Red Hat Bugzilla – Bug 418621
Valgrind reports invalid reads in python
Last modified: 2008-08-02 19:40:36 EDT
Description of problem: I get this segfault when running rhythmbox: Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 0xb7fc5720 (LWP 3129)] 0x02514c61 in malloc_consolidate (av=0x2603120) at malloc.c:4834 4834 unlink(p, bck, fwd); (gdb) bt #0 0x02514c61 in malloc_consolidate (av=0x2603120) at malloc.c:4834 #1 0x02516d8d in _int_malloc (av=0x2603120, bytes=1048) at malloc.c:4166 #2 0x02518b7b in __libc_malloc (bytes=1048) at malloc.c:3553 #3 0x00feea56 in IA__g_malloc (n_bytes=1048) at gmem.c:131 #4 0x0100366b in IA__g_slice_alloc (mem_size=1048) at gslice.c:824 #5 0x00bdf519 in pango_script_iter_new (text=0x92ac2d8 "Adam Richman (1)", length=16) at pango-script.c:130 #6 0x00bcdb8f in itemize_state_init (state=0xbfe84be0, context=<value optimized out>, text=<value optimized out>, base_dir=PANGO_DIRECTION_LTR, start_index=0, length=16, attrs=0xb5db2900, cached_iter=0xb5ddf4c0, desc=0x0) at pango-context.c:852 #7 0x00bcea2b in pango_itemize_with_base_dir (context=0x903d950, base_dir=PANGO_DIRECTION_LTR, text=0x92ac2d8 "Adam Richman (1)", start_index=0, length=16, attrs=0xb5db2900, cached_iter=0xb5ddf4c0) at pango-context.c:1423 #8 0x00bd7c04 in pango_layout_check_lines (layout=0x92d3db0) at pango-layout.c:3618 #9 0x00bd8da7 in pango_layout_get_extents_internal (layout=0x92d3db0, ink_rect=0x0, logical_rect=0xbfe84f3c, line_extents=0x0) at pango-layout.c:2311 #10 0x00722cce in get_size (cell=0x8ca6398, widget=0x8c38d08, cell_area=0x0, layout=0x92d3db0, x_offset=0x0, y_offset=0x0, width=0xbfe8503c, height=0xbfe85040) at gtkcellrenderertext.c:1540 ---Type <return> to continue, or q <return> to quit--- #11 0x00722fb5 in gtk_cell_renderer_text_get_size (cell=0x8ca6398, widget=0x8c38d08, cell_area=0x0, x_offset=0x0, y_offset=0x0, width=0xbfe8503c, height=0xbfe85040) at gtkcellrenderertext.c:1606 #12 0x00719d93 in IA__gtk_cell_renderer_get_size (cell=0x8ca6398, widget=0x8c38d08, cell_area=0x0, x_offset=0x0, y_offset=0x0, width=0xbfe8503c, height=0xbfe85040) at gtkcellrenderer.c:508 #13 0x0099d85e in IA__gtk_tree_view_column_cell_get_size ( tree_column=0x8caceb0, cell_area=0x0, x_offset=0x0, y_offset=0x0, width=0xbfe850f4, height=0xbfe850f0) at gtktreeviewcolumn.c:2606 #14 0x00980817 in validate_row (tree_view=0x8c38d08, tree=0x8ccaae8, node=0x92bc480, iter=0xbfe85280, path=0x92f57d8) at gtktreeview.c:5609 #15 0x009817bb in validate_visible_area (tree_view=0x8c38d08) at gtktreeview.c:5924 #16 0x00982761 in do_presize_handler (tree_view=0x8c38d08) at gtktreeview.c:6237 #17 0x009828ce in presize_handler_callback (data=0x8c38d08) at gtktreeview.c:6259 #18 0x00ae88ff in gdk_threads_dispatch (data=0x92c9270) at gdk.c:470 #19 0x00fe5541 in g_idle_dispatch (source=0x903d068, callback=0, user_data=0x92c9270) at gmain.c:4132 #20 0x00fe710c in IA__g_main_context_dispatch (context=0x8a975f8) at gmain.c:2061 #21 0x00fea54f in g_main_context_iterate (context=0x8a975f8, block=1, ---Type <return> to continue, or q <return> to quit--- dispatch=1, self=0x89cc410) at gmain.c:2694 #22 0x00fea8f9 in IA__g_main_loop_run (loop=0x923cb38) at gmain.c:2898 #23 0x008283ca in IA__gtk_main () at gtkmain.c:1163 #24 0x0805f7ed in main (argc=Cannot access memory at address 0x1 ) at main.c:340 Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
Created attachment 283071 [details] valgrind log Here's a valgrind log. Somehow it didn't crash when I ran it in valgrind so maybe there's a race here somewhere?
*** Bug 418931 has been marked as a duplicate of this bug. ***
I upgraded my whole machine to rawhide, and I can't reproduce the problem here...
You mean you don't see invalid reads in valgrind logs or that it doesn't crash?
Doesn't crash for me. I haven't tried running it under valgrind, as most of the invalid reads are actually from python, not from us. Tried disabling all the plugins? Any other errors on the command-line?
When I was having this problem I had only applied half the available updates from the repo since there were a few broken deps - Now that I've updated all packages it seems to be fine. Maybe there was a missing Requires on a pacakge somewhere along the way.
I don't see any crashes either, but I do still see the invalid reads when running it in valgrind. They all come from python related code it seems
Cool. I'll reassign this to Python, as they should fix the valgrind errors there. The valgrind log is in comment #1.
Changing the summary to reflect reality. Adding a log from running python and a small script under valgrind.
Created attachment 290944 [details] valgrind log python run with debuginfo installed
See the Misc/README.valgrind and Misc/valgrind-python.supp file in the python tarball, we probably should package these as part of python-devel ... I'll do that for Fed-9. And I've put them on my people page now, so it's easier for you: http://people.redhat.com/jantill/fedora/ Moving this bug back though.
Kjartan, any errors with that python helper for valgrind?
I'll close this here and file a bug upstream if I find any problems.