[PATCH] D24662: [Sparc][LEON] Hardware erratum fix: Insert NOPs around double precision instructions
Daniel Cederman via llvm-commits
llvm-commits at lists.llvm.org
Fri Sep 30 07:20:25 PDT 2016
dcederman added a subscriber: llvm-commits.
dcederman added inline comments.
> LeonPasses.cpp:281
> +// This erratum fix for some earlier LEON processors fixes a problem where a
> +// double precision load will not yield the correct result if used in FMUL,
> +// FDIV, FADD, FSUB or FSQRT instructions later. If this sequence is detected,
It should be a single precision load according to the errata document.
From http://www.atmel.com/Images/doc4409.pdf :
"Data dependency is not properly checked between a load singleword floating - point instruction (LDF) involving an odd-numbered floating-point register as a destination of the load and an immediately following double-precision floating-point instruction".
> LeonPasses.cpp:338-339
> + Modified = true;
> + }
> + }
> + } else if (NextOpcode == SP::FSQRTD) {
The indentation looks wrong here. You can use clang format to automatically indent and format your patch.
Repository:
rL LLVM
https://reviews.llvm.org/D24662
More information about the llvm-commits
mailing list