Re: [Valgrind-developers] asymmetric --malloc-fill/--free-fill behavior with mempools

[Julian S. <js...@ac...>]

> Is this behavior expected?  I'll can create a bugzilla for it either way,
> since a clarifying doc fix makes sense to me if it is expected behavior.

Yes, please file a bug report.  If you have a tentative fix pls add that
as a patch.

> It's clear how this is happening in the Valgrind code; "MC_(new_block)" and
> "die_and_free_mem" have asymmetric behavior when "MC_AllocCustom" is
> passed as "kind" to "MC_(new_block)".  It looks easy enough to fix in
> either function, depending on what behavior is desired.

I'm unclear why you speak of the possibility of two different behaviours.
Shouldn't the clreqs also simply respect the value of --malloc-fill?  If
so, why are there two possible fixes?

J