linux-kernel - Re: [PATCH] driver core: support bus manage deferred probe

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Wed, 18 Jul 2018 17:16:47 +0000
From:   Dmitry Torokhov <dmitry.torokhov@...il.com>
To:     ning.a.zhang@...el.com
Cc:     gregkh@...uxfoundation.org, rafael.j.wysocki@...el.com,
        linux-kernel@...r.kernel.org, mark.gross@...el.com
Subject: Re: [PATCH] driver core: support bus manage deferred probe

Hi,

On Fri, Jun 29, 2018 at 03:02:26PM +0800, ning.a.zhang@...el.com wrote:
> From: Zhang Ning <ning.a.zhang@...el.com>
> 
> deferred probe will be hanlded by deferred_probe_initcall,
> starts at late_initcall.
> 
> this is too late to handle deferred probe, this will block kernel exec
> to userspace, make kernel initial time longer.
> 
> if we know when required resources are ready for a list of devices,
> related deferred drivers can be probe earlier.
> 
> add these kinds of drivers into logical list,
> eg, bus_type->deferred_probe_pending_list,
> then it can be easily handled by bus driver.

I do not see why bus driver would have a better knowledge about its
device dependencies to accelerate deferred probe retries; it all depends
on the platform. On some platform PCI is essential and devices on PCI
bus are needed early, on others we can push PCI initialization to the
very end as there only a couple devices hanging off PCI bus.

You need to find a better way to decide when to reprobe a device, and
that requires analyzing and tracking the dependencies between devices.

Thanks.

-- 
Dmitry