[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [PATCH] target-i386: Fix compilation with --enable-debu

From: Blue Swirl
Subject: Re: [Qemu-devel] [PATCH] target-i386: Fix compilation with --enable-debug
Date: Tue, 3 Jul 2012 18:43:53 +0000

On Sat, Jun 30, 2012 at 1:46 PM, Stefan Weil <address@hidden> wrote:
> Am 30.06.2012 14:16, schrieb Dunrong Huang:
>> 2012/6/30 Stefan Weil <address@hidden>:
>>> commit c4baa0503d9623f1ce891f525ccd140c598bc29a improved SSE table type
>>> safety which now raises compiler errors when latest QEMU was configured with
>>> --enable-debug.
>> The error occurs when building with --enable-werror, which is default
>> option, not --enable-debug.
> The default configuration (configure called without any options)
> compiles perfectly (no warnings, no errors) on x86_64 or i386
> hosts.
> I only see warnings or errors when QEMU was configured with
> --enable-debug-tcg. Option --enable-debug includes --enable-debug-tcg,
> so it triggers the same warnings or errors.
> Stefan H., I suggest running some of the buildbots (for example
> default_x86_64_debian_6_0 and default_i386_debian_6_0) with --enable-debug.
> That option provides better code checks for TCG related code,
> and it also results in much shorter build times because compiler
> optimizations are disabled. gcc won't detect some kinds of errors
> which it only detects when it optimizes the code - that's why
> using --enable-debug for all buildbots would not be reasonable.
> Blue, are you planning to apply my patch as it is, or would you
> prefer an implementation using if / else blocks instead of
> arrays of function pointers?

I looked briefly at how the tables could be converted switch() cases,
but it looks like there would be a need for some refactoring if we
don't want to just duplicate the common code.

Your patch could still be used as the first step regardless.

> Regards,
> Stefan W.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]