linux-kernel - Re: [Xen-devel] [PATCH 2/2] xen-pciback: Allow enabling/disabling expansion ROM

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] [day] [month] [year] [list]

Date:   Mon, 03 Dec 2018 08:04:32 -0700
From:   "Jan Beulich" <JBeulich@...e.com>
To:     "Marek Marczykowski" <marmarek@...isiblethingslab.com>
Cc:     "Dwayne Litzenberger" <dlitz@...tz.net>,
        "Stefano Stabellini" <sstabellini@...nel.org>,
        "xen-devel" <xen-devel@...ts.xenproject.org>,
        "Boris Ostrovsky" <boris.ostrovsky@...cle.com>,
        "Juergen Gross" <jgross@...e.com>, <linux-kernel@...r.kernel.org>
Subject: Re: [Xen-devel] [PATCH 2/2] xen-pciback: Allow
 enabling/disabling expansion ROM

>>> On 03.12.18 at 15:47, <marmarek@...isiblethingslab.com> wrote:
> On Mon, Dec 03, 2018 at 04:01:07AM -0700, Jan Beulich wrote:
>> >>> On 02.12.18 at 18:47, <marmarek@...isiblethingslab.com> wrote:
>> > From: Dwayne Litzenberger <dlitz@...tz.net>
>> > 
>> > Newer AMD GPUs store their initialization routines as bytecode on the
>> > ROM.  This fixes the following initialization error inside the VM when
>> > doing PCI passthrough:
>> > 
>> >     radeon 0000:00:05.0: Invalid PCI ROM header signature: expecting 0xaa55, got 0xffff
>> >     radeon 0000:00:05.0: Invalid PCI ROM header signature: expecting 0xaa55, got 0xffff
>> >     [drm:radeon_get_bios [radeon]] *ERROR* Unable to locate a BIOS ROM
>> >     radeon 0000:00:05.0: Fatal error during GPU init
>> 
>> Isn't it that qemu is supposed to surface the ROM image to guests,
>> making it unnecessary to allow guests control over the physical
>> enable bit? 
> 
> Unless that qemu is in stubdomain, where it use pciback to access
> everything about the device...

Would be quite helpful to explain this in the description.

>> Also why would allowing to alter the bit depend on
>> whether the address portion of the value is non-zero?
> 
> That's a good question. According to commit message I think it should be
> a ROM presence check instead. If needed at this point at all - is this
> function even called if there is no ROM?

I suppose this was a question to Dwayne, not me.

Jan