[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <YrWPg3xIHbm9bFxP@arm.com>
Date: Fri, 24 Jun 2022 11:18:43 +0100
From: Catalin Marinas <catalin.marinas@....com>
To: Yee Lee <yee.lee@...iatek.com>
Cc: Patrick Wang <patrick.wang.shcn@...il.com>,
akpm@...ux-foundation.org, linux-mm@...ck.org,
linux-kernel@...r.kernel.org, Marco Elver <elver@...gle.com>
Subject: Re: [PATCH v4 3/4] mm: kmemleak: add rbtree and store physical
address for objects allocated with PA
On Thu, Jun 23, 2022 at 07:25:15PM +0800, Yee Lee wrote:
> On Thu, 2022-06-23 at 16:45 +0800, Yee Lee wrote:
> > Now we have seperated rb_tree for phys and virts addresses. But why
> > can't we have kmemleak_free_phys()? It may apply the same format to
> > delete_object_full().
> >
> > Some users would request to remove the kmemleak object from the phys
> > tree but we don't have this one.
>
> Please check this, an issue happened at kfence with the latest kmemleak
> patches. kfence pool allocated memory from memblock but have no way to
> free it from the phys tree.
> https://lkml.org/lkml/2022/6/23/486
I don't think I was cc'ed on the other thread but at a quick look, what
you probably want is:
kmemleak_ignore_phys(__kfence_pool);
instead of the current kmemleak_free(). With Patrick's changes, you can
no longer tell kmemleak about an object with a physical address and free
it with the virtual one.
--
Catalin
Powered by blists - more mailing lists