如何在2张C500部署 MiniMax-H3 量化8模型
如何在2张C500部署 MiniMax-H3 量化8模型
更新了3.9驱动和maca ,还是一样的问题
mx-smi 回显
dmesg 信息
[Wed Sep 23 05:36:51 2026] METAX.BASE.INFO MetaX driver exit
[Wed Sep 23 05:36:51 2026] METAX.PCI.INFO metax_reset exit
[Wed Sep 23 05:36:51 2026] METAX.BASE.INFO unload by process rmmod
[Wed Sep 23 05:36:51 2026] METAX.B6300.D0.BASE.INFO finishing device
[Wed Sep 23 05:36:51 2026] METAX.B6300.D0.BASE.INFO native device fini ok
[Wed Sep 23 05:36:51 2026] MXCD.TOPO.INFO Update layout SYSFS entries
[Wed Sep 23 05:36:51 2026] METAX.B6300.D0.MC.INFO ttm finalized
[Wed Sep 23 05:36:51 2026] METAX.B6300.D0.BASE.INFO native device fini ok
[Wed Sep 23 05:36:51 2026] MXCD.TOPO.INFO Destroy sysfs manager folder
[Wed Sep 23 05:36:55 2026] nvme 0000:15:00.0: found device 0000:15:00.0 MPS:512B exceeds 256B!
[Wed Sep 23 05:36:55 2026] mlx5_core 0000:2f:00.0: found device 0000:2f:00.0 MPS:512B exceeds 256B!
[Wed Sep 23 05:36:55 2026] nvme 0000:65:00.0: set MRRS from 4096B to 256B
[Wed Sep 23 05:36:55 2026] METAX.PCI.INFO found 1 metax pcie devices.
[Wed Sep 23 05:36:55 2026] MXCD.TOPO.INFO Update layout SYSFS entries
[Wed Sep 23 05:36:55 2026] MXCD.TOPO.INFO layout: add 2 CPU node
[Wed Sep 23 05:36:55 2026] MXCD.TOPO.INFO finish initializing layout
[Wed Sep 23 05:36:55 2026] MXCD.PEER.INFO Try to initialize peer memory client support
[Wed Sep 23 05:36:55 2026] failing symbol_get of non-GPLONLY symbol ib_register_peer_memory_client.
[Wed Sep 23 05:36:55 2026] failing symbol_get of non-GPLONLY symbol ib_register_peer_memory_client.
[Wed Sep 23 05:36:55 2026] failing symbol_get of non-GPLONLY symbol ib_unregister_peer_memory_client.
[Wed Sep 23 05:36:55 2026] failing symbol_get of non-GPLONLY symbol ib_unregister_peer_memory_client.
[Wed Sep 23 05:36:55 2026] MXCD.PEER.INFO Peer memory client interface was not detected
[Wed Sep 23 05:36:55 2026] MXCD.TOPO.INFO Create sysfs manager folder
[Wed Sep 23 05:36:55 2026] METAX.BASE.INFO load by process kworker/72:9
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.BASE.INFO initializing device (0x9999:0x4001 0x9999:0x4001 0x01) cap 0x0000000c000006cf
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.PCI.INFO pci bar5 mmio base 0x9a200000 size 0x100000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.PCI.INFO register mmio CPU va 0xffa0000033500000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.VIRT.INFO virt caps 0x0
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.INT.INFO MSI vectors 1 MSI-X vectors 3
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.INT.INFO using MSI
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.XCORE.INFO xcore_version 0
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.PCI.INFO MX_IOV pos e90
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.BASE.INFO PCIe atomic ops is supported
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_mc]
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_atu]
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_int]
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_sdma]
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_ccx]
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_smp]
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_xcore]
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.SMP.INFO wait SMP0 change to be idle start 76901 wait 30 secs
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.SMP.INFO SMP0 idle now
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.SMP.INFO wait SMP1 change to be idle start 75378 wait 30 secs
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.SMP.INFO SMP1 idle now
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MGPU.INFO dnoc 0x100 topo_id 0 socket_id 0 die_id 0 die_num 1
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MGPU.INFO total die 1 available die 1 int hop base 0x0
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MGPU.INFO mgpu topo id 0 socket id 0 die_id 0
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MGPU.INFO topo_sw_id 0 sockets 1 sriov[0]
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.BASE.INFO available die num 1 activated die num 1
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.SOC.INFO vbios version 1.33.5.0, driver version 3.10.14
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MON.INFO the active status of this card is changed, new status:1
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.RAS.INFO XSC Proxy Supported
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.PCI.INFO pci bar2 doorbell base 0x1f4000000000 size 0x800000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.PCI.INFO doorbell ptr 0xffa0000030482000 doorbell num 1024
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO XVM PT block size is fixed at 9-bit, page size is 2048 KB
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO get vram size 0x1000000000 from fuse
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO mc vram size 0x1000000000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO pcie bar0 vram original base 0x1f3000000000 size 0x1000000000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO pcie bar0 vram final base 0x1f3000000000 size 0x1000000000 cpu visible 0x1000000000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO unmapped one 0x0-0xfffffffff, off 0x0 unmapped two 0x0-0x0
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.BASE.INFO detected vram size 65536M, bar size 65536M
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.BASE.INFO vram width 4096bits HBM
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO buffer object driver init succeeded, dma32 0
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO unmapped_aperture_kaddr[0] 0xffa0000040000000-0xffa000103fffffff
[Wed Sep 23 05:36:55 2026] METAX.MC.INFO 481946M of (XTT + Userptr) memory available
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO 65536M of vram memory ready
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO 0M of MXME memory ready
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.RAS.INFO sync bad pages from SMP
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO reserved memory index 5: offset 0x0
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO reserved memory index 6: offset 0x400000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO reserved memory index 7: offset 0xff3800000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO reserved memory index 8: offset 0x1400000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO reserved memory index 9: offset 0x1800000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO reserved memory index 10: offset 0x1c00000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO reserved memory index 11: offset 0x2000000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO reserved memory index 12: offset 0x2400000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO reserved memory index 13: offset 0x7400000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO reserved memory index 14: offset 0xc400000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO reserved memory index 15: offset 0xc800000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO reserved xtt share memory index 21: cpu_va 0xffa0000030f19000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_mc> init done!
