From Newsgroup: muc.lists.freebsd.stable

Seems like installing drm-66-kmod instead fixed the stabiliy issues

$ pkg info drm-66-kmod
drm-66-kmod-6.6.25.1501501_10
Name : drm-66-kmod
Version : 6.6.25.1501501_10
Installed on : Wed Aug 12 19:03:11 2026 CEST
Origin : graphics/drm-66-kmod

FreeBSD garfield 15.1-STABLE FreeBSD 15.1-STABLE stable/15-n284839-a8c598de78fd GENERIC amd64


--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-admin@muc.de
--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: muc.lists.freebsd.stable


The problem was fixed after recompiling drm-latest-kmod against the
latest sources.

Best,
R


--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-admin@muc.de
--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: muc.lists.freebsd.stable

Hello again,i have managed a "hack" to get drm-6.12-kmod working on our old IGPUs,the hack is to simply return version 3 unconditionally for i915_gem_mmap_gtt_version (in src/drivers/gpu/drm/i915/gem/i915_gem_mman.c)
the reason this is a hack because it will break for any modern intel gpu because you can't get the legacy ioctl working for them which version 3 uses
an actual fix would be to fix LinuxKPI's page sharing behaviour(currently it seems linuxkpi can't represent two mappings of the same page)
On Thursday, August 13th, 2026 at 1:32 AM, Anders Gulden Olstad <andersgo@abbor.no> wrote:

Hi

Yes my GPU/PC is getting quite old

Replacing the drm-latest with drm-66 seems to have fixed the issue. I installed it from ports.

Br
Anders

On 8/12/26 20:56, polyduekes@proton.me wrote:

this is interesting, i was thinking lkpi_vmf_insert_pfn_prot_locked: page rename failed: page is mapped was something only i was having ONLY on drm-6.12-kmod(drm 6.6 and drm 6.1 work fine) because my igpu is old, i have filed a github issue for this sometimes ago btw if you are interested
https://github.com/freebsd/drm-kmod/issues/481

i have been trying to debug this for some days now and i am 90% sure what the issue is

-------- Original Message --------
On Wednesday, 08/12/26 at 12:16 Anders Gulden Olstad <andersgo@abbor.no> wrote:
On 8/12/26 08:32, Anders Gulden Olstad wrote:

On 8/9/26 20:18, Mark Millard wrote:

On 8/9/26 02:58, Ricardo Branco wrote:

Attached core.txt.5.gz

Dump header from device: /dev/ada0p3
Architecture: amd64
Architecture Version: 2
Dump Length: 1162715136
Blocksize: 512
Compression: none
Dumptime: 2026-08-08 10:02:28 +0200
Hostname:
Magic: FreeBSD Kernel Dump
Version String: FreeBSD 15.1-STABLE stable/15-n284803-c0f892003486 >>>> GENERIC

Got the same issue. Compiled drm-latest-kmod from ports to get X running >> again.

But Xorg is now crashing/respawning from time to time when I'm using
Firefox (video streaming etc)

drmn0: [drm] GPU HANG: ecode 7:1:85dffffc, in Renderer [104909]
drmn0: [drm] Resetting rcs0 for stopped heartbeat on rcs0
drmn0: [drm] Renderer[104909] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 7:1:85dffff8, in MainThread [100895]
drmn0: [drm] Resetting rcs0 for stopped heartbeat on rcs0
drmn0: [drm] Xorg[100895] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 7:1:85dffff8, in MainThread [100895]
drmn0: [drm] Resetting rcs0 for stopped heartbeat on rcs0
drmn0: [drm] MainThread[100895] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 7:1:85dffff8, in MainThread [100895]
drmn0: [drm] Resetting rcs0 for stopped heartbeat on rcs0
drmn0: [drm] MainThread[100895] context reset due to GPU hang
pid 3662 (Xorg), jid 0, uid 0: exited on signal 6 (core dumped)

This is something I have never experienced before

$ pkg info drm-latest-kmod
drm-latest-kmod-6.12.1501501
Name : drm-latest-kmod
Version : 6.12.1501501
Installed on : Tue Aug 11 09:11:53 2026 CEST
Origin : graphics/drm-latest-kmod

Now X/Firefox went totally haywire and becomes unresponsive. Firefox
churns away at 100% CPU and it doesn't take any kill -9 signals .

