。我猜这种用法不是线程安全的。
根据文档的建议,我现在使用CGImageSource API:
self.imageSrcRef = CGImageSourceCreateWithData((__bridge CFDataRef)imgData, NULL);
// get imagePropertiesDictionary
CFDictionaryRef imagePropertiesDictionary = CGImageSourceCopyPropertiesAtIndex(m_imageSrcRef,0, NULL);
self.imageRef = CGImageSourceCreateImageAtIndex(m_imageSrcRef, 0, imagePropertiesDictionary);