Since upgrading to glibc-2.13-r4, valgrind's output is noisier (as is tradition with newer glibc's). Upgrading to 3.6.1-r2 with the glibc 2.14 patch quiets it back down. Example of noise on 'valgrind ls' (repeats twice here) ==15250== Conditional jump or move depends on uninitialised value(s) ==15250== at 0x40161D6: index (in /lib64/ld-2.13.so) ==15250== by 0x400720A: expand_dynamic_string_token (in /lib64/ld-2.13.so) ==15250== by 0x4007A9F: _dl_map_object (in /lib64/ld-2.13.so) ==15250== by 0x400135D: map_doit (in /lib64/ld-2.13.so) ==15250== by 0x400DA25: _dl_catch_error (in /lib64/ld-2.13.so) ==15250== by 0x4001276: do_preload (in /lib64/ld-2.13.so) ==15250== by 0x40037D0: dl_main (in /lib64/ld-2.13.so) ==15250== by 0x4013DA7: _dl_sysdep_start (in /lib64/ld-2.13.so) ==15250== by 0x4004886: _dl_start (in /lib64/ld-2.13.so) ==15250== by 0x4000BA7: ??? (in /lib64/ld-2.13.so) Reproducible: Always
I haven't forgotten about this, I'm busy cleaning up some issues.
Appreciate it. If it makes more sense to wait for 3.7.0, by all means don't divide your (I'm sure) limited time :)
Created attachment 303669 [details] build.log as requested
Oops, this was not meant for this bug, please remove/disregard (I can't seem to remove it myself)
The only problem is with ppc32 and ppc64. Let's begin stabilization on amd64 and x86. I'm not sure about bug 387659. valgrind is a debugging program and removing the -g flag may not be a good idea. This will need more testing and right now we need to proceed forward with -r3 because of glibc-2.14.
amd64 stable
x86 stable
We're in the middle of stabilizing a newer version, bug 430822, so this is obsolete.