Re: [PATCH] drm/dumb-buffers: Increase size limits to match current devices

[Thomas Zimmermann <tzimmermann@suse.de>]

Hi

Am 02.06.26 um 12:53 schrieb Jani Nikula:
> On Tue, 02 Jun 2026, "Saarinen, Jani" <jani.saarinen@intel.com> wrote:
>> Hi,
>>> -----Original Message-----
>>> From: Thomas Zimmermann <tzimmermann@suse.de>
>>> Sent: Tuesday, 2 June 2026 10.30
>>> To: rajat.gupta@oss.qualcomm.com; jani.nikula@linux.intel.com; Saarinen,
>>> Jani <jani.saarinen@intel.com>; simona@ffwll.ch; airlied@gmail.com;
>>> mripard@kernel.org; maarten.lankhorst@linux.intel.com
>>> Cc: dri-devel@lists.freedesktop.org; intel-gfx@lists.freedesktop.org; intel-
>>> xe@lists.freedesktop.org; Thomas Zimmermann <tzimmermann@suse.de>
>>> Subject: [PATCH] drm/dumb-buffers: Increase size limits to match current
>>> devices
>>>
>>> Increase the limits for dumb-buffer arguments to sizes of 16384x16384 at 64
>>> bpp. Such buffer sizes are still within 32-bit unsigned limits.
>>> Fixes CI and high-end devices.
>>>
>>> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
>>> Fixes: 5ab62dd3687b ("drm: prevent integer overflows in dumb buffer
>>> creation helpers")
>>> Reported-by: Jani Nikula <jani.nikula@linux.intel.com>
>>> Closes: https://lore.kernel.org/dri-
>>> devel/ddf0233e50044059c85279f928661563ef6a55bf@intel.com/
>> Seems not helping on i915 CI BAT issue https://lore.kernel.org/intel-gfx/178038951063.34290.3796875109342598028@6beec6c84f66/T/#u
>> => https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_167719v1/index.html?
> Right, so vgem_basic has this, among other things:
>
> 	bo.width = 1<<15;
> 	bo.height = 1<<15;
> 	bo.bpp = 16;
> 	vgem_create(fd, &bo);
>
> i.e. 32768x32768.
>
> Not so sure about this one.
>
> Anyone else care to chime in?

I give up. I'll send out a patch to remove the size limits in a bit. The 
rest of the overflow fixes in the broken patch should not be a problem, 
I think.

But TBH allocating a dumb buffer of that size makes no sense. Who uses a 
2-GiB buffer for software rendering?

And as I mentioned elsewhere, there are reports about devices with GiBs 
of VRAM but only a few hundered MiB of PCI-BAR ranges. As we don't have 
DMA here, there will be some size limits required to make mmap work 
correctly.

Best regards
Thomas

>
>
> BR,
> Jani.
>
>>> Cc: Rajat Gupta <rajat.gupta@oss.qualcomm.com>
>>> Cc: Thomas Zimmermann <tzimmermann@suse.de>
>>> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
>>> Cc: Maxime Ripard <mripard@kernel.org>
>> Br,
>> Jani
>>
>>> ---
>>>   drivers/gpu/drm/drm_dumb_buffers.c | 9 +++++----
>>>   1 file changed, 5 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/drivers/gpu/drm/drm_dumb_buffers.c
>>> b/drivers/gpu/drm/drm_dumb_buffers.c
>>> index e60130b2bb0c..9c4cfb3210aa 100644
>>> --- a/drivers/gpu/drm/drm_dumb_buffers.c
>>> +++ b/drivers/gpu/drm/drm_dumb_buffers.c
>>> @@ -201,11 +201,12 @@ int drm_mode_create_dumb(struct drm_device
>>> *dev,
>>>   	if (!args->width || !args->height || !args->bpp)
>>>   		return -EINVAL;
>>>
>>> -	/* Reject unreasonable inputs early.  Dumb buffers are for software
>>> -	 * rendering; nothing legitimate needs more than 8192x8192 at
>>> 32bpp.
>>> -	 * This prevents overflows in downstream alignment helpers.
>>> +	/*
>>> +	 * Reject unreasonable inputs early. Dumb buffers are for software
>>> +	 * rendering; nothing legitimate needs more than 16384x16384 at
>>> +	 * 64 bpp. This prevents overflows in downstream alignment helpers.
>>>   	 */
>>> -	if (args->width >= 8192 || args->height >= 8192 || args->bpp > 32)
>>> +	if (args->width > 16384 || args->height > 16384 || args->bpp > 64)
>>>   		return -EINVAL;
>>>
>>>   	/* overflow checks for 32bit size calculations */
>>>
>>> base-commit: a980196655477a8f5067112946401fe52e510664
>>> --
>>> 2.54.0

-- 
--
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Frankenstr. 146, 90461 Nürnberg, Germany, www.suse.com
GF: Jochen Jaser, Andrew McDonald, Werner Knoblich, (HRB 36809, AG Nürnberg)