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.ATU.INFO system memory size is 0x9ce235c000, GPA starts from 0x9d00000000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.ATU.INFO ATUL2_CTRL 0x10004 APTURE_0 BASE 0x0 TOP 0xfffffffff OFFSET 9d00000000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.ATU.INFO ATUL2_CTRL 0x10004 APTURE_0 BASE 0x0 TOP 0xfffffffff OFFSET 9d00000000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.HAG.INFO hag ATUL1 CTRL1 0x400080
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.HAG.INFO hag init done, ATUL1 CTRL3 0x82
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.HAG.INFO hag init done, ATUL1 XPID 0x0
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.HAG.INFO hag init done, ATUL1 SPACE 0x4
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO metax_kmem_init: pd_base 0x200000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.ATU.INFO KMEM PAGE BASE ENABLE (table at 0x200000) pd_pa 0x200000.
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_atu> init done!
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.INT.INFO ring-wptr gpu va:0x4000200000 cpu va:0xff1106305b1af000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.INT.INFO ringbuf gpu va 0x4000400000 cpu va 0xffa0000030f5a000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.INT.INFO ringbuf gpu pa 0x63060180000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.INT.INFO ring-wptr addr gpu va 0x4000200000 gpu pa 0x6305b1af000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_int> init done!
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_sdma> init done!
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.FW.INFO load /lib/firmware/metax/mxc500/vpud0_fw.bin firmware
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.FW.INFO load /lib/firmware/metax/mxc500/vpue_fw.bin firmware
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.FW.INFO load /lib/firmware/metax/mxc500/vpud1_fw.bin firmware
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_ccx> init done!
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.SMP.INFO SMP0, log_pa:c400000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.SMP.INFO SMP1, log_pa:c800000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_smp> init done!
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.SMP.INFO wait XSC change to be idle start 328348 wait 10 secs
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.SMP.INFO XSC idle now
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_xcore> init done!
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.RAS.INFO RAS CE Monitor is on
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.SMI.INFO smi init ok
[Wed Sep 23 05:36:55 2026] MXCD.BASE.INFO allocated 1572864 bytes for tba memory in xtt.
[Wed Sep 23 05:36:55 2026] MXCD.B6300.D0.DOORBELL.INFO doorbell initialization:
[Wed Sep 23 05:36:55 2026] MXCD.B6300.D0.DOORBELL.INFO doorbell base = 0x1f4000000000
[Wed Sep 23 05:36:55 2026] MXCD.B6300.D0.DOORBELL.INFO doorbell_id_offset = 0x00000000
[Wed Sep 23 05:36:55 2026] MXCD.B6300.D0.DOORBELL.INFO db_process_ida_end = 0x00000400
[Wed Sep 23 05:36:55 2026] MXCD.B6300.D0.DOORBELL.INFO doorbell vmrange size = 0x00800000
[Wed Sep 23 05:36:55 2026] METAX.B6300.D0.MC.INFO aper: base 1f3000000000 limit 1f4000000000 pub size 1000000000 pri size 0 addr mask ffff000000000000
[Wed Sep 23 05:36:55 2026] MXCD.TOPO.INFO Adding GPU node 2(ID: 58793) to layout
[Wed Sep 23 05:36:55 2026] MXCD.B6300.D0.TOPO.INFO Added [157] GPU cache entries
[Wed Sep 23 05:36:55 2026] MXCD.TOPO.INFO Update layout SYSFS entries
[Wed Sep 23 05:36:56 2026] MXCD.B6300.D0.RINGBUF.INFO alloc private memory size 1c000000
[Wed Sep 23 05:36:56 2026] MXCD.B6300.D0.RINGBUF.INFO alloc GPU reserved memory, size=0x1000000
[Wed Sep 23 05:36:56 2026] METAX.B6300.D0.MC.INFO reserved utc memory index 23: offset 0xff0000000
[Wed Sep 23 05:36:56 2026] MXCD.B6300.D0.RINGBUF.INFO Shader UTC Time: 0x18d7baa65eb2ce25 [2026-09-22_19:16:40]
[Wed Sep 23 05:36:56 2026] METAX.XCORE.INFO set xcore page size to 2MB.
[Wed Sep 23 05:36:56 2026] MXCD.BASE.INFO added device 9999:4001
[Wed Sep 23 05:36:56 2026] workqueue: work_for_cpu_fn hogged CPU for >10000us 11 times, consider switching to WQ_UNBOUND
[Wed Sep 23 05:36:56 2026] METAX.BASE.INFO metax pro feature device page support 1
[Wed Sep 23 05:36:56 2026] METAX.B6300.D0.BASE.INFO pcba sn: AENA2510000498, gpuid 58793, socket_id 0, topo_id 0
[Wed Sep 23 05:36:56 2026] METAX.B6300.D0.BASE.INFO native device init ok
[Wed Sep 23 05:36:56 2026] METAX.B6300.D0.BASE.INFO device pm enabled.
