Request for review (S) 7017193: Small memory leak in get_stack_bounds // os::create_stack_guard_pages

Poonam Bajaj poonam.bajaj at oracle.com
Mon Feb 21 23:15:37 PST 2011


Hi Dmitry,

Few comments:

- These changes read the file character by character instead of one line 
at a time. Did you check if there is any performance impact due to that ?

- I think the comments should be outside the corresponding if blocks. 
And indentation of comment at line 2675 needs to be corrected.

2673     if ( bup == buf+sizeof(buf)-1 ){
2674        /* Address part of /proc/maps couldn't be more than 128 bytes,
2675             * flush the buffer
2676         */
2677       assert(checkme != true, "Should never happen");
2678       bup = buf;
2679       continue;
2680     }
2681 
2682     if (*bup == '\n'){
2683       /* We already seen '[' so it's safe to step back. 
2684        * if we don't have ack] keyword in this line just flush the buffer
2685        * buffer. 
2686        */
2687        if (checkme == false ||  *((int*)(bup-4)) != *((int*) "ack]") ){
2688          bup = buf;
2689          checkme = false;
2690          continue;
2691        }


- bup variable may be renamed to 'buf_pointer' or something like that. 



Thanks,
Poonam


On 2/21/2011 5:50 PM, Dmitry Samersoff wrote:
> Hi Everyone,
>
> http://cr.openjdk.java.net/~dsamersoff/7017193
>
> I'm uncomfortable with one malloc/free cycle per line so decided to 
> get rid of getline() call completely.
>
> -Dmitry
>
>


More information about the hotspot-runtime-dev mailing list