linux-kernel - Re: [PATCH v2 1/2] drm: Add a drm_drv_enabled() to check if drivers should be enabled

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: <877ddmapfj.fsf@intel.com>
Date:   Fri, 05 Nov 2021 12:04:48 +0200
From:   Jani Nikula <jani.nikula@...ux.intel.com>
To:     Javier Martinez Canillas <javierm@...hat.com>,
        Thomas Zimmermann <tzimmermann@...e.de>,
        linux-kernel@...r.kernel.org
Cc:     David Airlie <airlied@...ux.ie>,
        Daniel Vetter <daniel.vetter@...ll.ch>,
        dri-devel@...ts.freedesktop.org,
        Gurchetan Singh <gurchetansingh@...omium.org>,
        Gerd Hoffmann <kraxel@...hat.com>,
        amd-gfx@...ts.freedesktop.org,
        VMware Graphics <linux-graphics-maintainer@...are.com>,
        Peter Robinson <pbrobinson@...il.com>,
        nouveau@...ts.freedesktop.org, Dave Airlie <airlied@...hat.com>,
        Ben Skeggs <bskeggs@...hat.com>,
        Michel Dänzer 
        <michel@...nzer.net>, Hans de Goede <hdegoede@...hat.com>,
        Rodrigo Vivi <rodrigo.vivi@...el.com>,
        virtualization@...ts.linux-foundation.org,
        Pekka Paalanen <pekka.paalanen@...labora.com>,
        "Pan\, Xinhui" <Xinhui.Pan@....com>,
        spice-devel@...ts.freedesktop.org,
        Alex Deucher <alexander.deucher@....com>,
        intel-gfx@...ts.freedesktop.org,
        Christian König <christian.koenig@....com>
Subject: Re: [PATCH v2 1/2] drm: Add a drm_drv_enabled() to check if drivers should be enabled

On Fri, 05 Nov 2021, Javier Martinez Canillas <javierm@...hat.com> wrote:
> Hello Thomas,
>
> On 11/5/21 09:43, Thomas Zimmermann wrote:
>> Hi
>> 
>> Am 04.11.21 um 21:09 schrieb Javier Martinez Canillas:
>>> Hello Jani,
>>>
>>> On 11/4/21 20:57, Jani Nikula wrote:
>>>> On Thu, 04 Nov 2021, Javier Martinez Canillas <javierm@...hat.com> wrote:
>>>>> +/**
>>>>> + * drm_drv_enabled - Checks if a DRM driver can be enabled
>>>>> + * @driver: DRM driver to check
>>>>> + *
>>>>> + * Checks whether a DRM driver can be enabled or not. This may be the case
>>>>> + * if the "nomodeset" kernel command line parameter is used.
>>>>> + *
>>>>> + * Return: 0 on success or a negative error code on failure.
>>>>> + */
>>>>> +int drm_drv_enabled(const struct drm_driver *driver)
>> 
>> Jani mentioned that i915 absolutely wants this to run from the 
>> module_init function. Best is to drop the parameter.
>>
>
> Ok. I now wonder though how much value would add this function since
> it will just be a wrapper around the nomodeset check.
>
> We talked about adding a new DRIVER_GENERIC feature flag and check for
> this, but as danvet mentioned that is not really needed. We just need
> to avoid testing for nomodeset in the simpledrm driver.
>
> Do you envision other condition that could be added later to disable a
> DRM driver ? Or do you think that just from a code readability point of
> view makes worth it ?

Taking a step back for perspective.

I think there's broad consensus in moving the parameter to drm, naming
the check function to drm_something_something(), and breaking the ties
to CONFIG_VGA_CONSOLE. I appreciate the work you're doing to that
effect.

I think everything beyond that is still a bit vague and/or
contentious. So how about making the first 2-3 patches just that?
Something we can all agree on, makes good progress, improves the kernel,
and gives us something to build on?

BR,
Jani.


>
>>>>> +{
>>>>> +	if (vgacon_text_force()) {
>>>>> +		DRM_INFO("%s driver is disabled\n", driver->name);
>>>>> +		return -ENODEV;
>>>>> +	}
>> 
>> If we run this from within a module_init function, we'd get plenty of 
>> these warnings if drivers are compiled into the kernel. Maybe simply 
>> remove the message. There's already a warning printed by the nomodeset 
>> handler.
>>
>
> Indeed. I'll just drop it.
>
>>>>> +
>>>>> +	return 0;
>>>>> +}
>>>>> +EXPORT_SYMBOL(drm_drv_enabled);
>>>>
>>>> The name implies a bool return, but it's not.
>>>>
>>>> 	if (drm_drv_enabled(...)) {
>>>> 		/* surprise, it's disabled! */
>>>> 	}
>>>>
>>>
>>> It used to return a bool in v2 but Thomas suggested an int instead to
>>> have consistency on the errno code that was returned by the callers.
>>>
>>> I should probably name that function differently to avoid confusion.
>> 
>> Yes, please.
>>
>
> drm_driver_check() maybe ?
>  
> Best regards,

-- 
Jani Nikula, Intel Open Source Graphics Center