CVE-2021-47421

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> drm/amdgpu: handle the case of pci_channel_io_frozen only in amdgpu_pci_resume<br /> <br /> In current code, when a PCI error state pci_channel_io_normal is detectd,<br /> it will report PCI_ERS_RESULT_CAN_RECOVER status to PCI driver, and PCI<br /> driver will continue the execution of PCI resume callback report_resume by<br /> pci_walk_bridge, and the callback will go into amdgpu_pci_resume<br /> finally, where write lock is releasd unconditionally without acquiring<br /> such lock first. In this case, a deadlock will happen when other threads<br /> start to acquire the read lock.<br /> <br /> To fix this, add a member in amdgpu_device strucutre to cache<br /> pci_channel_state, and only continue the execution in amdgpu_pci_resume<br /> when it&amp;#39;s pci_channel_io_frozen.