**Body:**
Hello Nvidia Linux team and community,
I am reporting a severe hard system lockup occurring under GPU load with the new Nvidia 595.58.03 driver on the `7.0.0-1-cachyos` kernel. The issue results in a complete loss of monitor signal, and the system logs indicate that the **GPU is dropping off the PCIe bus**. I have generated and attached the `nvidia-bug-report.log.gz` file for your detailed analysis.
**System Specifications:**
* **OS:** CachyOS x86_64
* **Kernel:** Linux 7.0.0-1-cachyos
* **Display Server / WM:** Wayland / KWin
* **DE:** KDE Plasma (Breeze Theme)
* **CPU:** AMD Ryzen 7 7800X3D (8C/16T @ 4.20 GHz)
* **GPU 1 (dGPU):** Gigabyte NVIDIA GeForce RTX 5070
* **GPU 2 (iGPU):** AMD Radeon Graphics (RADV RAPHAEL_MENDOCINO)
* **Monitor:** 3840x2160 @ 160Hz (External, Wayland scaling set to 2x)
* **RAM:** Patriot 32GB (2x16) 6000MHz CL28 DDR5
* **Motherboard:** MSI B650M Gaming Plus WiFi
**Steps to Reproduce & Description of the Issue:**
Whenever the dGPU (RTX 5070) is put under heavy load, the monitor suddenly loses signal completely (it does not output a black screen; it displays “No Signal”). The system becomes entirely unresponsive:
1. I cannot switch to a TTY.
2. The Magic SysRq keys (REISUB) are completely ignored.
3. The system requires a hard reset via the physical power button.
**Key Log Findings (The “GPU Lost” Error):**
Checking the `journalctl` logs after a forced reboot reveals massive NVRM assertions, specifically citing `NV_ERR_GPU_IS_LOST`. It is also intertwined with AMD’s SRSO mitigations (`srso_alias_return_thunk`) and kernel panic symptoms (`Unable to access opcode bytes`).
**Relevant Journalctl Snippet:**
```text
Apr 22 22:28:27 msi-r7-5070 kernel: ? srso_alias_return_thunk+0x5/0xfbef5
Apr 22 22:28:27 msi-r7-5070 kernel: RIP: 0033:0x7fe49c6b2422
Apr 22 22:28:27 msi-r7-5070 kernel: Code: Unable to access opcode bytes at 0x7fe49c6b23f8.
…
Apr 22 22:28:27 msi-r7-5070 kernel: NVRM: GPU0 nvAssertFailedNoLog: Assertion failed: (status == NV_OK) || (status == NV_ERR_GPU_IN_FULLCH>
Apr 22 22:28:27 msi-r7-5070 kernel: NVRM: GPU0 nvCheckOkFailedNoLog: Check failed: GPU lost from the bus [NV_ERR_GPU_IS_LOST] (0x0000000F)>
Apr 22 22:28:27 msi-r7-5070 dbus-broker-launch[797]: Activation request for ‘org.freedesktop.nm_dispatcher’ failed.
```
**Attachments:**
* `nvidia-bug-report.log.gz`
Given the hardware combination (Ryzen 7000 series with iGPU, RTX 5000 series dGPU, Gen4 NVMe, and PCIe Gen4/5 motherboard slots), I suspect this could be related to ASPM (Active State Power Management) conflicts, GSP firmware issues under load, or a severe Wayland synchronization crash that knocks the card off the bus.
Thank you for your time and support. Please advise on any specific kernel parameters (e.g., disabling GSP or tweaking PCIe power management) I should test.
**Body:**
Hello Nvidia Linux team and community,
I am reporting a severe hard system lockup occurring under GPU load with the new Nvidia 595.58.03 driver on the `7.0.0-1-cachyos` kernel. The issue results in a complete loss of monitor signal, and the system logs indicate that the **GPU is dropping off the PCIe bus**. I have generated and attached the `nvidia-bug-report.log.gz` file for your detailed analysis.
**System Specifications:**
* **OS:** CachyOS x86_64
* **Kernel:** Linux 7.0.0-1-cachyos
* **Display Server / WM:** Wayland / KWin
* **DE:** KDE Plasma (Breeze Theme)
* **CPU:** AMD Ryzen 7 7800X3D (8C/16T @ 4.20 GHz)
* **GPU 1 (dGPU):** Gigabyte NVIDIA GeForce RTX 5070
* **GPU 2 (iGPU):** AMD Radeon Graphics (RADV RAPHAEL_MENDOCINO)
* **Monitor:** 3840x2160 @ 160Hz (External, Wayland scaling set to 2x)
* **RAM:** Patriot 32GB (2x16) 6000MHz CL28 DDR5
* **Motherboard:** MSI B650M Gaming Plus WiFi
**Steps to Reproduce & Description of the Issue:**
Whenever the dGPU (RTX 5070) is put under heavy load, the monitor suddenly loses signal completely (it does not output a black screen; it displays “No Signal”). The system becomes entirely unresponsive:
1. I cannot switch to a TTY.
2. The Magic SysRq keys (REISUB) are completely ignored.
3. The system requires a hard reset via the physical power button.
**Key Log Findings (The “GPU Lost” Error):**
Checking the `journalctl` logs after a forced reboot reveals massive NVRM assertions, specifically citing `NV_ERR_GPU_IS_LOST`. It is also intertwined with AMD’s SRSO mitigations (`srso_alias_return_thunk`) and kernel panic symptoms (`Unable to access opcode bytes`).
