linux-kernel - Re: [PATCH v5 2/2] dax/kmem: allow kmem to add memory with memmap_on

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: <923d65270ad08d47adae0d82ab4b508d01e9cc00.camel@intel.com>
Date:   Tue, 17 Oct 2023 00:31:04 +0000
From:   "Verma, Vishal L" <vishal.l.verma@...el.com>
To:     "Williams, Dan J" <dan.j.williams@...el.com>,
        "Jiang, Dave" <dave.jiang@...el.com>,
        "osalvador@...e.de" <osalvador@...e.de>,
        "david@...hat.com" <david@...hat.com>,
        "akpm@...ux-foundation.org" <akpm@...ux-foundation.org>
CC:     "dave.hansen@...ux.intel.com" <dave.hansen@...ux.intel.com>,
        "Huang, Ying" <ying.huang@...el.com>,
        "linux-mm@...ck.org" <linux-mm@...ck.org>,
        "aneesh.kumar@...ux.ibm.com" <aneesh.kumar@...ux.ibm.com>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "linux-cxl@...r.kernel.org" <linux-cxl@...r.kernel.org>,
        "Hocko, Michal" <mhocko@...e.com>,
        "nvdimm@...ts.linux.dev" <nvdimm@...ts.linux.dev>,
        "jmoyer@...hat.com" <jmoyer@...hat.com>,
        "Jonathan.Cameron@...wei.com" <Jonathan.Cameron@...wei.com>
Subject: Re: [PATCH v5 2/2] dax/kmem: allow kmem to add memory with
 memmap_on_memory

On Thu, 2023-10-05 at 14:16 -0700, Dan Williams wrote:
> Vishal Verma wrote:
> > 
<..>

> > +
> > +       rc = kstrtobool(buf, &val);
> > +       if (rc)
> > +               return rc;
> 
> Perhaps:
> 
> if (dev_dax->memmap_on_memory == val)
>         return len;
> 
> ...and skip the check below when it is going to be a nop
> 
> > +
> > +       device_lock(dax_region->dev);
> > +       if (!dax_region->dev->driver) {
> 
> Is the polarity backwards here? I.e. if the device is already attached to
> the kmem driver it is too late to modify memmap_on_memory policy.

Hm this sounded logical until I tried it. After a reconfigure-device to
devdax (i.e. detach kmem), I get the -EBUSY if I invert this check.

> 
> > +               device_unlock(dax_region->dev);
> > +               return -ENXIO;
> 
> I would expect -EBUSY since disabling the device allows the property to be
> set and -ENXIO implies a more permanent error.
> 
> > +       }
> > +
> > +       dev_dax->memmap_on_memory = val;
> > +
> > +       device_unlock(dax_region->dev);
> > +       return len;
> > +}
> > +static DEVICE_ATTR_RW(memmap_on_memory);
> 
> This new attribute needs a new Documentation/ABI/ entry... in fact all of
> these attributes need Documentation/ entries. I can help with that base
> document to get things started.
> 
> Perhaps split this sysfs ABI into its own patch and, depending on how fast
> we can pull the Documentation together, start with the
> region-driver-conveyed approach in the meantime.

Yep I'll split this out and I can do a separate series to add the ABI
docs for /sys/bus/dax, and include this new ABI in that as well.

Agreed with all other comments.