On Mon, Dec 25, 2017 at 12:52:56AM +0100, Patrick Wildt wrote:
> Hi,
>
> gerhard@ found a bug in LLVM's i386 code. He noticed that some floating
> point instructions were "optimized" and thus reordered between other FP
> instructions that changed floating point modes, so that those reordered
> in
> Date: Mon, 25 Dec 2017 00:52:56 +0100
> From: Patrick Wildt
>
> Hi,
>
> gerhard@ found a bug in LLVM's i386 code. He noticed that some floating
> point instructions were "optimized" and thus reordered between other FP
> instructions that changed floating point modes, so that those reordered
>
Hi,
gerhard@ found a bug in LLVM's i386 code. He noticed that some floating
point instructions were "optimized" and thus reordered between other FP
instructions that changed floating point modes, so that those reordered
instructions were running with an incorrent mode.
I don't completely grasp t