Aug 12 08:23:51 garfield kernel: pid 3662 (Xorg), jid 0, uid 0: exited
on signal 6 (core dumped)

Aug 12 08:23:51 garfield devd[2121]: notify_clients: send() failed; dropping unresponsive client

Aug 12 08:23:51 garfield syslogd: last message repeated 1 times

Aug 12 08:23:51 garfield devd[2121]: check_clients: dropping
disconnected client

Aug 12 08:23:51 garfield syslogd: last message repeated 1 times

Aug 12 08:35:42 garfield pulseaudio[12113]: [] oss-util.c: '/dev/dsp0' doesn't support full duplex

Aug 12 08:35:42 garfield pulseaudio[12113]: [] oss-util.c: '/dev/dsp2' doesn't support full duplex

Aug 12 08:35:42 garfield pulseaudio[12113]: [] cli-command.c: stat('/usr/local/etc/pulse/default.pa.d'): No such file or directory

Aug 12 08:36:24 garfield kernel: ked: page rename failed: page is mapped

Aug 12 08:36:24 garfield kernel: lkpi_vmf_insert_pfn_prot_locked: page rename failed: page is mapped

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-admin@muc.de
--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: muc.lists.freebsd.stable

hi! please file a bug about this so we can track it. Which specific
iGPUs are you seeing this issue on?


-a

On Fri, 14 Aug 2026 at 12:18, <polyduekes@proton.me> wrote:

Hello again,i have managed a "hack" to get drm-6.12-kmod working on our old IGPUs,the hack is to simply return version 3 unconditionally for i915_gem_mmap_gtt_version (in src/drivers/gpu/drm/i915/gem/i915_gem_mman.c)
the reason this is a hack because it will break for any modern intel gpu because you can't get the legacy ioctl working for them which version 3 uses
an actual fix would be to fix LinuxKPI's page sharing behaviour(currently it seems linuxkpi can't represent two mappings of the same page)
On Thursday, August 13th, 2026 at 1:32 AM, Anders Gulden Olstad <andersgo@abbor.no> wrote:

Hi

Yes my GPU/PC is getting quite old

Replacing the drm-latest with drm-66 seems to have fixed the issue. I installed it from ports.

Br
Anders

On 8/12/26 20:56, polyduekes@proton.me wrote:

this is interesting, i was thinking lkpi_vmf_insert_pfn_prot_locked: page rename failed: page is mapped was something only i was having ONLY on drm-6.12-kmod(drm 6.6 and drm 6.1 work fine) because my igpu is old, i have filed a github issue for this sometimes ago btw if you are interested
https://github.com/freebsd/drm-kmod/issues/481

i have been trying to debug this for some days now and i am 90% sure what the issue is

-------- Original Message --------
On Wednesday, 08/12/26 at 12:16 Anders Gulden Olstad <andersgo@abbor.no> wrote:
On 8/12/26 08:32, Anders Gulden Olstad wrote:

On 8/9/26 20:18, Mark Millard wrote:

On 8/9/26 02:58, Ricardo Branco wrote:

Attached core.txt.5.gz

Dump header from device: /dev/ada0p3
Architecture: amd64
Architecture Version: 2
Dump Length: 1162715136
Blocksize: 512
Compression: none
Dumptime: 2026-08-08 10:02:28 +0200
Hostname:
Magic: FreeBSD Kernel Dump
Version String: FreeBSD 15.1-STABLE stable/15-n284803-c0f892003486 >>>> GENERIC

Got the same issue. Compiled drm-latest-kmod from ports to get X running >> again.

