netdev - Re: [PATCH v2 0/5] net: qrtr: Add distant node support

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]

Message-ID: <20201110094427.5ffb1d1b@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
Date:   Tue, 10 Nov 2020 09:44:27 -0800
From:   Jakub Kicinski <kuba@...nel.org>
To:     Loic Poulain <loic.poulain@...aro.org>,
        Manivannan Sadhasivam <manivannan.sadhasivam@...aro.org>
Cc:     David Miller <davem@...emloft.net>,
        Bjorn Andersson <bjorn.andersson@...aro.org>,
        cjhuang@...eaurora.org,
        Network Development <netdev@...r.kernel.org>
Subject: Re: [PATCH v2 0/5] net: qrtr: Add distant node support

On Tue, 10 Nov 2020 10:03:29 +0100 Loic Poulain wrote:
> On Mon, 9 Nov 2020 at 19:39, Jakub Kicinski <kuba@...nel.org> wrote:
> >
> > On Mon, 9 Nov 2020 09:49:24 +0100 Loic Poulain wrote:  
> > > > Looks like patch 1 is a bug fix and patches 2-5 add a new feature.
> > > > Is that correct?  
> > >
> > > That's correct, though strictly speaking 2-5 are also bug fix since remote node
> > > communication is supposed to be supported in QRTR to be compatible with
> > > other implementations (downstream or private implementations).  
> >
> > Is there a spec we can quote to support that, or is QRTR purely
> > a vendor interface?  
> 
> There is no public spec AFAIK, this is a vendor interface.
> 
> > What's the end user issue that we're solving? After firmware upgrade
> > things stop working? Things don't work on HW platforms on which this
> > was not tested? Don't work on new HW platforms?  
> 
> QRTR is usually something used in SoC context as communication
> protocol for accessing the differents IPs (modem, WiFi, DSP, etc)
> around the CPU. In that case, these components (nodes), identified
> with a 'node ID', are directly reachable by the CPU (QRTR over shared
> memory). This case is not impacted by the series, all nodes beeing CPU
> immediate neighbours.
> 
> But today QRTR is no more a ARCH_QCOM thing only, It is also exposed
> as communication channel for QCOM based wireless modules (e.g. SDX55
> modem), over PCIe/MHI. In that case, the host is only connected to the
> Modem CPU QRTR endpoint that in turn gives access to other embedded
> Modem endpoints, acting as a gateway/bridge for accessing
> non-immediate nodes from the host. currently, this case is not working
> and the series fix it.
> 
> However, AFAIK, the only device would request this support is the
> SDX55 PCIe module, that just landed in mhi-next. So I assume it's fine
> if the related part of the series targets net-next.

Thanks! Sounds like net-next will work just fine, but won't you need
these changes in mhi-next, then? In which case you should send a pull
request based on Linus' tree so that both Mani and netdev can pull it
in.

Mani, WDYT?