The issue is that DRI using apps, started complaining about mesa. E.g.: glxgears complains on start: MESA: error: ZINK: failed to choose pdev glx: failed to create drisw screen failed to load driver: zink This also happens in some other apps, and also with mpv, which prints the similar error and then falls back to vo=x11. I think it's started with 23.3.0. I've also updated once again recently, and now it's at 23.3.1, but the problem persists. The card I have: vgapci0@pci0:0:2:0: class=0x030000 rev=0x0c hdr=0x00 vendor=0x8086 device=0x4692 subvendor=0x103c subdevice=0x89b5 vendor = 'Intel Corporation' device = 'Alder Lake-S GT1 [UHD Graphics 730]' class = display subclass = VGA
I have this issue too. In my case it's enough to run emacs (from the shell command line - no X11): ~ % emacs & [1] 92567 marek@simba:~ % MESA: error: ZINK: failed to choose pdev glx: failed to create drisw screen failed to load driver: zink [1] Done emacs ~ % 13.2-RELEASE-p9 FreeBSD simba 13.2-RELEASE-p8 FreeBSD 13.2-RELEASE-p8 GENERIC amd64 vgapci0@pci0:0:2:0: class=0x030000 rev=0x09 hdr=0x00 vendor=0x8086 device=0x0166 subvendor=0x103c subdevice=0x179b vendor = 'Intel Corporation' device = '3rd Gen Core processor Graphics Controller' class = display subclass = VGA
Can you reproduce with graphics/mesa-devel? If may have an upstream fix. Otherwise, it doesn't include zink and swrast, so should fail hard instead of giving buggy experience. Can you reproduce under x11-wm/ctwm? It doesn't have OpenGL or XRENDER compositor thus relying on very primitive framebuffer compositing built into Xorg itself. May be related to bug 267915 i.e., mesa-dri regression before it was conflated with drm-kmod regression.
1. Rectification to my comment (#1): it is not true that there was no X11. It's true that I started emacs from the command line but that was from within xterm run under xming (on Windows 11). 2. After installing graphics/mesa-devel the same steps lead to the following messages: ~ % emacs & [1] 43239 ~ % MESA-LOADER: failed to open zink: Cannot open "/usr/local/lib/dri-devel/zink_dri.so" (search paths /usr/local/lib/dri-devel, suffix _dri) failed to load driver: zink MESA-LOADER: failed to open swrast: Cannot open "/usr/local/lib/dri-devel/swrast_dri.so" (search paths /usr/local/lib/dri-devel, suffix _dri) failed to load driver: swrast [1] Done emacs ~ % 3. I installed x11-wm/ctwm, but I can't seem to get it to work. I tried a simple xinit, but it seems to run twm, not ctwm (no error messages above, by the way). 4. I noticed the issue today right after pkg upgrade'ing (if I noticed correctly one of the updated packages was mesa). There was no such problem before the upgrade.
Created attachment 247597 [details] Xorg.0.log of ctwm startup (and termination)
(In addition to marek, comment #4) I was finally able to run ctwm (echo "exec ctwm" > ~/.xinitrc; xinit). See the attached Xorg.0.log (I don't see any mesa-related errors in it).
Try actually running something OpenGL *under* ctwm e.g., if emacs no longer prints zink warning it'd be a new finding. The point is to check if X11 compositor sabotages regular X11 clients. For example, kwin_x11 (Plasma) or mutter (GNOME) do too much and are known sources of Xorg-specific bugs like bug 251836 or bug 253746.
Alternatively, neither comment 0 nor comment 1 provided enough details to reproduce (not even drm-*kmod version). DRI/X11 works fine on my Intel iGPU (Skylake, 0x1912) with drm-515-kmod and mesa-devel: tested Xwayland (sway-devel) and Xorg (dwm + picom or ctwm, modesetting or xf86-video-intel + SNA + DRI3). Like comment 0 I use mpv (mainly with --vo=gpu-next --gpu-api=vulkan --hwdec=vulkan) and like comment 1 I use emacs (mainly -devel and PGTK=on).
