[ug-bosug] Bluetooth Stack in userspace

[Ananth Shrinivas]

realfree() is the libc function that coalesces free blocks and maintains 
them in a splay tree. If your program crashes here,  it means you have 
written past or before a valid pointer, and modified metadata about 
allocated blocks that libc maintains.

Makes sense because a redundant malloc() alters the behavior of your 
program.

Run your program under libumem or watchmalloc - The point of failure 
will be closer to your root-cause due to the red-zone protection code.

Cheers,
Ananth

Anand Bheemarajaiah wrote:
> Hello all,
>
> I am currently building the bluetooth stack support in solaris for usb bluetooth dongles using the libusb api as a part of my undergraduate final year project.
>
> I have completed the hci-usb layer, hci layer, l2cap layer, sdp layer. I am currently doing the rfcomm layer.
>
> My query is with the code, it has started to behave in an unpredictable manner,
>
> for example,
>
> In the main function i am first inquiring for devices, then extracting their human readable names, then connecting to them and going on up the various layers of the bluetooth stack.
>
> here in the code, main function looks something like this,
>
> .....
> hci_inquiry(hd,req,resp);  // inquire for devices
> print_inq(resp);        // extracts the human readable names of devices discovered
> ....
>
>
> This code started crashing within the print_inq function after some time, now if i embed a malloc in between the two calls it works, 
>
> .....
> hci_inquiry(hd,req,resp);  // inquire for devices
> malloc(sizeof(struct hci_evt)); 
> // hci_evt is the structure holding info of incoming event pkts.
> print_inq(resp);        // extracts the human readable names of devices discovered
> ....
>
>
> can you tell why it is happening so.... and the runtime error is
>
> sigsegv with mapper code ... - it actually shows the callstack to be at realfree.
>
> i am sure it is not running out of memory because other commands and memory allocations work just fine.
>
> Any help would be great because i believe this is a stupid error and don't want to waste too much time on this.
>
> regards,
> Anand B
>  
>  
> This message posted from opensolaris.org
> _______________________________________________
> ug-bosug mailing list
> List-Unsubscribe: mailto:ug-bosug-unsubscribe at opensolaris.org
> List-Owner: mailto:ug-bosug-owner at opensolaris.org
> List-Archives: http://www.opensolaris.org/jive/forum.jspa?forumID=54
>