suppress based on any callstack in indirect leak group

[derekbruening]

From [email protected] on June 08, 2011 06:19:09

Today only the leak whose callstack is reported will be matched vs
suppression and no indirect leaks will. Since the primary leak reported is
determined by malloc iteration order and can thus be relatively random it
can be difficult to suppress a leak when the primary has a fixed callstack
but there are many, many indirect leaks with very different callstacks.
Apparently this has been a problem: Timur can you fill in details if
possible?

Original issue: http://code.google.com/p/drmemory/issues/detail?id=444

[derekbruening]

From [email protected] on June 08, 2011 03:36:30

xref issue #283 (somewhat related)

Some details: in Chromium, some third-party components like NSS, GTK, Pango, FcConfig and others have many memory leaks. These leaks usually have just a few causes (e.g. a group of objects is intentionally leaked; or someone has forgotten to Unref the root object) but the allocation stacks are scattered all over the place, making the suppressions rather difficult.