<html> <head> <base href="https://bugs.freedesktop.org/"> </head> <body> <div> <a class="bz_bug_link bz_status_ASSIGNED " title="ASSIGNED - memory leak with intel i965 mesa when running android container in Ubuntu" href="https://bugs.freedesktop.org/show_bug.cgi?id=104884#c19">Comment # 19</a> on <a class="bz_bug_link bz_status_ASSIGNED " title="ASSIGNED - memory leak with intel i965 mesa when running android container in Ubuntu" href="https://bugs.freedesktop.org/show_bug.cgi?id=104884">bug 104884</a> from <a class="email" href="mailto:94389147@qq.com" title="Jiancong <94389147@qq.com>"> Jiancong</a> <pre>(In reply to Eero Tamminen from <a href="show_bug.cgi?id=104884#c17">comment #17</a>) > (In reply to Kenneth Graunke from <a href="show_bug.cgi?id=104884#c15">comment #15</a>) > > (In reply to Jiancong from <a href="show_bug.cgi?id=104884#c14">comment #14</a>) > > > Testing with Simon's patch,removing ralloc_strdup, it still be found > > > following possible leaking points. > ... > > That happens when you forget to glXDestroyContext and glXMakeCurrent to NULL > > to unbind it so it can actually get freed. It's not a driver bug. > > > (In reply to Jiancong from <a href="show_bug.cgi?id=104884#c16">comment #16</a>) > > Created <a href="attachment.cgi?id=137522" name="attach_137522" title="Patched with previous fix, and rerun the valgrind memcheck.">attachment 137522</a> <a href="attachment.cgi?id=137522&action=edit" title="Patched with previous fix, and rerun the valgrind memcheck.">[details]</a> > > Patched with previous fix, and rerun the valgrind memcheck. > > Please fix the issue pointed out by Kenneth first, and some of the other > application bugs visible in the valgrind output first. Otherwise it's hard > to say which of those numerous items in your log are relevant. > > Alternative, you could kill the valgrind process with a signal that the > application doesn't catch, but Valgrind can catch (e.g. SIGHUP or SIGXCPU > could be such). That way application cannot mess up the leakage info with > bugs it has in its exit path (which are irrelevant for tracking down > run-time leakage). > > Note that for tracking down larger leaks, Valgrind Massif output is often > more useful than Valgrind Memcheck output. Thanks for your advice. I've tested it again with massif and killed valgrind process with SIGHUP. After 4 hours testing, And I attached the massif.out file in attachment named . From the file, the #1 memory occupation is ralloc_size. As android app open and exit, the ralloc_size increase dramtically. The #2 memory occupation is eglCreateContext. But the weird is the callstack shows mesa called _swrast_CreateContext finally. In the configuration process, I've disabled swrast compile with "./configure --enable-glx=dri --enable-dri --with-dri-drivers=i915,i965 --with-gallium-drivers=i915 --disable-gallium-llvm". Is there anything wrong in my configuration?</pre> </div> <hr> You are receiving this mail because: <ul> <li>You are the QA Contact for the bug.</li> </ul> </body> </html>