qemu-devel
[Top][All Lists]
Advanced

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

Re: [Qemu-devel] [PATCH v4] avoid compilation warning/errors on up to da


From: Jean-Christophe Dubois
Subject: Re: [Qemu-devel] [PATCH v4] avoid compilation warning/errors on up to date
Date: Thu, 18 Jun 2009 01:16:37 +0200
User-agent: KMail/1.11.2 (Linux/2.6.28-13-generic; KDE/4.2.2; x86_64; ; )

Le jeudi 18 juin 2009 00:51:43 Anthony Liguori, vous avez écrit :
> Jean-Christophe Dubois wrote:
> > Once again, when all warnings are suppressed I doubt anybody will
> > volunteer for this boring task. And all future code added to the project
> > without system call error check will get unnoticed because you silenced
> > GCC.
> >
> > But at the end it is your call.
>
> Silencing the warnings != fixing the warnings.

Correct.

> To fix the warning, the errors have to be dealt with in a graceful way.
> This means either propagating the errors and releases any acquired
> resources or aborting.

Well, I believe my patch is trying to deal with errors in a graceful way and 
releasing acquired resources whenever possible/required (actually, it does fix 
a few fd leaks). If you find this is not the case just point me at the faulty 
file and I'll fix it.

> It is not an improvement to just add dummy error handling in a few
> spots.  It does not result in making the code any more robust.

I don't think I am doing "dummy" error handling. I have made all the error 
handling you requested and more (EINTR error, partial read/write) and tried to 
clean resources in an ordered way when I found it was required.

And whatever I have added is a lot more robust than what was before. Once 
again, please just show me where my patch is incorrect and I will fix it. 

But don't just point at lseek() and friends. There are many of these system 
calls that are not handled correctly in the qemu code and I don't think I'll 
fix them all at once. It seems you are thinking my patch is already too big.

>
> Regards,
>
> Anthony Liguori
>
> >> Regards,
> >>
> >> Anthony Liguori





reply via email to

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