On Mon, 2 Jun 2025 10:06:38 +1000
Jonathan Gray <j...@jsg.id.au> wrote:
> On Sat, May 31, 2025 at 07:15:22PM +0900, YASUOKA Masahiko wrote:
>> I noticed just after the changes on Feb 7 works fine.  So I did bisect
>> to find a commit that introduced the problem.  The commit is
>> 
>> On Tue, 11 Feb 2025 00:54:52 -0700 (MST)
>> Jonathan Gray <j...@cvs.openbsd.org> wrote:
>> > CVSROOT:   /cvs
>> > Module name:       src
>> > Changes by:        j...@cvs.openbsd.org    2025/02/11 00:54:52
>> > 
>> > Modified files:
>> >    sys/dev/pci/drm/include/generated: autoconf.h 
>> > 
>> > Log message:
>> > define CONFIG_FRAMEBUFFER_CONSOLE
>> > 
>> > used in inteldrm when deciding if an fb is too big for stolen memory
>> 
>> After revert this diff, the problem doesn't happen.
>> 
>> I have not see the detail of the flags yet, but it seems enabling the
>> flag causes the problem.
>> 
>> I'm sorry for my first diff was wrong direction.
> 
> The memory placement of the non-bios created framebuffer seems to be
> the problem.  I'll commit the below diff for now.
> 
> My Alder Lake machine has a lower resolution framebuffer, which may
> explain why I didn't see the same problem.

That makes sense.

> real mem = 16847925248 (16067MB)
> avail mem = 16309862400 (15554MB)
> random: good seed from bootblocks
> mpath0 at root
> scsibus0 at mpath0: 256 targets
> mainbus0 at root
> bios0 at mainbus0: SMBIOS rev. 3.4 @ 0x8f8a3000 (80 entries)
> bios0: vendor LENOVO version "N3MET18W (1.17 )" date 10/24/2023
> bios0: LENOVO 21AHCTO1WW
> ...
> inteldrm0 at pci0 dev 2 function 0 "Intel Graphics" rev 0x0c
> drm0 at inteldrm0
> inteldrm0: msi, ALDERLAKE_P, gen 12
> ...
> [drm] GGTT size = 4096M
> [drm] GMADR size = 256M
> [drm] DSM size = 64M
> [drm] GEN6_STOLEN_RESERVED = 0x00000000a0600087
> [drm] Memory reserved for graphics device: 65536K, usable: 63488K
> [drm] Memory region(0): system: 15562 MiB [0x0-0x3ccabefff], io: n/a
> [drm] Memory region(5): stolen-system: 62 MiB [0x9c800000-0xa05fffff], io: n/a
> ...
> inteldrm0: 1920x1200, 32bpp
> 
> size * 2 (18432000) < dsm usable (65011712)

my vaio:

real mem = 16843354112 (16063MB)
avail mem = 16306778112 (15551MB)
random: good seed from bootblocks
mpath0 at root
scsibus0 at mpath0: 256 targets
mainbus0 at root
bios0 at mainbus0: SMBIOS rev. 3.4 @ 0x2ee40000 (38 entries)
bios0: vendor Phoenix Technologies Ltd. version "R0160ZR" date 09/06/2023
bios0: VAIO Corporation VJPK23
...
inteldrm0 at pci0 dev 2 function 0 "Intel Graphics" rev 0x04
drm0 at inteldrm0
inteldrm0: msi, ALDERLAKE_P, gen 12
...
GTT size = 4096M
GMADR size = 256M
DSM size = 60M
GEN6_STOLEN_RESERVED = 0x0000000050200087
Memory reserved for graphics device: 61440K, usable: 59392K
Memory region(0): system: 15559 MiB [0x0-0x3cc7d1fff], io: n/a
Memory region(5): stolen-system: 58 MiB [0x4c800000-0x501fffff], io: n/a
...
inteldrm0: 3840x2160, 32bpp

size * 2 (66355200) > dsm usable (60817408)

Thanks,

> Index: sys/dev/pci/drm/include/generated/autoconf.h
> ===================================================================
> RCS file: /cvs/src/sys/dev/pci/drm/include/generated/autoconf.h,v
> diff -u -p -r1.18 autoconf.h
> --- sys/dev/pci/drm/include/generated/autoconf.h      14 Mar 2025 03:39:23 
> -0000      1.18
> +++ sys/dev/pci/drm/include/generated/autoconf.h      1 Jun 2025 23:55:06 
> -0000
> @@ -6,7 +6,12 @@
>  #define CONFIG_BACKLIGHT_CLASS_DEVICE                1
>  #define CONFIG_DRM_FBDEV_EMULATION           1
>  #define CONFIG_DRM_CLIENT_SETUP                      1
> +
> +#ifdef notyet
> +/* causes Intel GuC init to fail with large fbs */
>  #define CONFIG_FRAMEBUFFER_CONSOLE           1
> +#endif
> +
>  #define CONFIG_DRM_PANEL                     1
>  #define CONFIG_DRM_I915_CAPTURE_ERROR                1
>  #define CONFIG_DRM_AMD_DC                    1
> 

Reply via email to