<html> <head> <base href="https://bugs.freedesktop.org/"> </head> <body> <div> <a class="bz_bug_link bz_status_NEW " title="NEW - Thread leak when changing context size" href="https://bugs.freedesktop.org/show_bug.cgi?id=107092#c3">Comment # 3</a> on <a class="bz_bug_link bz_status_NEW " title="NEW - Thread leak when changing context size" href="https://bugs.freedesktop.org/show_bug.cgi?id=107092">bug 107092</a> from <a class="email" href="mailto:sroland@vmware.com" title="Roland Scheidegger <sroland@vmware.com>"> Roland Scheidegger</a> <pre>I'm near certain this is a bug in the xlib state tracker rather than llvmpipe. swr will leak as well, you just don't notice because it doesn't create additional threads. glXChooseFBConfig will call choose_visual(), which will call xmesa_init(), which will create the pipe screen (via xmesa_init_display()). (llvmpipe threads are per screen, not per context.) (mesa/src/gallium/state_trackers/glx/xlib/ xm_api.c and glx_api.c) Note that there's 2 places where the screens would get destroyed (one in xmesa_close_display(), reached via XCloseDisplay(), the other is in XMesaDestroyContext(), reached via glXDestroyContext()). Both are commented out and marked as FIXME, since apparently it's not safe to destroy the screen at this point since surfaces can still be around... This looks apparently broken but it's not obvious to me how to fix it. At first glance it looks like screen destruction should be delayed until there's neither contexts nor surfaces referencing it are around but I guess it isn't that easy...</pre> </div> <hr> You are receiving this mail because: <ul> <li>You are the QA Contact for the bug.</li> <li>You are the assignee for the bug.</li> </ul> </body> </html>