But Xorg is now crashing/respawning from time to time when I'm using
Firefox (video streaming etc)

drmn0: [drm] GPU HANG: ecode 7:1:85dffffc, in Renderer [104909]
drmn0: [drm] Resetting rcs0 for stopped heartbeat on rcs0
drmn0: [drm] Renderer[104909] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 7:1:85dffff8, in MainThread [100895]
drmn0: [drm] Resetting rcs0 for stopped heartbeat on rcs0
drmn0: [drm] Xorg[100895] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 7:1:85dffff8, in MainThread [100895]
drmn0: [drm] Resetting rcs0 for stopped heartbeat on rcs0
drmn0: [drm] MainThread[100895] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 7:1:85dffff8, in MainThread [100895]
drmn0: [drm] Resetting rcs0 for stopped heartbeat on rcs0
drmn0: [drm] MainThread[100895] context reset due to GPU hang
pid 3662 (Xorg), jid 0, uid 0: exited on signal 6 (core dumped)

This is something I have never experienced before

$ pkg info drm-latest-kmod
drm-latest-kmod-6.12.1501501
Name : drm-latest-kmod
Version : 6.12.1501501
Installed on : Tue Aug 11 09:11:53 2026 CEST
Origin : graphics/drm-latest-kmod

Now X/Firefox went totally haywire and becomes unresponsive. Firefox churns away at 100% CPU and it doesn't take any kill -9 signals .

Aug 12 08:23:51 garfield kernel: pid 3662 (Xorg), jid 0, uid 0: exited
on signal 6 (core dumped)

Aug 12 08:23:51 garfield devd[2121]: notify_clients: send() failed; dropping unresponsive client

Aug 12 08:23:51 garfield syslogd: last message repeated 1 times

Aug 12 08:23:51 garfield devd[2121]: check_clients: dropping disconnected client

Aug 12 08:23:51 garfield syslogd: last message repeated 1 times

Aug 12 08:35:42 garfield pulseaudio[12113]: [] oss-util.c: '/dev/dsp0' doesn't support full duplex

Aug 12 08:35:42 garfield pulseaudio[12113]: [] oss-util.c: '/dev/dsp2' doesn't support full duplex

Aug 12 08:35:42 garfield pulseaudio[12113]: [] cli-command.c: stat('/usr/local/etc/pulse/default.pa.d'): No such file or directory

Aug 12 08:36:24 garfield kernel: ked: page rename failed: page is mapped

Aug 12 08:36:24 garfield kernel: lkpi_vmf_insert_pfn_prot_locked: page rename failed: page is mapped

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-admin@muc.de
--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: muc.lists.freebsd.stable

if you mean i should create a bug report against drm 6.12 kmod, then i have already filed one on bugzilla https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=296448 and one on drm-kmod gh https://github.com/freebsd/drm-kmod/issues/481, the igpu is intel i5 650, if you meant that i should file a pr against base then i would refrain to do until someone more knowledgeable then me can confirm this is indeed due to lkpi being unable to do page sharing
-------- Original Message --------
On Sunday, 08/16/26 at 05:59 Adrian Chadd <adrian@freebsd.org> wrote:
hi! please file a bug about this so we can track it. Which specific
iGPUs are you seeing this issue on?
-a
On Fri, 14 Aug 2026 at 12:18, <polyduekes@proton.me> wrote:

Hello again,i have managed a "hack" to get drm-6.12-kmod working on our old IGPUs,the hack is to simply return version 3 unconditionally for i915_gem_mmap_gtt_version (in src/drivers/gpu/drm/i915/gem/i915_gem_mman.c)
the reason this is a hack because it will break for any modern intel gpu because you can't get the legacy ioctl working for them which version 3 uses
an actual fix would be to fix LinuxKPI's page sharing behaviour(currently it seems linuxkpi can't represent two mappings of the same page)
On Thursday, August 13th, 2026 at 1:32 AM, Anders Gulden Olstad <andersgo@abbor.no> wrote:

Hi

Yes my GPU/PC is getting quite old

Replacing the drm-latest with drm-66 seems to have fixed the issue. I installed it from ports.

