The attached test-case shows memory leaking from _cairo_gradient_pattern_init_copy. Here's the valgrind output: ==8450== 88 bytes in 1 blocks are definitely lost in loss record 3 of 5 ==8450== at 0x1B902338: malloc (vg_replace_malloc.c:130) ==8450== by 0x1B9239FE: _cairo_gradient_pattern_init_copy (cairo-pattern.c:166) ==8450== by 0x1B923B20: _cairo_pattern_init_copy (cairo-pattern.c:205) ==8450== by 0x1B913F00: _cairo_gstate_copy_transformed_pattern (cairo-gstate.c:699) ==8450== by 0x1B913F86: _cairo_gstate_copy_transformed_source (cairo-gstate.c:713) ==8450== by 0x1B915772: _cairo_gstate_fill (cairo-gstate.c:1569) ==8450== by 0x1B90F960: cairo_fill_preserve (cairo.c:1598) ==8450== by 0x1B90F926: cairo_fill (cairo.c:1576) ==8450== by 0x80487B0: main (in /home/remenic/cairo-pattern-leak)
Created attachment 3850 [details] test-case
Thanks. This is now fixed in CVS: 2005-11-21 Carl Worth <cworth@cworth.org> * ROADMAP: Add reminder to fix all memory leaks (as determined by the test suite) before release 1.2.0. * src/cairo-gstate.c: (_cairo_gstate_stroke), (_cairo_gstate_fill): Patch leak of patterns as noted in bug #5100 (thanks to Richard Stellingwerff <remenic@gmail.com>). Note. You might also find that 'make check-valgrind' in cairo's source tree is useful for examining memory leaks. Back with 1.0, this check was clean, but it's obviously not anymore. I've added a note to the cairo 1.2 ROADMAP that these leaks should all be fixed before that release.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.