linux strlen man page accepts NULL?

linux strlen man page accepts NULL?

[Jon Grant]

Looking at this man page:

http://linux.die.net/man/3/strlen

Should it not mention that a NULL address is a valid param? Or is it
not a valid param? On most systems NULL is a special error pointer.

Cheers, Jon
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: linux strlen man page accepts NULL?

[Michael Kerrisk]

Jon,

On Fri, Aug 7, 2009 at 1:55 PM, Jon Grant<jg-hus3n9K41k0@public.gmane.org> wrote:
> Looking at this man page:
>
> http://linux.die.net/man/3/strlen
>
> Should it not mention that a NULL address is a valid param? Or is it
> not a valid param?

What makes you think it is a valid parameter?

> On most systems NULL is a special error pointer.

I don't understand what you mean with this last sentence. Please explain.

Thanks,

Michael


-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Watch my Linux system programming book progress to publication!
http://blog.man7.org/
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: linux strlen man page accepts NULL?

[Jon Grant]

Hello

2009/8/7 Michael Kerrisk <mtk.manpages-gM/Ye1E23mwN+BqQ9rBEUg@public.gmane.org>:
> Jon,
>
> On Fri, Aug 7, 2009 at 1:55 PM, Jon Grant<jg-hus3n9K41k0@public.gmane.org> wrote:
>> Looking at this man page:
>>
>> http://linux.die.net/man/3/strlen
>>
>> Should it not mention that a NULL address is a valid param? Or is it
>> not a valid param?
>
> What makes you think it is a valid parameter?

NULL points to 0x0, which could be mapped to something. On my embedded
platform it is the beginning of the boot ROM. However typically 0x0 is
an invalid address, in which case strlen should check for NULL, and
return 0

e.g.:

size_t strlen(const char *str)
{
        const char *s;

if(str == NULL)
{
    return 0;
}

        for (s = str; *s; ++s)
                ;
        return (s - str);
}

>> On most systems NULL is a special error pointer.
>
> I don't understand what you mean with this last sentence. Please explain.

Well, the purpose of strlen is to count chars (excluding terminating
'\0') but on most systems address 0x0 [which is what NULL is, (void*)0
].

So if something is mapped by the hardware at 0x0 address then strlen
could be used to count the number of characters at that location.
However, on most systems his 0x0 address (NULL) indicates an invalid
address.

I am not on this mailing list, so please keep my email address in any replies.

Best regards, Jon
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: linux strlen man page accepts NULL?

[Michael Kerrisk]

Hello Jon,

On Mon, Aug 10, 2009 at 2:06 PM, Jon Grant <jg-hus3n9K41k0@public.gmane.org> wrote:
> Hello
>
> 2009/8/7 Michael Kerrisk <mtk.manpages-gM/Ye1E23mwN+BqQ9rBEUg@public.gmane.org>:
>> Jon,
>>
>> On Fri, Aug 7, 2009 at 1:55 PM, Jon Grant<jg-hus3n9K41k0@public.gmane.org> wrote:
>>> Looking at this man page:
>>>
>>> http://linux.die.net/man/3/strlen
>>>
>>> Should it not mention that a NULL address is a valid param? Or is it
>>> not a valid param?
>>
>> What makes you think it is a valid parameter?
>
> NULL points to 0x0, which could be mapped to something.

D'oh! Sorry -- now I'm with you. Yes, it could.

> On my embedded
> platform it is the beginning of the boot ROM. However typically 0x0 is
> an invalid address, in which case strlen should check for NULL, and
> return 0

I don't think it should check for this. If the addres is invalid, it
should be treated like any other invalid address -- usually a SIGSEGV
results.

> e.g.:
>
> size_t strlen(const char *str)
> {
>        const char *s;
>
> if(str == NULL)
> {
>    return 0;
> }
>
>        for (s = str; *s; ++s)
>                ;
>        return (s - str);
> }
>
>>> On most systems NULL is a special error pointer.
>>
>> I don't understand what you mean with this last sentence. Please explain.
>
> Well, the purpose of strlen is to count chars (excluding terminating
> '\0') but on most systems address 0x0 [which is what NULL is, (void*)0
> ].
>
> So if something is mapped by the hardware at 0x0 address then strlen
> could be used to count the number of characters at that location.
> However, on most systems his 0x0 address (NULL) indicates an invalid
> address.
>
> I am not on this mailing list, so please keep my email address in any replies.

As far as I can see, there's nothing in the documentation that needs
to be fixed. I'm doubtful that glibc's strlen() should be fixed
either. I agree that mappings at 0x0 are a strange corner case that
can produce unusual results in cases like these.

Cheers,

Michael

-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Watch my Linux system programming book progress to publication!
http://blog.man7.org/
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: linux strlen man page accepts NULL?

[Mike Frysinger]

[-- Attachment #1: Type: Text/Plain, Size: 717 bytes --]

On Saturday 19 September 2009 22:41:41 Michael Kerrisk wrote:
> On Mon, Aug 10, 2009 at 2:06 PM, Jon Grant wrote:
> > On my embedded
> > platform it is the beginning of the boot ROM. However typically 0x0 is
> > an invalid address, in which case strlen should check for NULL, and
> > return 0
> 
> I don't think it should check for this. If the addres is invalid, it
> should be treated like any other invalid address -- usually a SIGSEGV
> results.

right.  if 0 is a valid address, then strlen() should work.  if it isnt valid 
and you called strlen() on it anyways, then your code sucks and it should 
crash.  POSIX does not require special handling of any address, so neither 
should Linux.
-mike

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 836 bytes --]