Regarding versions, it is (after today's upgrade): - mesa-devel-23.3.b.3608 - mesa-dri-23.3.2 - mesa-libs-23.3.2 - drm-510-kmod-5.10.163_8 - libdrm-2.4.119,1 No error messages are diplayed when emacs and glxgears run locally (directly on FreeBSD laptop). However, when run remotely, in an environment provided by XMING or MOBAXTERM (on Windows 11), they display the following: ~ % emacs & [1] 59259 ~ % MESA-LOADER: failed to open zink: Cannot open "/usr/local/lib/dri-devel/zink_dri.so" (search paths /usr/local/lib/dri-devel, suffix _dri) failed to load driver: zink MESA-LOADER: failed to open swrast: Cannot open "/usr/local/lib/dri-devel/swrast_dri.so" (search paths /usr/local/lib/dri-devel, suffix _dri) failed to load driver: swrast [1] Done emacs ~ % glxgears & [1] 64849 ~ % MESA-LOADER: failed to open zink: Cannot open "/usr/local/lib/dri-devel/zink_dri.so" (search paths /usr/local/lib/dri-devel, suffix _dri) failed to load driver: zink MESA-LOADER: failed to open swrast: Cannot open "/usr/local/lib/dri-devel/swrast_dri.so" (search paths /usr/local/lib/dri-devel, suffix _dri) failed to load driver: swrast Running synchronized to the vertical refresh. The framerate should be approximately the same as the monitor refresh rate. X connection to localhost:10.0 broken (explicit kill or server shutdown). [1] Exit 1 glxgears ~ %
Don't you think it's the difference in driver modules? You have drm-515-kmod and I have drm-510-kmod (and zink and swrast are really missing in the location stated in error messages).
(In reply to marek from comment #8) Remove mesa-devel. It's not suitable for X11 forwarding: DRI mostly expects local GPU, otherwise falls back to swrast (llvmpipe) which mesa-devel (unlike mesa-dri) doesn't provide. Sorry, I can't test remote X11 on my side, let alone via XMING or MOBAXTERM. (In reply to marek from comment #9) Unlikely. Remote X11 (except VirtualGL, WSLg) with 3D rendering (via OpenGL) instead of 2D rendering (via pixman, cairo, skia, etc) is simply not tested much. Regressions usually come from mesa-* or xorg-server updates.
(In reply to Jan Beich from comment #7) I have: mesa-dri-23.3.1 OpenGL hardware acceleration drivers for DRI2+ mesa-libs-23.3.1 OpenGL libraries that support GLX and EGL clients drm-515-kmod-5.15.118_2 DRM drivers modules gpu-firmware-kmod-20230210_1,1 Firmware modules for the drm-kmod drivers libdrm-2.4.118,1 Userspace interface to kernel Direct Rendering Module services I use openbox. I haven't yet updated after reporting the issue. IIRC, I've tried mesa-devel as well but it didn't help. I think I'll update this weekend and check things again.
(In reply to Roman Bogorodskiy from comment #11) I forgot to mention that I'm using x11-wm/openbox: $ cat ~/.xinitrc exec ssh-agent openbox-session $
(In reply to Roman Bogorodskiy from comment #0) It's a bit weird that it tries to use zink. Note that I don't have any problem using x11 on my skylake or apollolake machine. Alderlake support in 5.15 is present but afaik not really stable so I wonder if previously you didn't had gpu acceleration and since the update mesa tries to do opengl via zink for some reason and obviously it doesn't work as you also have no Vulkan. Can you post the output of glxinfo -B please ? (pkg install mesa-demos).
