[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20231019075432.4d975762@hermes.local>
Date: Thu, 19 Oct 2023 07:54:32 -0700
From: Stephen Hemminger <stephen@...workplumber.org>
To: Antoine Tenart <atenart@...nel.org>
Cc: davem@...emloft.net, kuba@...nel.org, pabeni@...hat.com,
edumazet@...gle.com, netdev@...r.kernel.org, gregkh@...uxfoundation.org,
mhocko@...e.com
Subject: Re: [RFC PATCH net-next 0/4] net-sysfs: remove
rtnl_trylock/restart_syscall use
On Thu, 19 Oct 2023 09:47:27 +0200
Antoine Tenart <atenart@...nel.org> wrote:
> > That doesn't mean the locking should not be fixed, just that better
> > to avoid the situation if possible.
>
> 100% agree on this one. I believe using netlink is the right way.
>
> Having said that, sysfs is still there and (quite some time ago) while
> having discussions with different projects, some were keen to switch to
> netlink, but some weren't and pushed back because "sysfs is a stable
> API" and "if there is a kernel issue it should be fixed in the kernel".
> Not blaming anyone really, they'd have to support the two interfaces for
> compatibility. My point is, yes, I would encourage everyone to use
> netlink too, but we don't control every user and it's not like sysfs
> will disappear anytime soon.
I have seen code doing discovery of new devices via netlink then poking
around in sysfs. But that usage is inherently racy from the application
point of view. By the time device is discovered, it might be removed or
worse renamed before the sysfs operations.
Powered by blists - more mailing lists