On Wed, Jul 29, 2020 at 05:34:13PM +0200, Oleg Nesterov wrote: > On 07/29, Luc Van Oostenryck wrote: > > > > The dissect client uses struct symbol's fields 'examined' & 'evaluated' > > to avoid reprocessing the same symbols. But these fields are used > > internally by sparse for type examination & evaluation and despite > > dissect not doing these operations explicitly, they can be done > > implicitly > > Yes. For example, test_dissect.c calls show_typename() and this can > lead to examine/evaluate. > > I didn't bother to fix this because test_dissect.c is the very basic > debugging tool. > > > So, add a new field to struct symbol: 'inspected' and use it, as > > well as the existing 'visited', instead of 'evaluated' & 'examined'. > > Thanks! Looks good to me. > > > Note: when used on the kernel, this patch avoids a lot of warnings: > > "warning: r_member bad sym type=7 kind=0" > > "warning: r_member bad mem->kind = 0" > > and creates substantially more normal output. > > So this is test-dissect, sindex should be fine even without this patch. Yes, I only used test-dissect but sindex should be affected too. The real problem is that, even without the show_typename() from test-dissect, some symbols can already be examined by __sparse(). In this case dissect.c:examine_sym_node() will be a no-op because node->examined will already be set. One of the symptoms is node->kind having a value of 0. I'll clarify the commit message. -- Luc
