Re: Need to support mixed memory mappings with HMM

2021-03-26 Thread Christian König
Am 26.03.21 um 15:49 schrieb Felix Kuehling: Am 2021-03-26 um 4:50 a.m. schrieb Christian König: Am 25.03.21 um 17:28 schrieb Felix Kuehling: Am 2021-03-25 um 12:22 p.m. schrieb Christian König: My idea is to change the amdgpu_vm_update_mapping interface to use some high-bit in the pages_addr

Re: Need to support mixed memory mappings with HMM

2021-03-26 Thread Felix Kuehling
Am 2021-03-26 um 4:50 a.m. schrieb Christian König: > > > Am 25.03.21 um 17:28 schrieb Felix Kuehling: >> Am 2021-03-25 um 12:22 p.m. schrieb Christian König: My idea is to change the amdgpu_vm_update_mapping interface to use some high-bit in the pages_addr array to indic

Re: Need to support mixed memory mappings with HMM

2021-03-26 Thread Christian König
Am 25.03.21 um 17:28 schrieb Felix Kuehling: Am 2021-03-25 um 12:22 p.m. schrieb Christian König: My idea is to change the amdgpu_vm_update_mapping interface to use some high-bit in the pages_addr array to indicate the page type. For the default page type (0) nothing really changes for the cal

Re: Need to support mixed memory mappings with HMM

2021-03-25 Thread Felix Kuehling
Am 2021-03-25 um 12:22 p.m. schrieb Christian König: >> My idea is to change the amdgpu_vm_update_mapping interface to use >> some >> high-bit in the pages_addr array to indicate the page type. For the >> default page type (0) nothing really changes for the callers. The >> "fla

Re: Need to support mixed memory mappings with HMM

2021-03-25 Thread Christian König
Am 25.03.21 um 17:20 schrieb Felix Kuehling: Am 2021-03-25 um 12:16 p.m. schrieb Christian König: Am 25.03.21 um 17:14 schrieb Felix Kuehling: Am 2021-03-25 um 12:10 p.m. schrieb Christian König: Am 25.03.21 um 17:03 schrieb Felix Kuehling: Hi, This is a long one with a proposal for a pretty

Re: Need to support mixed memory mappings with HMM

2021-03-25 Thread Felix Kuehling
Am 2021-03-25 um 12:16 p.m. schrieb Christian König: > Am 25.03.21 um 17:14 schrieb Felix Kuehling: >> Am 2021-03-25 um 12:10 p.m. schrieb Christian König: >>> >>> Am 25.03.21 um 17:03 schrieb Felix Kuehling: Hi, This is a long one with a proposal for a pretty significant redesi

Re: Need to support mixed memory mappings with HMM

2021-03-25 Thread Christian König
Am 25.03.21 um 17:14 schrieb Felix Kuehling: Am 2021-03-25 um 12:10 p.m. schrieb Christian König: Am 25.03.21 um 17:03 schrieb Felix Kuehling: Hi, This is a long one with a proposal for a pretty significant redesign of how we handle migrations and VRAM management with HMM. This is based on my

Re: Need to support mixed memory mappings with HMM

2021-03-25 Thread Felix Kuehling
Am 2021-03-25 um 12:10 p.m. schrieb Christian König: > > > Am 25.03.21 um 17:03 schrieb Felix Kuehling: >> Hi, >> >> This is a long one with a proposal for a pretty significant redesign of >> how we handle migrations and VRAM management with HMM. This is based on >> my own debugging and reading of

Re: Need to support mixed memory mappings with HMM

2021-03-25 Thread Christian König
Am 25.03.21 um 17:03 schrieb Felix Kuehling: Hi, This is a long one with a proposal for a pretty significant redesign of how we handle migrations and VRAM management with HMM. This is based on my own debugging and reading of the migrate_vma helpers, as well as Alex's problems with migrations o

Need to support mixed memory mappings with HMM

2021-03-25 Thread Felix Kuehling
Hi, This is a long one with a proposal for a pretty significant redesign of how we handle migrations and VRAM management with HMM. This is based on my own debugging and reading of the migrate_vma helpers, as well as Alex's problems with migrations on A+A. I hope we can discuss this next Monday aft