linux-kernel - Re: [PATCH v1 3/4] drm/ls2kbmc: Add support for Loongson-2K BMC display

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 for Android: free password hash cracker in your pocket
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAMpQs4L38rDEDYM64jJ6pO+g=M4+etKN9v9+ygzkLY6RQgu94A@mail.gmail.com>
Date: Mon, 6 Jan 2025 15:03:23 +0800
From: Binbin Zhou <zhoubb.aaron@...il.com>
To: Thomas Zimmermann <tzimmermann@...e.de>
Cc: Binbin Zhou <zhoubinbin@...ngson.cn>, Huacai Chen <chenhuacai@...ngson.cn>, 
	Lee Jones <lee@...nel.org>, Corey Minyard <minyard@....org>, 
	Maarten Lankhorst <maarten.lankhorst@...ux.intel.com>, Maxime Ripard <mripard@...nel.org>, 
	David Airlie <airlied@...il.com>, Simona Vetter <simona@...ll.ch>, Huacai Chen <chenhuacai@...nel.org>, 
	linux-kernel@...r.kernel.org, openipmi-developer@...ts.sourceforge.net, 
	dri-devel@...ts.freedesktop.org, Xuerui Wang <kernel@...0n.name>, 
	loongarch@...ts.linux.dev, Chong Qiao <qiaochong@...ngson.cn>
Subject: Re: [PATCH v1 3/4] drm/ls2kbmc: Add support for Loongson-2K BMC display

Hi Thomas:

The last reply email was incomplete, sorry for the incomplete reply
due to my mistake.

On Thu, Jan 2, 2025 at 9:32 PM Thomas Zimmermann <tzimmermann@...e.de> wrote:
>
> Hi
>
>
> Am 02.01.25 um 13:55 schrieb Binbin Zhou:
> > Hi Thomas:
> >
> > Thanks for your reply.
> >
> > On Thu, Jan 2, 2025 at 5:07 PM Thomas Zimmermann <tzimmermann@...e.de> wrote:
> >> Hi
> >>
> >>
> >> Am 30.12.24 um 10:31 schrieb Binbin Zhou:
> >> [...]
> >>> +
> >>> +static struct platform_driver ls2kbmc_platform_driver = {
> >>> +     .driver = {
> >>> +             .name = "ls2kbmc-framebuffer",
> >> The driver is mostly a copy of simpledrm. Why don't you use
> >> "simple-framebuffer" for your device name? You could use simpledrm
> >> directly then.
> > Ah, indeed, the driver is based on simpledrm.
> >
> > Initially, I also tried to use simpledrm directly, but it will fail in
> > drm memory acquire.
>
> Could you point to the exact call that fails within simpledrm?

If we use simpledrm directly, the following error occurs:

[    8.289823] simple-framebuffer simple-framebuffer.0: [drm] *ERROR*
could not acquire memory range [mem 0xe0031200000-0xe00315fffff flags
0x200]: -16
[    8.312681] simple-framebuffer simple-framebuffer.0: probe with
driver simple-framebuffer failed with error -16

The reason for the failure: overlapping resources.

https://elixir.bootlin.com/linux/v6.12.6/source/drivers/video/aperture.c#L175
>
> > Because although we register the driver in platform form, its memory
> > belongs to pci space and we can see the corresponding pci probe and
> > resource allocation in Patch-1.
>
> I don't understand. Graphics memory is often located on the PCI bus.
> What is so special about this one?
>
> > Therefore, we need to use aperture_remove_conflicting_pci_devices().
>
> So there is already a device that represents the graphics card? That's
> what you'd remove here? If you only add that MFD device, who owns the
> framebuffer? If it's the PCI device from patch 1 ("ls2k-bmc"), why does
> aperture_remove_conflicting_pci_devices() not remove that device? I'm
> somewhat confused, because the logic in your driver mostly looks like it
> binds to a pre-configured framebuffer, but some of the code doesn't.

Perhaps the use of aperture_remove_conflicting_pci_devices() is wrong,
as there is only one display device for the LS2K BMC and there will be
no phase conflict.

When I tried to use that API before, it was partly due to the error
above, and partly because I referenced that other display drivers via
pci_driver.probe() would have it, just in case I used it, which was
probably the wrong choice.

The resources for pci bar0 are as follows:
BAR0: e0030000000/SZ_32M

0x0              0x600000  0xf00001c    16M            32M
+----+--------------+--------+-----------+---+-----------------+
| 2M | simpldrm |           | IPMI      |     | video env     |
+-----------------------------------------------------------------+

The mfd driver registers the ls2kbmc-framebuffer and ls2k-ipmi-si
devices according to the resource allocation shown above. At the same
time, the ls2kbmc drm is bound to the pre-configured “simpldrm”
resource in the above figure, which is passed through the
ls2kbmc-framebuffer driver. In addition, the resolution is read from
“video env” for the time being, and the resolution adaption is planned
to be added later.

> Best regards Thomas
>
> >
> > Also, since we are using BMC display, the display will be disconnected
> > when BMC reset, at this time we need to push the display data (crtc,
> > connector, etc.) manually as shown in Patch-4.
> >
> > Probably it's not the most suitable way to implement it.
> >
> >> Best regards
> >> Thomas
> >>
> >>> +     },
> >>> +     .probe = ls2kbmc_probe,
> >>> +     .remove = ls2kbmc_remove,
> >>> +};
> >>> +
> >>> +module_platform_driver(ls2kbmc_platform_driver);
> >>> +
> >>> +MODULE_DESCRIPTION("DRM driver for Loongson-2K BMC");
> >>> +MODULE_LICENSE("GPL");
> >> --
> >> --
> >> Thomas Zimmermann
> >> Graphics Driver Developer
> >> SUSE Software Solutions Germany GmbH
> >> Frankenstrasse 146, 90461 Nuernberg, Germany
> >> GF: Ivo Totev, Andrew Myers, Andrew McDonald, Boudien Moerman
> >> HRB 36809 (AG Nuernberg)
> >>
> >
>
> --
> --
> Thomas Zimmermann
> Graphics Driver Developer
> SUSE Software Solutions Germany GmbH
> Frankenstrasse 146, 90461 Nuernberg, Germany
> GF: Ivo Totev, Andrew Myers, Andrew McDonald, Boudien Moerman
> HRB 36809 (AG Nuernberg)
>


--
Thanks.
Binbin