This is merely a historical archive of years 2008-2021, before the migration to mailman3.
A maintained and still updated list archive can be found at https://lists.osmocom.org/hyperkitty/list/OpenBSC@lists.osmocom.org/.
Holger Freyther holger at freyther.de
> On 04 Apr 2016, at 15:30, msuraev at sysmocom.de wrote:
>
Hi Max,
> +const char *bsc_nat_resolve_path(void *ctx, const char *base, const char *name)
> +{
> + return ('/' == name[0]) ? name : talloc_asprintf(ctx, "%s/%s", base,
> + name);
> +}
I had hinted to you to repeat config reload (and relative paths) and observing memory consumption by using top -p `pidof osmo-bsc_nat`.
This routine is problematic either way in C.
Semantic:
Sometimes new memory will be allocated sometimes not
Responsibility:
Sometimes the callers needs to call talloc_free, sometimes not.
Result:
a.) Caller will never call talloc_free => sometimes leaks
b.) Caller will always call talloc_free => sometimes crashes
Solution:
Make the semantic consistent
So please, take this leak and make yourself familiar with valgrind, talloc leak check, etc. It is important to learn from it. See my previous mail for an example how to script it.
kind regards
holger