<div>Hello there, <br></div><div> </div><div>I reviewed intel driver in Mesa7.2 and found there seems a problem in <br></div><div>======================================================<br>intel_region_release(struct intel_region **region)<br>{<br> if (!*region)<br> return;<br><br> DBG("%s %d\n", __FUNCTION__, (*region)->refcount - 1);<br><br> ASSERT((*region)->refcount > 0);<br> (*region)->refcount--;<br><br> if ((*region)->refcount == 0) {<br> assert((*region)->map_refcount == 0);<br><br> if ((*region)->pbo)<br> (*region)->pbo->region = NULL;<br> (*region)->pbo = NULL;<br> dri_bo_unreference((*region)->buffer);<br> free(*region);<br> }<br> *region = NULL;<br>}<br>==========================================<br>If
there are two contexts made current on this region of rb, after
"(*region)->refcount--;" is it possible refcount still be larger
than "0", and it will not be free then leaked?<br><br>thanks,<br></div><br><br><span title="neteasefooter"><hr>
<a href="http://fanxian.163.com/fanxian/?keyfrom=163mail.footer">网易邮箱用户购物独享现金返还</a>
</span><br><br><span title="neteasefooter"/><hr/>
<a href="http://fanxian.163.com/fanxian/?keyfrom=163mail.footer">网易邮箱用户购物独享现金返还</a>
</span>