linux-kernel - Re: [PATCH] x86/resctrl: Fix rdt_find

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:   Mon, 10 Dec 2018 12:42:14 -0800
From:   Reinette Chatre <reinette.chatre@...el.com>
To:     Borislav Petkov <bp@...en8.de>
Cc:     tglx@...utronix.de, fenghua.yu@...el.com, tony.luck@...el.com,
        jithu.joseph@...el.com, mingo@...hat.com, hpa@...or.com,
        x86@...nel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] x86/resctrl: Fix rdt_find_domain() return value checks

Hi Boris,

On 12/10/2018 11:13 AM, Borislav Petkov wrote:
> On Wed, Nov 28, 2018 at 10:20:27AM -0800, Reinette Chatre wrote:
>> rdt_find_domain() may return an ERR_PTR(), NULL, or a pointer to struct
>> rdt_domain. It is thus required that the return value be checked for the
>> possibility of an ERR_PTR as well as NULL.
> 
> Well, it returns ERR_PTR(id) but code which uses ERR_PTR passes in an -E
> value, for example ERR_PTR(-EINVAL) or so, and not an id.
> 
> And that might work now if id fits within that MAX_ERRNO range - I'm
> looking at include/linux/err.h - but that's still fragile.
> 

Thank you for catching this.

It does seem as though things work at this time since rdt_find_domain()
contains:

if (id < 0)
        return ERR_PTR(id);

and from what I can tell the only possible negative value of id is -1.

As you note, this is fragile. Additionally the error, if intending to
use -E values, does not reflect the error (since -1 would mean EPERM).

Would you be ok if the above is changed to

if (id < 0)
        return ERR_PTR(-ENOENT);

as part of this patch?

Looking at rdtgroup_mondata_show() is does seem as though ENOENT is the
actual intended error value, although ENODEV could perhaps also be
considered since such a result reflects that a particular cache instance
could not be found.

Thank you!

Reinette