Hi everyone, I’m posting to document a couple of severe regressions I hit after pulling the latest system updates. I’ve already rolled back via Btrfs snapshots which resolved both issues, but I wanted to get this on the radar for anyone else experiencing them or for the devs tracking upstream bugs. System Information: Hardware: Lenovo Legion 5 (15IRX10) GPU: NVIDIA RTX 5060 (Mobile) + Intel iGPU Compositor: Wayland (Niri / Hyprland) Broken Kernel: linux-cachyos 7.0.1-1 Broken NVIDIA Driver: nvidia-open-dkms 595.58.03-2 Issue 1: Massive VRAM Leak & Vulkan Crash in CS2 (nvidia-open) After the update, launching CS2 in hybrid mode under Wayland resulted in a massive memory leak, ultimately crashing the game with a VK_ERROR_DEVICE_LOST error. Looking at dmesg , the nvidia-open driver completely fails to manage memory (likely a DMA buffer/explicit sync issue with Wayland frame copying to the iGPU), filling up VRAM until the GSP firmware crashes and throws an MMU fault. Relevant dmesg logs right before the crash: [ 5.095284] NVRM: GPU0 nvAssertOkFailedNoLog: Assertion failed: Invalid data passed [NV_ERR_INVALID_DATA] (0x00000025) returned from PlatformRequestHandler failed to get target temp from SBIOS @ platform_request_handler_ctrl.c:2171 [22389.283459] NVRM: Xid (PCI:0000:01:00): 31, pid=98700, name=cs2, channel 0x0000000c, intr 00000000. MMU Fault: ENGINE GRAPHICS GPC1 GPCCLIENT_PROP_0 faulted @ 0x0_1ee04000. Fault is of type FAULT_PDE ACCESS_TYPE_VIRT_READ [22431.245396] NVRM: GPU0 nvCheckOkFailedNoLog: Check failed: Out of memory [NV_ERR_NO_MEMORY] (0x00000051) returned from _memdescAllocInternal(pMemDesc) @ mem_desc.c:1336 [22585.433368] NVRM: _kgspProcessRpcEvent: Attempted to process RPC event from GPU0: 0x101a (PFM_REQ_HNDLR_STATE_SYNC_CALLBACK) during bootup without API lock [22611.091146] NVRM: nvCheckOkFailedNoLog: Check failed: Out of memory [NV_ERR_NO_MEMORY] (0x00000051) returned from _memdescAllocInternal(pMemDesc) @ mem_desc.c:1336 Note: If I use my laptop’s hardware MUX switch to force discrete-only graphics, the issue completely disappears, confirming it’s tied to hybrid mode frame copying under Wayland. Issue 2: Internal Microphone Broken (Kernel 7.0.1) Alongside the GPU issue, the update to linux-cachyos 7.0.1-1 completely broke my internal microphone. It seems to be a regression in the snd-hda-intel ALSA drivers for the Realtek audio codec used in these Lenovo Legion chassis. WirePlumber lost track of the hardware node entirely, leaving the mic unrecognized/unusable. Workaround I rolled back to my pre-update Btrfs snapshot, which downgraded the kernel, Wayland packages, and the NVIDIA drivers. Both the audio and the Vulkan memory leak issues are completely fixed on the older snapshot. I’ve added the kernel and nvidia packages to my IgnorePkg list in pacman.conf for now. Hopefully, this helps anyone else debugging Xid 31 crashes or broken audio on the newest packages!