[Wed Sep 23 05:36:56 2026] [drm] Initialized METAX 3.10.14 for 0000:63:00.0 on minor 1
[Wed Sep 23 05:36:56 2026] METAX.BASE.INFO MetaX driver startup! built by jenkins_sw_bot, git commit d92bb957
[Wed Sep 23 05:36:56 2026] METAX.B6300.D0.BASE.INFO systemd-logind Open kms should open mxcd first
[Wed Sep 23 05:37:21 2026] EXT4-fs (nvme1n1): Supports (experimental) DIO atomic writes awu_min: 4096, awu_max: 4096
[Wed Sep 23 05:37:21 2026] EXT4-fs (nvme1n1): mounted filesystem 1eb3de61-a114-4ad7-86b6-4670845bc993 r/w with ordered data mode. Quota mode: none.
[Wed Sep 23 05:39:07 2026] Oops: general protection fault, probably for non-canonical address 0x7473616c00736d77: 0000 [#1] PREEMPT SMP NOPTI
gpu 直接 storage 测试,fio没有输出”的dmesg日志如下
[Wed Sep 23 03:43:09 2026] METAX.B6300.D0.MC.INFO ttm finalized
[Wed Sep 23 03:43:09 2026] METAX.B6300.D0.BASE.INFO native device fini ok
[Wed Sep 23 03:43:09 2026] MXCD.TOPO.INFO Destroy sysfs manager folder
[Wed Sep 23 03:43:13 2026] nvme 0000:15:00.0: found device 0000:15:00.0 MPS:512B exceeds 256B!
[Wed Sep 23 03:43:13 2026] mlx5_core 0000:2f:00.0: found device 0000:2f:00.0 MPS:512B exceeds 256B!
[Wed Sep 23 03:43:13 2026] nvme 0000:65:00.0: set MRRS from 4096B to 256B
[Wed Sep 23 03:43:13 2026] METAX.PCI.INFO found 1 metax pcie devices.
[Wed Sep 23 03:43:13 2026] MXCD.TOPO.INFO Update layout SYSFS entries
[Wed Sep 23 03:43:13 2026] MXCD.TOPO.INFO layout: add 2 CPU node
[Wed Sep 23 03:43:13 2026] MXCD.TOPO.INFO finish initializing layout
[Wed Sep 23 03:43:13 2026] MXCD.PEER.INFO Try to initialize peer memory client support
[Wed Sep 23 03:43:13 2026] failing symbol_get of non-GPLONLY symbol ib_register_peer_memory_client.
[Wed Sep 23 03:43:13 2026] failing symbol_get of non-GPLONLY symbol ib_register_peer_memory_client.
[Wed Sep 23 03:43:13 2026] failing symbol_get of non-GPLONLY symbol ib_unregister_peer_memory_client.
[Wed Sep 23 03:43:13 2026] failing symbol_get of non-GPLONLY symbol ib_unregister_peer_memory_client.
[Wed Sep 23 03:43:13 2026] MXCD.PEER.INFO Peer memory client interface was not detected
[Wed Sep 23 03:43:13 2026] MXCD.TOPO.INFO Create sysfs manager folder
[Wed Sep 23 03:43:13 2026] METAX.BASE.INFO load by process kworker/72:0
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.BASE.INFO initializing device (0x9999:0x4001 0x9999:0x4001 0x01) cap 0x0000000c000006cf
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.PCI.INFO pci bar5 mmio base 0x9a200000 size 0x100000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.PCI.INFO register mmio CPU va 0xffa0000035200000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.VIRT.INFO virt caps 0x0
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.INT.INFO MSI vectors 1 MSI-X vectors 3
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.INT.INFO using MSI
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.XCORE.INFO xcore_version 0
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.PCI.INFO MX_IOV pos e90
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.BASE.INFO PCIe atomic ops is supported
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_mc]
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_atu]
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_int]
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_sdma]
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_ccx]
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_smp]
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_xcore]
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.SMP.INFO wait SMP0 change to be idle start 14529 wait 30 secs
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.SMP.INFO SMP0 idle now
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.SMP.INFO wait SMP1 change to be idle start 14241 wait 30 secs
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.SMP.INFO SMP1 idle now
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MGPU.INFO dnoc 0x100 topo_id 0 socket_id 0 die_id 0 die_num 1
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MGPU.INFO total die 1 available die 1 int hop base 0x0
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MGPU.INFO mgpu topo id 0 socket id 0 die_id 0
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MGPU.INFO topo_sw_id 0 sockets 1 sriov[0]
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.BASE.INFO available die num 1 activated die num 1
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.SOC.INFO vbios version 1.33.5.0, driver version 3.8.30
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MON.INFO the active status of this card is changed, new status:1
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.RAS.INFO XSC Proxy Supported
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.PCI.INFO pci bar2 doorbell base 0x1f4000000000 size 0x800000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.PCI.INFO doorbell ptr 0xffa000001b40e000 doorbell num 1024
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO XVM PT block size is fixed at 9-bit, page size is 2048 KB
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO get vram size 0x1000000000 from fuse
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO mc vram size 0x1000000000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO pcie bar0 vram original base 0x1f3000000000 size 0x1000000000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO pcie bar0 vram final base 0x1f3000000000 size 0x1000000000 cpu visible 0x1000000000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO unmapped one 0x0-0xfffffffff, off 0x0 unmapped two 0x0-0x0
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.BASE.INFO detected vram size 65536M, bar size 65536M
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.BASE.INFO vram width 4096bits HBM
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO buffer object driver init succeeded, dma32 0
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO unmapped_aperture_kaddr[0] 0xffa0001080000000-0xffa000207fffffff
[Wed Sep 23 03:43:13 2026] METAX.MC.INFO 481946M of (XTT + Userptr) memory available
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO 65536M of vram memory ready
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO 0M of MXME memory ready
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.RAS.INFO sync bad pages from SMP
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO reserved memory index 5: offset 0x0
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO reserved memory index 6: offset 0x400000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO reserved memory index 7: offset 0xff3800000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO reserved memory index 8: offset 0x1400000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO reserved memory index 9: offset 0x1800000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO reserved memory index 10: offset 0x1c00000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO reserved memory index 11: offset 0x2000000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO reserved memory index 12: offset 0x2400000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO reserved memory index 13: offset 0x7400000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO reserved memory index 14: offset 0xc400000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO reserved memory index 15: offset 0xc800000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO reserved xtt share memory index 20: cpu_va 0xffa000001f981000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_mc> init done!
