[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <dd08257d-6110-43ee-9949-2c879575cbef@rbox.co>
Date: Mon, 16 Dec 2024 13:03:36 +0100
From: Michal Luczaj <mhal@...x.co>
To: Stefano Garzarella <sgarzare@...hat.com>
Cc: netdev@...r.kernel.org
Subject: Re: [PATCH net-next 4/4] vsock/test: Add test for MSG_ZEROCOPY
completion memory leak
On 12/13/24 15:33, Stefano Garzarella wrote:
> On Thu, Dec 12, 2024 at 07:26:39PM +0100, Michal Luczaj wrote:
>> [...]
>> That said, I really think this test should be scrapped. I'm afraid it will
>> break sooner or later. And since kmemleak needs root anyway, perhaps it's
>> better to use failslab fault injection for this?
>
> As you prefer!
>
> I'd be for merging this new version, but I would ask you to put a
> comment above the function with your concerns about possible failures
> in the future and possibly how to implement it with more privileges.
> If they happen we always have time to remove the test or extend it to
> use more privileged things.
All right, here's v2: https://lore.kernel.org/netdev/20241216-test-vsock-leaks-v2-0-55e1405742fc@rbox.co/
>> [*] It's the only caller. Should @size be dropped from sock_omalloc()?
>
> Oh, I see, more a question for net maintainer, but I'd agree with you.
> So I think you can try sending a patch to net-next for that.
I digged more and it seems this (i.e. sock_omalloc() only being called
with @size=0) was the case from the very beginning[*]. I understand it was
deliberate and just awaits for some other users of ancillary sbks.
[*] https://lore.kernel.org/netdev/20170803202945.70750-1-willemdebruijn.kernel@gmail.com/
Powered by blists - more mailing lists