[Ilugc] Structure copy behaviour
- From: vaidhy@xxxxxxxxxx (Vaidhy)
- Date: Thu May 24 12:22:26 2007
From what I remember, gcc always does a shallow copy (case 1) and as an
optimization, it implements copy-on-write.
Basically, assigning a[0] = t will have t and a[0] pointed to the same
location till you change either t or a[0]. At that point, there will be a
memcpy for sizeof(tmp) from &t to &a[0] and then the write will happen.
If you want to have a deep copy, only option I can think of right now is to
do it in C++ and implement the copy constructor and also overload the
assignment operator. (this is your case 2).
Regards,
Vaidhy
PS: My C and C++ are kind of rusty since I have not coded in them for the
last 4+ years. If things have changed since then, someone will be able to
correct it :)
On 5/24/07, Sujith Manoharan <m.sujith@xxxxxxxxx> wrote:
Hi,
Say I have a structure :
typedef struct {
...
...
...
} tmp;
And I do this :
tmp a[256];
tmp t;
a[0] = t;
Now, for this structure copy how does gcc decide that :
1) It should generate a call to memcpy
2) Generate instructions to manually do the copy.
What is the threshold for the structure's size that changes gcc's
behaviour ?
And is there any gcc parameter to change this threshold value ?
regards,
Sujith.
_______________________________________________
To unsubscribe, email ilugc-request@xxxxxxxxxxxxx with
"unsubscribe <password> <address>"
in the subject or body of the message.
http://www.ae.iitm.ac.in/mailman/listinfo/ilugc
Other related posts:
