linux-kernel - Re: [PATCH RESEND] PCI: endpoint: Fix clearing start entry in configfs

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

Date:   Tue, 25 Feb 2020 19:05:07 +0900
From:   Kunihiko Hayashi <hayashi.kunihiko@...ionext.com>
To:     Lorenzo Pieralisi <lorenzo.pieralisi@....com>,
        Kishon Vijay Abraham I <kishon@...com>
Cc:     <linux-pci@...r.kernel.org>, <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH RESEND] PCI: endpoint: Fix clearing start entry in configfs

Hi Lorenzo,

On Fri, 21 Feb 2020 14:15:53 +0000 <lorenzo.pieralisi@....com> wrote:

> On Fri, Dec 20, 2019 at 09:24:37PM +0900, Kunihiko Hayashi wrote:
> > The value of 'start' entry is no change whenever writing 0 to configfs.
> > So the endpoint that stopped once can't restart.
> > 
> > The following command lines are an example restarting endpoint and
> > reprogramming configurations after receiving bus-reset.
> > 
> >     echo 0 > controllers/66000000.pcie-ep/start
> >     rm controllers/66000000.pcie-ep/func1
> >     ln -s functions/pci_epf_test/func1 controllers/66000000.pcie-ep/
> >     echo 1 > controllers/66000000.pcie-ep/start
> > 
> > However, the first 'echo' can't set 0 to 'start', so the last 'echo' can't
> > restart endpoint.
> 
> I think your description is not correct - pci_epc_group->start is
> just used to check if an endpoint has been started or not (in
> pci_epc_epf_unlink() and that's a WARN_ON) but nonetheless this
> looks like a bug and ought to be fixed.

When pci_epc_group->start keeps 1 after starting this controller with
'echo 1 > start', we can never unlink the func1 from the controller
because of WARN_ON.

I think that unlink/re-link play initialization role of configfs
through 'unbind' and 'bind' functions. However, we can't re-initialize
configfs.

If this is the intended behavior, my patch will make no sense.

> I need Kishon's ACK to proceed.

I think so, too.

Thank you,

> 
> Thanks,
> Lorenzo
> 
> > Fixes: d74679911610 ("PCI: endpoint: Introduce configfs entry for configuring EP functions")
> > Cc: Kishon Vijay Abraham I <kishon@...com>
> > Signed-off-by: Kunihiko Hayashi <hayashi.kunihiko@...ionext.com>
> > ---
> >  drivers/pci/endpoint/pci-ep-cfs.c | 1 +
> >  1 file changed, 1 insertion(+)
> > 
> > diff --git a/drivers/pci/endpoint/pci-ep-cfs.c b/drivers/pci/endpoint/pci-ep-cfs.c
> > index d1288a0..4fead88 100644
> > --- a/drivers/pci/endpoint/pci-ep-cfs.c
> > +++ b/drivers/pci/endpoint/pci-ep-cfs.c
> > @@ -58,6 +58,7 @@ static ssize_t pci_epc_start_store(struct config_item *item, const char *page,
> >  
> >  	if (!start) {
> >  		pci_epc_stop(epc);
> > +		epc_group->start = 0;
> >  		return len;
> >  	}
> >  
> > -- 
> > 2.7.4
> > 

---
Best Regards,
Kunihiko Hayashi