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.ATU.INFO system memory size is 0x9ce235c000, GPA starts from 0x9d00000000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.ATU.INFO ATUL2_CTRL 0x10004 APTURE_0 BASE 0x0 TOP 0xfffffffff OFFSET 9d00000000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.ATU.INFO ATUL2_CTRL 0x10004 APTURE_0 BASE 0x0 TOP 0xfffffffff OFFSET 9d00000000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.HAG.INFO hag ATUL1 CTRL1 0x400080
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.HAG.INFO hag init done, ATUL1 CTRL3 0x82
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.HAG.INFO hag init done, ATUL1 XPID 0x0
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.HAG.INFO hag init done, ATUL1 SPACE 0x4
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO metax_kmem_init: pd_base 0x200000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.ATU.INFO KMEM PAGE BASE ENABLE (table at 0x200000) pd_pa 0x200000.
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_atu> init done!
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.INT.INFO ring-wptr addr 0x4000200000 0xff1106303a79c000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.INT.INFO gpu va 0x4000400000 cpu va 0xffa0000030129000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.INT.INFO page index 0 pa 0x630eb700000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.INT.INFO wptr_addr 0x4000200000 pa 0x6303a79c000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_int> init done!
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_sdma> init done!
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.FW.INFO load /lib/firmware/metax/mxc500/vpud0_fw.bin firmware
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.FW.INFO load /lib/firmware/metax/mxc500/vpue_fw.bin firmware
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.FW.INFO load /lib/firmware/metax/mxc500/vpud1_fw.bin firmware
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_ccx> init done!
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.SMP.INFO SMP0, log_pa:c400000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.SMP.INFO SMP1, log_pa:c800000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_smp> init done!
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.SMP.INFO wait XSC change to be idle start 7288 wait 10 secs
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.SMP.INFO XSC idle now
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_xcore> init done!
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.RAS.INFO RAS CE Monitor is on
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.SMI.INFO smi init ok
[Wed Sep 23 03:43:13 2026] MXCD.BASE.INFO allocated 1572864 bytes for tba memory in xtt.
[Wed Sep 23 03:43:13 2026] MXCD.B6300.D0.DOORBELL.INFO doorbell initialization:
[Wed Sep 23 03:43:13 2026] MXCD.B6300.D0.DOORBELL.INFO doorbell base = 0x1f4000000000
[Wed Sep 23 03:43:13 2026] MXCD.B6300.D0.DOORBELL.INFO doorbell_id_offset = 0x00000000
[Wed Sep 23 03:43:13 2026] MXCD.B6300.D0.DOORBELL.INFO db_process_ida_end = 0x00000400
[Wed Sep 23 03:43:13 2026] MXCD.B6300.D0.DOORBELL.INFO doorbell vmrange size = 0x00800000
[Wed Sep 23 03:43:13 2026] METAX.B6300.D0.MC.INFO aper: base 1f3000000000 limit 1f4000000000 pub size 1000000000 pri size 0 addr mask ffff000000000000
[Wed Sep 23 03:43:13 2026] MXCD.TOPO.INFO Adding GPU node 2(ID: 58793) to layout
[Wed Sep 23 03:43:13 2026] MXCD.B6300.D0.TOPO.INFO Added [157] GPU cache entries
[Wed Sep 23 03:43:13 2026] MXCD.TOPO.INFO Update layout SYSFS entries
[Wed Sep 23 03:43:14 2026] MXCD.B6300.D0.RINGBUF.INFO alloc private memory size 1c000000
[Wed Sep 23 03:43:14 2026] MXCD.B6300.D0.RINGBUF.INFO alloc GPU reserved memory, size=0x1000000
[Wed Sep 23 03:43:14 2026] METAX.XCORE.INFO set xcore page size to 2MB.
[Wed Sep 23 03:43:14 2026] MXCD.BASE.INFO added device 9999:4001
[Wed Sep 23 03:43:14 2026] METAX.BASE.INFO metax pro feature device page support 1
[Wed Sep 23 03:43:14 2026] METAX.B6300.D0.BASE.INFO pcba sn: AENA2510000498, gpuid 58793, socket_id 0, topo_id 0
[Wed Sep 23 03:43:14 2026] METAX.B6300.D0.BASE.INFO native device init ok
[Wed Sep 23 03:43:14 2026] METAX.B6300.D0.BASE.INFO device pm enabled.
[Wed Sep 23 03:43:14 2026] [drm] Initialized METAX 3.8.30 for 0000:63:00.0 on minor 1
[Wed Sep 23 03:43:14 2026] METAX.BASE.INFO MetaX driver startup! built by jenkins_sw_bot, git commit ff4d0144
[Wed Sep 23 03:43:14 2026] METAX.B6300.D0.BASE.INFO systemd-logind Open kms should open mxcd first
[Wed Sep 23 03:45:31 2026] EXT4-fs (nvme1n1): Supports (experimental) DIO atomic writes awu_min: 4096, awu_max: 4096
[Wed Sep 23 03:45:31 2026] EXT4-fs (nvme1n1): recovery complete
[Wed Sep 23 03:45:31 2026] EXT4-fs (nvme1n1): mounted filesystem 1eb3de61-a114-4ad7-86b6-4670845bc993 r/w with ordered data mode. Quota mode: none.