**Relevant Journalctl Snippet:**
```text
Apr 22 22:28:27 msi-r7-5070 kernel: ? srso_alias_return_thunk+0x5/0xfbef5
Apr 22 22:28:27 msi-r7-5070 kernel: RIP: 0033:0x7fe49c6b2422
Apr 22 22:28:27 msi-r7-5070 kernel: Code: Unable to access opcode bytes at 0x7fe49c6b23f8.
…
Apr 22 22:28:27 msi-r7-5070 kernel: NVRM: GPU0 nvAssertFailedNoLog: Assertion failed: (status == NV_OK) || (status == NV_ERR_GPU_IN_FULLCH>
Apr 22 22:28:27 msi-r7-5070 kernel: NVRM: GPU0 nvCheckOkFailedNoLog: Check failed: GPU lost from the bus [NV_ERR_GPU_IS_LOST] (0x0000000F)>
Apr 22 22:28:27 msi-r7-5070 dbus-broker-launch[797]: Activation request for ‘org.freedesktop.nm_dispatcher’ failed.
```
**Attachments:**
* `nvidia-bug-report.log.gz`
Given the hardware combination (Ryzen 7000 series with iGPU, RTX 5000 series dGPU, Gen4 NVMe, and PCIe Gen4/5 motherboard slots), I suspect this could be related to ASPM (Active State Power Management) conflicts, GSP firmware issues under load, or a severe Wayland synchronization crash that knocks the card off the bus.
Thank you for your time and support. Please advise on any specific kernel parameters (e.g., disabling GSP or tweaking PCIe power management) I should test.
**Body:**
Hello Nvidia Linux team and community,
I am reporting a severe hard system lockup occurring under GPU load with the new Nvidia 595.58.03 driver on the `7.0.0-1-cachyos` kernel. The issue results in a complete loss of monitor signal, and the system logs indicate that the **GPU is dropping off the PCIe bus**. I have generated and attached the `nvidia-bug-report.log.gz` file for your detailed analysis.
**System Specifications:**
* **OS:** CachyOS x86_64
* **Kernel:** Linux 7.0.0-1-cachyos
* **Display Server / WM:** Wayland / KWin
* **DE:** KDE Plasma (Breeze Theme)
* **CPU:** AMD Ryzen 7 7800X3D (8C/16T @ 4.20 GHz)
* **GPU 1 (dGPU):** Gigabyte NVIDIA GeForce RTX 5070
* **GPU 2 (iGPU):** AMD Radeon Graphics (RADV RAPHAEL_MENDOCINO)
* **Monitor:** 3840x2160 @ 160Hz (External, Wayland scaling set to 2x)
* **RAM:** Patriot 32GB (2x16) 6000MHz CL28 DDR5
* **Motherboard:** MSI B650M Gaming Plus WiFi
**Steps to Reproduce & Description of the Issue:**
Whenever the dGPU (RTX 5070) is put under heavy load, the monitor suddenly loses signal completely (it does not output a black screen; it displays “No Signal”). The system becomes entirely unresponsive:
1. I cannot switch to a TTY.
2. The Magic SysRq keys (REISUB) are completely ignored.
3. The system requires a hard reset via the physical power button.
**Key Log Findings (The “GPU Lost” Error):**
Checking the `journalctl` logs after a forced reboot reveals massive NVRM assertions, specifically citing `NV_ERR_GPU_IS_LOST`. It is also intertwined with AMD’s SRSO mitigations (`srso_alias_return_thunk`) and kernel panic symptoms (`Unable to access opcode bytes`).
**Relevant Journalctl Snippet:**
```text
Apr 22 22:28:27 msi-r7-5070 kernel: ? srso_alias_return_thunk+0x5/0xfbef5
Apr 22 22:28:27 msi-r7-5070 kernel: RIP: 0033:0x7fe49c6b2422
Apr 22 22:28:27 msi-r7-5070 kernel: Code: Unable to access opcode bytes at 0x7fe49c6b23f8.
…
Apr 22 22:28:27 msi-r7-5070 kernel: NVRM: GPU0 nvAssertFailedNoLog: Assertion failed: (status == NV_OK) || (status == NV_ERR_GPU_IN_FULLCH>
Apr 22 22:28:27 msi-r7-5070 kernel: NVRM: GPU0 nvCheckOkFailedNoLog: Check failed: GPU lost from the bus [NV_ERR_GPU_IS_LOST] (0x0000000F)>
Apr 22 22:28:27 msi-r7-5070 dbus-broker-launch[797]: Activation request for ‘org.freedesktop.nm_dispatcher’ failed.
```
**Attachments:**
nvidia-bug-report.log.gz (402.6 KB)