本文整理汇总了C++中MInstructionIterator::toLsh方法的典型用法代码示例。如果您正苦于以下问题：C++ MInstructionIterator::toLsh方法的具体用法？C++ MInstructionIterator::toLsh怎么用？C++ MInstructionIterator::toLsh使用的例子？那么, 这里精选的方法代码示例或许可以为您提供帮助。您也可以进一步了解该方法所在类MInstructionIterator的用法示例。
在下文中一共展示了MInstructionIterator::toLsh方法的2个代码示例，这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞，您的评价将有助于系统推荐出更棒的C++代码示例。
示例1: AnalyzeLsh
// This analysis converts patterns of the form:
// truncate(x + (y << {0,1,2,3}))
// truncate(x + (y << {0,1,2,3}) + imm32)
// into a single lea instruction, and patterns of the form:
// asmload(x + imm32)
// asmload(x << {0,1,2,3})
// asmload((x << {0,1,2,3}) + imm32)
// asmload((x << {0,1,2,3}) & mask) (where mask is redundant with shift)
// asmload(((x << {0,1,2,3}) + imm32) & mask) (where mask is redundant with shift + imm32)
// into a single asmload instruction (and for asmstore too).
//
// Additionally, we should consider the general forms:
// truncate(x + y + imm32)
// truncate((y << {0,1,2,3}) + imm32)
bool
EffectiveAddressAnalysis::analyze()
{
for (ReversePostorderIterator block(graph_.rpoBegin()); block != graph_.rpoEnd(); block++) {
for (MInstructionIterator i = block->begin(); i != block->end(); i++) {
if (i->isLsh())
AnalyzeLsh(graph_.alloc(), i->toLsh());
}
}
return true;
}示例2: if
// This analysis converts patterns of the form:
// truncate(x + (y << {0,1,2,3}))
// truncate(x + (y << {0,1,2,3}) + imm32)
// into a single lea instruction, and patterns of the form:
// asmload(x + imm32)
// asmload(x << {0,1,2,3})
// asmload((x << {0,1,2,3}) + imm32)
// asmload((x << {0,1,2,3}) & mask) (where mask is redundant with shift)
// asmload(((x << {0,1,2,3}) + imm32) & mask) (where mask is redundant with shift + imm32)
// into a single asmload instruction (and for asmstore too).
//
// Additionally, we should consider the general forms:
// truncate(x + y + imm32)
// truncate((y << {0,1,2,3}) + imm32)
bool
EffectiveAddressAnalysis::analyze()
{
for (ReversePostorderIterator block(graph_.rpoBegin()); block != graph_.rpoEnd(); block++) {
for (MInstructionIterator i = block->begin(); i != block->end(); i++) {
// Note that we don't check for MAsmJSCompareExchangeHeap
// or MAsmJSAtomicBinopHeap, because the backend and the OOB
// mechanism don't support non-zero offsets for them yet.
if (i->isLsh())
AnalyzeLsh(graph_.alloc(), i->toLsh());
else if (i->isAsmJSLoadHeap())
analyzeAsmHeapAccess(i->toAsmJSLoadHeap());
else if (i->isAsmJSStoreHeap())
analyzeAsmHeapAccess(i->toAsmJSStoreHeap());
}
}
return true;
}