[Wed Sep 23 03:46:54 2026] Oops: general protection fault, probably for non-canonical address 0x7473616c00736d77: 0000 [#1] PREEMPT SMP NOPTI
[Wed Sep 23 03:46:54 2026] CPU: 1 UID: 0 PID: 10504 Comm: fio Tainted: G W OE 6.14.0-27-generic #27~24.04.1-Ubuntu
[Wed Sep 23 03:46:54 2026] Tainted: [W]=WARN, [O]=OOT_MODULE, [E]=UNSIGNED_MODULE
[Wed Sep 23 03:46:54 2026] Hardware name: Intel Corporation AvenueCity/AvenueCity, BIOS BHSDCRB1.IPC.3544.P25.2411252204 11/25/2024
[Wed Sep 23 03:46:54 2026] RIP: 0010:xas_start+0x34/0x130
[Wed Sep 23 03:46:54 2026] Code: 48 83 ec 08 4c 8b 67 18 4c 89 e0 83 e0 03 0f 84 85 00 00 00 49 81 fc 05 c0 ff ff 76 06 48 83 f8 02 74 5f 48 8b 03 48 8b 73 08 <48> 8b 40 08 48 89 c2 83 e2 03 48 83 fa 02 75 08 48 3d 00 10 00 00
[Wed Sep 23 03:46:54 2026] RSP: 0018:ffa000003c91f4d8 EFLAGS: 00010213
[Wed Sep 23 03:46:54 2026] RAX: 7473616c00736d6f RBX: ffa000003c91f540 RCX: 0000000000000002
[Wed Sep 23 03:46:54 2026] RDX: 0000000000000000 RSI: 0000000000006500 RDI: ffa000003c91f540
[Wed Sep 23 03:46:54 2026] RBP: ffa000003c91f4f0 R08: 0000000000000100 R09: 0000000000000000
[Wed Sep 23 03:46:54 2026] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000003
[Wed Sep 23 03:46:54 2026] R13: ff1100011d8af0c8 R14: ff1100011d8af0c8 R15: ffffffff82d25b7d
[Wed Sep 23 03:46:54 2026] FS: 00007171dd102800(0000) GS:ff11005e7c880000(0000) knlGS:0000000000000000
[Wed Sep 23 03:46:54 2026] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[Wed Sep 23 03:46:54 2026] CR2: 00005c068cd21f88 CR3: 000000012684b004 CR4: 0000000000f73ef0
[Wed Sep 23 03:46:54 2026] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[Wed Sep 23 03:46:54 2026] DR3: 0000000000000000 DR6: 00000000fffe07f0 DR7: 0000000000000400
[Wed Sep 23 03:46:54 2026] PKRU: 55555554
[Wed Sep 23 03:46:54 2026] Call Trace:
[Wed Sep 23 03:46:54 2026] <TASK>
[Wed Sep 23 03:46:54 2026] ? allocate_slab+0x132/0x310
[Wed Sep 23 03:46:54 2026] xas_load+0x17/0x100
[Wed Sep 23 03:46:54 2026] xa_load+0x55/0xa0
[Wed Sep 23 03:46:54 2026] pci_p2pdma_map_segment+0x139/0x1a0
[Wed Sep 23 03:46:54 2026] dma_direct_map_sg+0xcb/0x140
[Wed Sep 23 03:46:54 2026] __dma_map_sg_attrs+0x66/0x1b0
[Wed Sep 23 03:46:54 2026] dma_map_sgtable+0x1d/0x40
[Wed Sep 23 03:46:54 2026] nvme_map_data+0xd6/0x510 [nvme]
[Wed Sep 23 03:46:54 2026] nvme_prep_rq.part.0+0x33/0x160 [nvme]
[Wed Sep 23 03:46:54 2026] nvme_queue_rqs+0x108/0x230 [nvme]
[Wed Sep 23 03:46:54 2026] blk_mq_flush_plug_list+0x1df/0x200
[Wed Sep 23 03:46:54 2026] blk_add_rq_to_plug+0xff/0x1c0
[Wed Sep 23 03:46:54 2026] blk_mq_submit_bio+0x619/0x820
[Wed Sep 23 03:46:54 2026] ? __es_insert_extent+0x26f/0x370
[Wed Sep 23 03:46:54 2026] __submit_bio+0x75/0x290
[Wed Sep 23 03:46:54 2026] submit_bio_noacct_nocheck+0x168/0x230
[Wed Sep 23 03:46:54 2026] submit_bio_noacct+0x16c/0x5c0
[Wed Sep 23 03:46:54 2026] submit_bio+0xb1/0x110
[Wed Sep 23 03:46:54 2026] iomap_dio_submit_bio+0x84/0xa0
[Wed Sep 23 03:46:54 2026] iomap_dio_bio_iter+0x3a0/0x570
[Wed Sep 23 03:46:54 2026] __iomap_dio_rw+0x3a4/0x710
[Wed Sep 23 03:46:54 2026] iomap_dio_rw+0x11/0x60
[Wed Sep 23 03:46:54 2026] ext4_file_read_iter+0x159/0x200
[Wed Sep 23 03:46:54 2026] do_iter_readv_writev+0x205/0x270
[Wed Sep 23 03:46:54 2026] vfs_iter_read+0x103/0x170
[Wed Sep 23 03:46:54 2026] metax_io_rw_sync+0x440/0x590 [metax]
[Wed Sep 23 03:46:54 2026] mxcd_ioctl_vram_io+0xf0/0x220 [metax]
[Wed Sep 23 03:46:54 2026] mxcd_ioctl+0x326/0x4d0 [metax]
[Wed Sep 23 03:46:54 2026] ? __pfx_mxcd_ioctl_vram_io+0x10/0x10 [metax]
[Wed Sep 23 03:46:54 2026] __x64_sys_ioctl+0xa4/0xe0
[Wed Sep 23 03:46:54 2026] x64_sys_call+0x131e/0x2650
[Wed Sep 23 03:46:54 2026] do_syscall_64+0x7e/0x170
