Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 30 Jun 2022 09:14:06 -0700
From:   Davidlohr Bueso <dave@stgolabs.net>
To:     Jonathan Cameron <Jonathan.Cameron@Huawei.com>
Cc:     Ira Weiny <ira.weiny@intel.com>,
        Dan Williams <dan.j.williams@intel.com>,
        Bjorn Helgaas <bhelgaas@google.com>,
        Lukas Wunner <lukas@wunner.de>,
        Alison Schofield <alison.schofield@intel.com>,
        Vishal Verma <vishal.l.verma@intel.com>,
        Dave Jiang <dave.jiang@intel.com>,
        Ben Widawsky <bwidawsk@kernel.org>,
        linux-kernel@vger.kernel.org, linux-cxl@vger.kernel.org,
        linux-pci@vger.kernel.org
Subject: Re: [PATCH V12 4/9] cxl/pci: Create PCI DOE mailbox's for memory
 devices
Message-ID: <20220630161406.yglzoxn2va3bhts4@offworld>
References: <20220628041527.742333-1-ira.weiny@intel.com>
 <20220628041527.742333-5-ira.weiny@intel.com>
 <20220628153317.00002e36@Huawei.com>
 <Yr01iZhuG5YHThDg@iweiny-desk3>
 <20220630163240.00003596@Huawei.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Disposition: inline
In-Reply-To: <20220630163240.00003596@Huawei.com>
User-Agent: NeoMutt/20220429
Precedence: bulk

On Thu, 30 Jun 2022, Jonathan Cameron wrote:

>On Wed, 29 Jun 2022 22:32:57 -0700 Ira Weiny <ira.weiny@intel.com> wrote:

>> I _thought_ that we did not care if some mailboxes failed or not.
>
>I have a different view to Dan on this.  In my view if your hardware is
>not working in any way at all scream like mad don't carry on... Dan
>is keen to try to muddle onwards.

I am also of the idea of not carrying on upon any indication of failure.

>>
>> If CDAT is not supported on any of the mailboxes found then CDAT will not show
>> up on sysfs (as per Dan's last comment).  If it was supported on a mailbox but
>> no data is found the sysfs will show up but be 0 length.
>>
>> At this layer I thought we agreed to skip over these errors.  If a protocol is
>> needed at a higher layer and it is not found on any of the mailboxes the errors
>> should show up in that layer.  In this series CDAT is not 100% necessary as
>> devices can work without it.  So the errors were mostly paper'ed over in favor
>> of just printing error messages and muddle on.
>>
>> The xa_insert() deserves a pci_err() though.
>
>That's probably the minimum we should do.  The xa_insert() failing is something
>horrible going wrong in our software / host afterall.

Yes. And in addition, devm_cxl_pci_create_doe() should return any error status, and
cxl_pci_probe() can choose to omit any errors, but it's still better to have it.

Thanks,
Davidlohr