[PATCH, PR69707] Handle -fdiagnostics-color in lto

[Jakub Jelinek]

On Mon, Feb 08, 2016 at 11:34:39AM +0100, Tom de Vries wrote:
> Hi,
> 
> when running libgomp.oacc-c-c++-common/parallel-dims.c with -flto
> -fno-use-linker-plugin, we run into a failing 'test for excess errors'.
> 
> The problem is that while -fdiagnostics-color=never is passed to gcc, it's
> not propagated to lto1, and the error message is annotated with color
> information, which confuses the test for excess errors.
> 
> This patch fixes the problem by making sure that -fdiagnostics-color is
> propagated to lto1, in the same way that -fdiagnostics-show-caret is
> propagated to lto1.
> 
> Bootstrapped and reg-tested on x86_64.
> 
> OK for trunk, stage1?

Doesn't that mean diagnostics from the driver itself will no longer honor
that option when deciding if the diagnostics should be colorized or not?

	Jakub