[Wed Sep 23 03:46:54 2026] ? arch_exit_to_user_mode_prepare.isra.0+0x22/0xd0
[Wed Sep 23 03:46:54 2026] ? syscall_exit_to_user_mode+0x38/0x1d0
[Wed Sep 23 03:46:54 2026] ? do_syscall_64+0x8a/0x170
[Wed Sep 23 03:46:54 2026] ? exc_page_fault+0x96/0x1e0
[Wed Sep 23 03:46:54 2026] entry_SYSCALL_64_after_hwframe+0x76/0x7e
[Wed Sep 23 03:46:54 2026] RIP: 0033:0x7171dcf24e1d
[Wed Sep 23 03:46:54 2026] Code: 04 25 28 00 00 00 48 89 45 c8 31 c0 48 8d 45 10 c7 45 b0 10 00 00 00 48 89 45 b8 48 8d 45 d0 48 89 45 c0 b8 10 00 00 00 0f 05 <89> c2 3d 00 f0 ff ff 77 1a 48 8b 45 c8 64 48 2b 04 25 28 00 00 00
[Wed Sep 23 03:46:54 2026] RSP: 002b:00007ffd828b3240 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
[Wed Sep 23 03:46:54 2026] RAX: ffffffffffffffda RBX: 0000000000100000 RCX: 00007171dcf24e1d
[Wed Sep 23 03:46:54 2026] RDX: 00007ffd828b3360 RSI: 00000000c0404b1b RDI: 0000000000000007
[Wed Sep 23 03:46:54 2026] RBP: 00007ffd828b3290 R08: 0000000000000000 R09: 0000000000000007
[Wed Sep 23 03:46:54 2026] R10: 00007171dce18808 R11: 0000000000000246 R12: 00005c06aacdf4e0
[Wed Sep 23 03:46:54 2026] R13: 0000000000000000 R14: 0000000000000000 R15: 00005c06aace26c0
[Wed Sep 23 03:46:54 2026] </TASK>
[Wed Sep 23 03:46:54 2026] Modules linked in: metax(OE) snd_seq_dummy snd_hrtimer snd_seq_midi snd_seq_midi_event snd_rawmidi snd_seq snd_seq_device snd_timer snd soundcore rdma_ucm(OE) rdma_cm(OE) iw_cm(OE) ib_ipoib(OE) ib_cm(OE) ib_umad(OE) mlx5_ib(OE) mlx5_fwctl(OE) mlx5_dpll(OE) fwctl(OE) intel_ifs i10nm_edac skx_edac_common nfit x86_pkg_temp_thermal intel_powerclamp coretemp qrtr joydev input_leds hid_generic macsec kvm_intel usbhid iaa_crypto hid kvm irqbypass polyval_clmulni polyval_generic ghash_clmulni_intel sha256_ssse3 sha1_ssse3 ib_uverbs(OE) aesni_intel crypto_simd cryptd ib_core(OE) intel_cstate intel_plr_tpmi isst_tpmi ses intel_uncore_frequency_tpmi intel_rapl_tpmi cmdlinepart dax_hmem enclosure intel_rapl_common intel_tpmi_power_domains intel_uncore_frequency_common isst_tpmi_core ttm isst_if_mmio binfmt_misc spd5118 cxl_pmem spi_nor mlx5_core(OE) mtd mlxdevm(OE) mlxfw(OE) ipmi_ssif psample qat_4xxx tls pci_hyperv_intf mlx_compat(OE) pmt_telemetry isst_if_common xhci_pci_renesas intel_sdsi intel_vsec_tpmi pmt_class
[Wed Sep 23 03:46:54 2026] i2c_i801 intel_qat cxl_acpi i2c_smbus crc8 cxl_port idxd spi_intel_pci acpi_ipmi authenc i2c_mux intel_vsec idxd_bus spi_intel cxl_core ipmi_si ipmi_devintf nls_iso8859_1 einj ipmi_msghandler pfr_update pfr_telemetry mac_hid sch_fq_codel dmap(OE) msr parport_pc ppdev lp parport nvme_fabrics nvme_keyring efi_pstore nfnetlink dmi_sysfs ip_tables x_tables autofs4 mpt3sas nvme raid_class igb scsi_transport_sas nvme_core i2c_algo_bit dca nvme_auth wmi pinctrl_intel_platform [last unloaded: metax(OE)]
[Wed Sep 23 03:46:54 2026] ---[ end trace 0000000000000000 ]---
[Wed Sep 23 03:46:55 2026] RIP: 0010:xas_start+0x34/0x130
[Wed Sep 23 03:46:55 2026] Code: 48 83 ec 08 4c 8b 67 18 4c 89 e0 83 e0 03 0f 84 85 00 00 00 49 81 fc 05 c0 ff ff 76 06 48 83 f8 02 74 5f 48 8b 03 48 8b 73 08 <48> 8b 40 08 48 89 c2 83 e2 03 48 83 fa 02 75 08 48 3d 00 10 00 00
[Wed Sep 23 03:46:55 2026] RSP: 0018:ffa000003c91f4d8 EFLAGS: 00010213
[Wed Sep 23 03:46:55 2026] RAX: 7473616c00736d6f RBX: ffa000003c91f540 RCX: 0000000000000002
[Wed Sep 23 03:46:55 2026] RDX: 0000000000000000 RSI: 0000000000006500 RDI: ffa000003c91f540
[Wed Sep 23 03:46:55 2026] RBP: ffa000003c91f4f0 R08: 0000000000000100 R09: 0000000000000000
[Wed Sep 23 03:46:55 2026] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000003
[Wed Sep 23 03:46:55 2026] R13: ff1100011d8af0c8 R14: ff1100011d8af0c8 R15: ffffffff82d25b7d
