:: Re: [DNG] The D in Systemd stands f…
Page principale
Supprimer ce message
Répondre à ce message
Auteur: aitor
Date:  
À: dng
Sujet: Re: [DNG] The D in Systemd stands for 'Dammmmit!'
Hi,

On 10/29/2018 09:37 PM, Daniel Abrecht wrote:
> On 27/10/2018 19.38, Steve Litt wrote:
>> I say: You must use strncpy()/strncat() because strcpy()/strcat() are
>>     soooooo old.
>>
>> What's it been now, 30 years since the strn versions of those
>> commands have been around? You'd think they'd have taken that in and
>> adopted it by now. But nooooooooooooooooooooooooo!
>
> strcpy/strcat can be used safely if the length of the strings
> beforehand is checked properly. I've seen people changing perfectly
> fine code because some tool warned about functions being unsave, and
> then introducing bugs in the process I had to fix afterwards. There
> was an instance where some combination of strlen and strcpy was
> replaced by a strncpy, and of course they forgot that strncpy doesn't
> add the null byte if the buffer is full, and they missed that it fills
> the remaining buffer with null bytes otherwise too. Sure strcpy is
> dangerous, but strncpy isn't any better either. In fact, most things
> in c involving pointers and/or arrays are dangerous to do, it's easy
> to misunderstand semantics and intent of such code.
>
> I think strlcpy/strlcat are the functions which are the easiest to not
> use wrongly, but sadly they are non-standard, and even with those,
> people will still manage to make mistakes.


I have had a lot of memory problems with both strcpy and strcat working
on simple-netaid:

https://git.devuan.org/aitor_czr/simple-netaid-gtk/blob/master/backend_src/netstat.c

I reached into things like:

*(my_data.wired_device)='\0';

Cheers,

Aitor.