linux-kernel - Re: [PATCH] libnvdimm: Clarify nd_pfn

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: <CAPcyv4iVg09ryVAqOnxXAGUJf7xZqzxL0CWxGY4WOTEman--8w@mail.gmail.com>
Date:   Mon, 21 Jan 2019 17:00:47 -0800
From:   Dan Williams <dan.j.williams@...el.com>
To:     Wei Yang <richardw.yang@...ux.intel.com>
Cc:     linux-nvdimm <linux-nvdimm@...ts.01.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] libnvdimm: Clarify nd_pfn_init() flow

On Mon, Jan 21, 2019 at 4:47 PM Wei Yang <richardw.yang@...ux.intel.com> wrote:
>
> On Mon, Jan 21, 2019 at 04:29:08PM -0800, Dan Williams wrote:
> >On Mon, Jan 21, 2019 at 4:26 PM Wei Yang <richardw.yang@...ux.intel.com> wrote:
> >[..]
> >> >@@ -706,6 +711,22 @@ static int nd_pfn_init(struct nd_pfn *nd_pfn)
> >> >               sig = DAX_SIG;
> >> >       else
> >> >               sig = PFN_SIG;
> >> >+
> >> >+      /*
> >> >+       * Check for an existing 'pfn' superblock before writing a new
> >> >+       * one. The intended flow is that on the first probe of an
> >> >+       * nd_{pfn,dax} device the superblock is calculated and written
> >> >+       * to the namespace. In this case nd_pfn_validate() returns
> >> >+       * -ENODEV because no valid superblock exists currently.
> >>
> >> As you replied in following mail:
> >>
> >> 3/ If present, nd_pfn_validate() returns 0 and nd_dax_probe()
> >> registers the dax0.1 device (this is a libnvdimm 'personality device).
> >>
> >> So at this point, nd_pfn_validate() return 0 or -ENODEV?
> >
> >In this case 0, because the configuration was successfully validated.
> >
> >-ENODEV, is only returned for the initial case where we want the
> >kernel to write the configuration.
> >
> >All other error codes are an actual failure and the probe procedure stops.
>
> To be honest, this maybe crystal clear for you. But I still feel a little
> confused. Especially on differentiating those cases. How many cases we have?

There are 4 cases:

1/ namespace probed, no personality info-block found, namespace
results in 'raw' mode

2/ namespace probed, {pfn,btt,dax} info-block found, {pfn,btt,dax}
device created and attached to the corresponding personality driver

3/ namespace probing ends in failure like -ENOMEM, or a -EIO error
reading namespace capacity, for example due to a bad block.

4/ namespace force assigned to a {pfn,btt,dax} device instance,
{pfn,btt,dax} device force enabled  leading to the info-block being
written out to the device

> And what's your first probe mean? This the nd_btt/pfn/dax_probe()? or the
> linux driver probe?

The first probe is always nd_pmem_probe(). nd_{pfn,btt,dax}_probe()
are secondary.