[Wed Sep 23 03:46:55 2026] FS: 00007171dd102800(0000) GS:ff11005e7c880000(0000) knlGS:0000000000000000
[Wed Sep 23 03:46:55 2026] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[Wed Sep 23 03:46:55 2026] CR2: 00005c068cd21f88 CR3: 000000012684b004 CR4: 0000000000f73ef0
[Wed Sep 23 03:46:55 2026] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[Wed Sep 23 03:46:55 2026] DR3: 0000000000000000 DR6: 00000000fffe07f0 DR7: 0000000000000400
[Wed Sep 23 03:46:55 2026] PKRU: 55555554
[Wed Sep 23 03:46:55 2026] ------------[ cut here ]------------
[Wed Sep 23 03:46:55 2026] WARNING: CPU: 1 PID: 10504 at kernel/exit.c:885 do_exit+0x47a/0x4e0
[Wed Sep 23 03:46:55 2026] Modules linked in: metax(OE) snd_seq_dummy snd_hrtimer snd_seq_midi snd_seq_midi_event snd_rawmidi snd_seq snd_seq_device snd_timer snd soundcore rdma_ucm(OE) rdma_cm(OE) iw_cm(OE) ib_ipoib(OE) ib_cm(OE) ib_umad(OE) mlx5_ib(OE) mlx5_fwctl(OE) mlx5_dpll(OE) fwctl(OE) intel_ifs i10nm_edac skx_edac_common nfit x86_pkg_temp_thermal intel_powerclamp coretemp qrtr joydev input_leds hid_generic macsec kvm_intel usbhid iaa_crypto hid kvm irqbypass polyval_clmulni polyval_generic ghash_clmulni_intel sha256_ssse3 sha1_ssse3 ib_uverbs(OE) aesni_intel crypto_simd cryptd ib_core(OE) intel_cstate intel_plr_tpmi isst_tpmi ses intel_uncore_frequency_tpmi intel_rapl_tpmi cmdlinepart dax_hmem enclosure intel_rapl_common intel_tpmi_power_domains intel_uncore_frequency_common isst_tpmi_core ttm isst_if_mmio binfmt_misc spd5118 cxl_pmem spi_nor mlx5_core(OE) mtd mlxdevm(OE) mlxfw(OE) ipmi_ssif psample qat_4xxx tls pci_hyperv_intf mlx_compat(OE) pmt_telemetry isst_if_common xhci_pci_renesas intel_sdsi intel_vsec_tpmi pmt_class
[Wed Sep 23 03:46:55 2026] i2c_i801 intel_qat cxl_acpi i2c_smbus crc8 cxl_port idxd spi_intel_pci acpi_ipmi authenc i2c_mux intel_vsec idxd_bus spi_intel cxl_core ipmi_si ipmi_devintf nls_iso8859_1 einj ipmi_msghandler pfr_update pfr_telemetry mac_hid sch_fq_codel dmap(OE) msr parport_pc ppdev lp parport nvme_fabrics nvme_keyring efi_pstore nfnetlink dmi_sysfs ip_tables x_tables autofs4 mpt3sas nvme raid_class igb scsi_transport_sas nvme_core i2c_algo_bit dca nvme_auth wmi pinctrl_intel_platform [last unloaded: metax(OE)]
[Wed Sep 23 03:46:55 2026] CPU: 1 UID: 0 PID: 10504 Comm: fio Tainted: G D W OE 6.14.0-27-generic #27~24.04.1-Ubuntu
[Wed Sep 23 03:46:55 2026] Tainted: [D]=DIE, [W]=WARN, [O]=OOT_MODULE, [E]=UNSIGNED_MODULE
[Wed Sep 23 03:46:55 2026] Hardware name: Intel Corporation AvenueCity/AvenueCity, BIOS BHSDCRB1.IPC.3544.P25.2411252204 11/25/2024
[Wed Sep 23 03:46:55 2026] RIP: 0010:do_exit+0x47a/0x4e0
[Wed Sep 23 03:46:55 2026] Code: 3b 42 00 48 8b 7d c0 84 c0 0f 85 ba fe ff ff e9 ab fe ff ff 48 89 87 60 02 00 00 e9 83 fd ff ff e8 db 5f 30 00 e9 9f fe ff ff <0f> 0b e9 fc fb ff ff 0f 0b e9 b5 fb ff ff 4c 89 ee bf 05 06 00 00
[Wed Sep 23 03:46:55 2026] RSP: 0018:ffa000003c91fed8 EFLAGS: 00010286
[Wed Sep 23 03:46:55 2026] RAX: 0000000000000000 RBX: ff1100016eaf0000 RCX: 0000000000000000
[Wed Sep 23 03:46:55 2026] RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000
[Wed Sep 23 03:46:55 2026] RBP: ffa000003c91ff18 R08: 0000000000000000 R09: 0000000000000000
[Wed Sep 23 03:46:55 2026] R10: 0000000000000000 R11: 0000000000000000 R12: ff11000121911680
[Wed Sep 23 03:46:55 2026] R13: 000000000000000b R14: ff11000121a32100 R15: 0000000000000000
[Wed Sep 23 03:46:55 2026] FS: 00007171dd102800(0000) GS:ff11005e7c880000(0000) knlGS:0000000000000000
[Wed Sep 23 03:46:55 2026] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[Wed Sep 23 03:46:55 2026] CR2: 00005c068cd21f88 CR3: 000000012684b004 CR4: 0000000000f73ef0