(In reply to Emmanuel Vadot from comment #13) $ glxinfo -B name of display: :0 MESA: error: ZINK: failed to choose pdev glx: failed to create drisw screen failed to load driver: zink display: :0 screen: 0 direct rendering: Yes Extended renderer info (GLX_MESA_query_renderer): Vendor: Mesa (0xffffffff) Device: llvmpipe (LLVM 15.0.7, 256 bits) (0xffffffff) Version: 23.3.1 Accelerated: no Video memory: 32768MB Unified memory: yes Preferred profile: core (0x1) Max core profile version: 4.5 Max compat profile version: 4.5 Max GLES1 profile version: 1.1 Max GLES[23] profile version: 3.2 Memory info (GL_ATI_meminfo): VBO free memory - total: 10 MB, largest block: 10 MB VBO free aux. memory - total: 30296 MB, largest block: 30296 MB Texture free memory - total: 0 MB, largest block: 0 MB Texture free aux. memory - total: 30296 MB, largest block: 30296 MB Renderbuffer free memory - total: 0 MB, largest block: 0 MB Renderbuffer free aux. memory - total: 30296 MB, largest block: 30296 MB Memory info (GL_NVX_gpu_memory_info): Dedicated video memory: 531525 MB Total available memory: 564293 MB Currently available dedicated video memory: 0 MB OpenGL vendor string: Mesa OpenGL renderer string: llvmpipe (LLVM 15.0.7, 256 bits) OpenGL core profile version string: 4.5 (Core Profile) Mesa 23.3.1 OpenGL core profile shading language version string: 4.50 OpenGL core profile context flags: (none) OpenGL core profile profile mask: core profile OpenGL version string: 4.5 (Compatibility Profile) Mesa 23.3.1 OpenGL shading language version string: 4.50 OpenGL context flags: (none) OpenGL profile mask: compatibility profile OpenGL ES profile version string: OpenGL ES 3.2 Mesa 23.3.1 OpenGL ES profile shading language version string: OpenGL ES GLSL ES 3.20 $
Right so I think that my theory is correct, I think that you never had gpu accel and now you have this weird side effect with newer mesa that it tries to use zink to have opengl because the main opengl didn't worked. AlderLake is better supported with drm-61-kmod but that needs -CURRENT.
(In reply to Emmanuel Vadot from comment #15) Yes, I think I didn't have the gpu accel too. However, I've been using the scfb driver and I *think* mpv used some more sane vo than x11 (at least it didn't print all these error messages). I'm actually running -CURRENT, but didn't follow drm-kmod development closely. I've switched to drm-61-kmod and it appears to work fine so far.
I believe I'm having this problem as well, but I have an AMD Radeon RX 550. After recent updates (a few weeks ago) on the quarterly branch, I was no longer able to launch Firefox from an XRDP session and saw the failure to load zink when launching it from a terminal. However, it launched fine locally. I also have a "guest" profile in Firefox that I use for testing that did load in the XRDP session, so I looked for preferences I'd changed in my main profile and found gfx.webrender.all: https://wiki.mozilla.org/Platform/GFX/Quantum_Render With that enabled, my main Firefox profile fails to load in an XRDP session. With it disabled, it loads fine. I can toggle the setting to provide output/logs, if needed. Maybe it would be helpful to see data from 1) working locally, 2) broken in XRDP, and 3) fixed in XRDP? I just need to know what information might be useful. To start, this is my video card: vgapci0@pci0:16:0:0: class=0x030000 rev=0xff hdr=0x00 vendor=0x1002 device=0x67ff subvendor=0x1682 subdevice=0x9550 vendor = 'Advanced Micro Devices, Inc. [AMD/ATI]' device = 'Baffin [Radeon RX 550 640SP / RX 560/560X]' class = display subclass = VGA cap 09[48] = vendor (length 8) cap 01[50] = powerspec 3 supports D0 D1 D2 D3 current D0 cap 10[58] = PCI-Express 2 legacy endpoint max data 256(256) RO NS max read 512 link x8(x8) speed 2.5(8.0) ASPM disabled(L1) cap 05[a0] = MSI supports 1 message, 64 bit enabled with 1 message ecap 000b[100] = Vendor [1] ID 0001 Rev 1 Length 16 ecap 0001[150] = AER 2 0 fatal 0 non-fatal 1 corrected ecap 0015[200] = Resizable BAR 1 ecap 0019[270] = PCIe Sec 1 lane errors 0 ecap 000f[2b0] = ATS 1 ecap 0013[2c0] = Page Page Request 1 ecap 001b[2d0] = Process Address Space ID 1 ecap 0018[320] = LTR 1 ecap 000e[328] = ARI 1 ecap 001e[370] = L1 PM Substates 1 DRM: pkg info | grep drm drm-510-kmod-5.10.163_8 DRM drivers modules drm-kmod-20220907_1 Metaport of DRM modules for the linuxkpi-based KMS components gpu-firmware-kmod-20230210_1,1 Firmware modules for the drm-kmod drivers Mesa: pkg info | grep mesa mesa-demos-8.4.0_3 OpenGL demos distributed with Mesa mesa-dri-23.3.4 OpenGL hardware acceleration drivers for DRI2+ mesa-libs-23.3.4 OpenGL libraries that support GLX and EGL clients If you believe my issue is unrelated to this one, just tell me to buzz off.