On Tue, Apr 26, 2011 at 4:27 PM, Daniel Baluta <daniel.baluta@xxxxxxxxx> wrote:
> On Tue, Apr 26, 2011 at 1:54 PM, ratheesh kannoth
> <ratheesh.ksz@xxxxxxxxx> wrote:
>> On Tue, Apr 26, 2011 at 4:10 PM, Daniel Baluta <daniel.baluta@xxxxxxxxx> wrote:
>>> On Tue, Apr 26, 2011 at 1:33 PM, Randi Botse <nightdecoder@xxxxxxxxx> wrote:
>>>> Hi All,
>>>>
>>>> I want to ask malloc() behaviour, consider these codes;
>>>>
>>>> ...
>>>> char *ptr = malloc(1);
>>>> strcpy(ptr, "what");
>>>> puts(ptr);
>>>> ....
>>>>
>>>> Confusingly, the strcpy() copied all bytes to ptr, but I just manage
>>>> to allocate ptr only for 1 byte, I guess I will have segfault here,
>>>> why this happen? why the string successfully copied into ptr? , is
>>>> those code legal?
>>>
>>> You didn't get segfault because you were lucky.
>>>
>>> Memory is allocated in multiples of page size (usually 4K).
>>> The memory after your allocated byte is valid in your case.
>>>
>>> thanks,
>>> Daniel.
>>> --
>>
>> U could read a little more about vm_page_struct. ( virtual address
>> space to physical page ).
>
> Can you elaborate on this?
>
> Daniel.
>
Daniel,
Note: Pls read Linux Kernel internals 2.6.
malloc() and free() works on virtual address space.
malloc(1) - this allotes a virtual address space of 4k.
strcpy(ptr, "what" ) - the 4k virtual address is mapped to a 4k
page frame thru page fault exception. Now you have a valid virtual
address of 4k.
free(ptr) - tells OS that this virtual address space (
4k ) can be reallotted if there is a need . But each program has a
virtual address space of 3GB ( 32 bit
,4GB minus 1GB (kernel) ). SO ptr
is a valid pointer unless
1) vitual address
space is realloted.
2) page frame is
realloted for some other page ( LRU algorithm )
so if ptr is valid , you could do following
*(ptr ++ ), *ptr ... etc .
-ratheesh
--
To unsubscribe from this list: send the line "unsubscribe linux-c-programming" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
