我在 ASRock X399 Taichi 上有两个 M.2 NVMe 驱动器（Intel 660P 2TB）。其中一个上的操作系统是 Ubuntu 20.04。当我添加第三个 M.2 NVMe（Seagate FireCuda 2TB）时，操作系统加载未完成，黑屏仍然存在。
正如我所检查的syslog，问题是gdm-x-session已经关闭。Xorg.0.log清楚地显示
[ 12.345] (EE) No devices detected.
[ 12.345] (II) Applying OutputClass "nvidia" to /dev/dri/card0
[ 12.345] loading driver: nvidia
[ 12.575] (==) Matched nvidia as autoconfigured driver 0
[ 12.575] (==) Matched nouveau as autoconfigured driver 1
[ 12.575] (==) Matched modesetting as autoconfigured driver 2
[ 12.575] (==) Matched fbdev as autoconfigured driver 3
[ 12.575] (==) Matched vesa as autoconfigured driver 4
[ 12.575] (==) Assigned the driver to the xf86ConfigLayout
[ 12.575] (II) LoadModule: "nvidia"
[ 12.575] (II) Loading /usr/lib/x86_64-linux-gnu/nvidia/xorg/nvidia_drv.so
[ 12.575] (II) Module nvidia: vendor="NVIDIA Corporation"
[ 12.575] compiled for 1.6.99.901, module version = 1.0.0
[ 12.575] Module class: X.Org Video Driver
[ 12.575] (II) UnloadModule: "nvidia"
[ 12.575] (II) Unloading nvidia
[ 12.575] (II) Failed to load module "nvidia" (already loaded, 0)
[ 12.576] (II) LoadModule: "nouveau"
[ 12.576] (II) Loading /usr/lib/xorg/modules/drivers/nouveau_drv.so
[ 12.577] (II) Module nouveau: vendor="X.Org Foundation"
[ 12.577] compiled for 1.20.3, module version = 1.0.16
[ 12.577] Module class: X.Org Video Driver
[ 12.577] ABI class: X.Org Video Driver, version 24.0
[ 12.577] (II) LoadModule: "modesetting"
[ 12.578] (II) Loading /usr/lib/xorg/modules/drivers/modesetting_drv.so
[ 12.578] (II) Module modesetting: vendor="X.Org Foundation"
[ 12.578] compiled for 1.20.9, module version = 1.20.9
[ 12.578] Module class: X.Org Video Driver
[ 12.578] ABI class: X.Org Video Driver, version 24.1
[ 12.578] (II) LoadModule: "fbdev"
[ 12.579] (II) Loading /usr/lib/xorg/modules/drivers/fbdev_drv.so
[ 12.579] (II) Module fbdev: vendor="X.Org Foundation"
[ 12.579] compiled for 1.20.1, module version = 0.5.0
[ 12.579] Module class: X.Org Video Driver
[ 12.579] ABI class: X.Org Video Driver, version 24.0
[ 12.579] (II) LoadModule: "vesa"
[ 12.579] (II) Loading /usr/lib/xorg/modules/drivers/vesa_drv.so
[ 12.580] (II) Module vesa: vendor="X.Org Foundation"
[ 12.580] compiled for 1.20.4, module version = 2.4.0
[ 12.580] Module class: X.Org Video Driver
[ 12.580] ABI class: X.Org Video Driver, version 24.0
[ 12.580] (II) NVIDIA dlloader X Driver 450.119.03 Mon Mar 29 17:58:11 UTC 2021
[ 12.580] (II) NVIDIA Unified Driver for all Supported NVIDIA GPUs
[ 12.580] (II) NOUVEAU driver Date: Mon Jan 28 23:25:58 2019 -0500
[ 12.580] (II) NOUVEAU driver for NVIDIA chipset families :
[ 12.580] RIVA TNT (NV04)
[ 12.580] RIVA TNT2 (NV05)
[ 12.580] GeForce 256 (NV10)
[ 12.580] GeForce 2 (NV11, NV15)
[ 12.580] GeForce 4MX (NV17, NV18)
[ 12.580] GeForce 3 (NV20)
[ 12.580] GeForce 4Ti (NV25, NV28)
[ 12.580] GeForce FX (NV3x)
[ 12.580] GeForce 6 (NV4x)
[ 12.580] GeForce 7 (G7x)
[ 12.581] GeForce 8 (G8x)
[ 12.581] GeForce 9 (G9x)
[ 12.581] GeForce GTX 2xx/3xx (GT2xx)
[ 12.581] GeForce GTX 4xx/5xx (GFxxx)
[ 12.581] GeForce GTX 6xx/7xx (GKxxx)
[ 12.581] GeForce GTX 9xx (GMxxx)
[ 12.581] GeForce GTX 10xx (GPxxx)
[ 12.581] (II) modesetting: Driver for Modesetting Kernel Drivers: kms
[ 12.581] (II) FBDEV: driver for framebuffer: fbdev
[ 12.581] (II) VESA: driver for VESA chipsets: vesa
[ 12.581] (WW) Falling back to old probe method for modesetting
[ 12.581] (WW) Falling back to old probe method for fbdev
[ 12.581] (WW) Falling back to old probe method for modesetting
[ 12.581] (WW) Falling back to old probe method for fbdev
[ 12.581] (EE) No devices detected.
[ 12.581] (EE)
Fatal server error:
[ 12.581] (EE) no screens found(EE)
[ 12.581] (EE)
Please consult the The X.Org Foundation support
at http://wiki.x.org
for help.
[ 12.581] (EE) Please also check the log file at "/var/log/Xorg.0.log" for additional information.
[ 12.581] (EE)
[ 12.586] (EE) Server terminated with error (1). Closing log file.
如您所见，xorg没有检测到图形输出设备。
如果我移除第三个 M.2 驱动器，一切都正常。
如果我从 Live USB 启动，一切都正常，并且所有三个 M.2 驱动器都被识别。
注意显示器已连接，但未xorg检测到。启动卡住时我可以看到闪烁的下划线。
为什么第三个 M.2 驱动器的存在会导致xorg无法识别图形设备？
答案1
我发布我的解决方案，因为它可能会对其他人有所帮助。如果您有描述性的答案，请添加。
问题在于 NVMe 驱动器和 PCIe 1 之间的总线通道存在冲突。
解决方案是将 GPU 放在 PCIe 插槽 #3 而不是 PCIe 插槽 #1（按照主板手册的指示）。
我不知道
- 如果这是正确的解决方案。
- 如果这是一种常见/预期的行为。
- 如果问题与使用 3 个 NVMe 驱动器或 Seagate FireCuda 有关。