Br
Anders

On 8/12/26 20:56, polyduekes@proton.me wrote:

this is interesting, i was thinking lkpi_vmf_insert_pfn_prot_locked: page rename failed: page is mapped was something only i was having ONLY on drm-6.12-kmod(drm 6.6 and drm 6.1 work fine) because my igpu is old, i have filed a github issue for this sometimes ago btw if you are interested
https://github.com/freebsd/drm-kmod/issues/481

i have been trying to debug this for some days now and i am 90% sure what the issue is

-------- Original Message --------
On Wednesday, 08/12/26 at 12:16 Anders Gulden Olstad <andersgo@abbor.no> wrote:
On 8/12/26 08:32, Anders Gulden Olstad wrote:

On 8/9/26 20:18, Mark Millard wrote:

On 8/9/26 02:58, Ricardo Branco wrote:

Attached core.txt.5.gz

Dump header from device: /dev/ada0p3
Architecture: amd64
Architecture Version: 2
Dump Length: 1162715136
Blocksize: 512
Compression: none
Dumptime: 2026-08-08 10:02:28 +0200
Hostname:
Magic: FreeBSD Kernel Dump
Version String: FreeBSD 15.1-STABLE stable/15-n284803-c0f892003486 >>>> GENERIC

Got the same issue. Compiled drm-latest-kmod from ports to get X running >> again.

But Xorg is now crashing/respawning from time to time when I'm using
Firefox (video streaming etc)

drmn0: [drm] GPU HANG: ecode 7:1:85dffffc, in Renderer [104909]
drmn0: [drm] Resetting rcs0 for stopped heartbeat on rcs0
drmn0: [drm] Renderer[104909] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 7:1:85dffff8, in MainThread [100895]
drmn0: [drm] Resetting rcs0 for stopped heartbeat on rcs0
drmn0: [drm] Xorg[100895] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 7:1:85dffff8, in MainThread [100895]
drmn0: [drm] Resetting rcs0 for stopped heartbeat on rcs0
drmn0: [drm] MainThread[100895] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 7:1:85dffff8, in MainThread [100895]
drmn0: [drm] Resetting rcs0 for stopped heartbeat on rcs0
drmn0: [drm] MainThread[100895] context reset due to GPU hang
pid 3662 (Xorg), jid 0, uid 0: exited on signal 6 (core dumped)

This is something I have never experienced before

$ pkg info drm-latest-kmod
drm-latest-kmod-6.12.1501501
Name : drm-latest-kmod
Version : 6.12.1501501
Installed on : Tue Aug 11 09:11:53 2026 CEST
Origin : graphics/drm-latest-kmod

Now X/Firefox went totally haywire and becomes unresponsive. Firefox churns away at 100% CPU and it doesn't take any kill -9 signals .

Aug 12 08:23:51 garfield kernel: pid 3662 (Xorg), jid 0, uid 0: exited
on signal 6 (core dumped)

Aug 12 08:23:51 garfield devd[2121]: notify_clients: send() failed; dropping unresponsive client

Aug 12 08:23:51 garfield syslogd: last message repeated 1 times

Aug 12 08:23:51 garfield devd[2121]: check_clients: dropping disconnected client

Aug 12 08:23:51 garfield syslogd: last message repeated 1 times

Aug 12 08:35:42 garfield pulseaudio[12113]: [] oss-util.c: '/dev/dsp0' doesn't support full duplex

Aug 12 08:35:42 garfield pulseaudio[12113]: [] oss-util.c: '/dev/dsp2' doesn't support full duplex

Aug 12 08:35:42 garfield pulseaudio[12113]: [] cli-command.c: stat('/usr/local/etc/pulse/default.pa.d'): No such file or directory

Aug 12 08:36:24 garfield kernel: ked: page rename failed: page is mapped

Aug 12 08:36:24 garfield kernel: lkpi_vmf_insert_pfn_prot_locked: page rename failed: page is mapped

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-admin@muc.de
--- Synchronet 3.22a-Linux NewsLink 1.2