[Wed Sep 23 03:46:55 2026] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[Wed Sep 23 03:46:55 2026] DR3: 0000000000000000 DR6: 00000000fffe07f0 DR7: 0000000000000400
[Wed Sep 23 03:46:55 2026] PKRU: 55555554
[Wed Sep 23 03:46:55 2026] Call Trace:
[Wed Sep 23 03:46:55 2026] <TASK>
[Wed Sep 23 03:46:55 2026] ? syscall_exit_to_user_mode+0x38/0x1d0
[Wed Sep 23 03:46:55 2026] make_task_dead+0x7e/0x140
[Wed Sep 23 03:46:55 2026] rewind_stack_and_make_dead+0x16/0x20
[Wed Sep 23 03:46:55 2026] RIP: 0033:0x7171dcf24e1d
[Wed Sep 23 03:46:55 2026] Code: 04 25 28 00 00 00 48 89 45 c8 31 c0 48 8d 45 10 c7 45 b0 10 00 00 00 48 89 45 b8 48 8d 45 d0 48 89 45 c0 b8 10 00 00 00 0f 05 <89> c2 3d 00 f0 ff ff 77 1a 48 8b 45 c8 64 48 2b 04 25 28 00 00 00
[Wed Sep 23 03:46:55 2026] RSP: 002b:00007ffd828b3240 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
[Wed Sep 23 03:46:55 2026] RAX: ffffffffffffffda RBX: 0000000000100000 RCX: 00007171dcf24e1d
[Wed Sep 23 03:46:55 2026] RDX: 00007ffd828b3360 RSI: 00000000c0404b1b RDI: 0000000000000007
[Wed Sep 23 03:46:55 2026] RBP: 00007ffd828b3290 R08: 0000000000000000 R09: 0000000000000007
[Wed Sep 23 03:46:55 2026] R10: 00007171dce18808 R11: 0000000000000246 R12: 00005c06aacdf4e0
[Wed Sep 23 03:46:55 2026] R13: 0000000000000000 R14: 0000000000000000 R15: 00005c06aace26c0
[Wed Sep 23 03:46:55 2026] </TASK>
[Wed Sep 23 03:46:55 2026] ---[ end trace 0000000000000000 ]---
设置了nokaslr, 虽然查询dmesg 信息符合预期,但是fio gpu direct storage测试异常,一直在准备阶段,无任何输出,但同环境改成posix,走cpu 则正常,
dmesg 查询符合预期
root@switch-PURICO:/home/switch/mxFIO/build# sudo dmesg -T |grep "device page"
[Wed Sep 23 01:58:56 2026] METAX.BASE.INFO metax pro feature device page support 1
gpu 直接 storage 测试,fio没有输出
root@switch-PURICO:/home/switch/mxFIO/build# ./fio --name=gds-test --directory=/mnt/gds/ext4 --ioengine=libcufile --cuda_io=cufile --gpu_dev_ids=0 --direct=1 --rw=write --bs=1M --size=4G --numjobs=1 --group_reporting --time_based --runtime=60 --ramp_time=4 --end_fsync=1
gds-test: (g=0): rw=write, bs=(R) 1024KiB-1024KiB, (W) 1024KiB-1024KiB, (T) 1024KiB-1024KiB, ioengine=libcufile, iodepth=1
fio-3.40
Starting 1 process
Killed1 (f=0): [/(1)][-.-%][eta 01m:04s]
mcfile 日志如下:
switch-PURICO:123148:123148 [MCFILE DEBUG] (mcFileDriverOpen:105) API_CALL
switch-PURICO:123148:123148 [MCFILE DEBUG] (mcFileBufRegister:83) API_CALL devPtr_base = 0x78ea60800000, length = 1056766, flags = 0
switch-PURICO:123148:123148 [MCFILE DEBUG] (mcFileHandleRegister:45) API_CALL descr->handle.fd = 11, descr->type = 1
switch-PURICO:123148:123148 [MCFILE DEBUG] (mcFileRead:209) API_CALL devPtr_base = 0x78ea60800000, size = 1048576, file_offset = 0, devPtr_offset = 0
相同环境,使用posix,存储storage先经过pwrite/pread 至内存,然后再通过libmcfile API至显存,fio输出正常
root@switch-PURICO:/home/switch/mxFIO/build# ./fio --name=gds-test --directory=/mnt/gds/ext4 --ioengine=libcufile --cuda_io=posix --gpu_dev_ids=0 --direct=1 --rw=read --bs=1M --size=4G --numjobs=1 --group_reporting --time_based --runtime=1 --ramp_time=4 --end_fsync=1
gds-test: (g=0): rw=read, bs=(R) 1024KiB-1024KiB, (W) 1024KiB-1024KiB, (T) 1024KiB-1024KiB, ioengine=libcufile, iodepth=1
fio-3.40
Starting 1 process
Jobs: 1 (f=1): [R(1)][100.0%][r=2262KiB/s][r=2 IOPS][eta 00m:00s]
gds-test: (groupid=0, jobs=1): err= 0: pid=123900: Wed Sep 23 02:13:16 2026
read: IOPS=68, BW=68.9MiB/s (72.3MB/s)(69.0MiB/1001msec)
clat (usec): min=481, max=447378, avg=14492.17, stdev=56946.02
lat (usec): min=481, max=447378, avg=14492.59, stdev=56946.03
clat percentiles (usec):
| 1.00th=[ 482], 5.00th=[ 510], 10.00th=[ 515], 20.00th=[ 529],
| 30.00th=[ 553], 40.00th=[ 635], 50.00th=[ 766], 60.00th=[ 807],
| 70.00th=[ 881], 80.00th=[ 26346], 90.00th=[ 27919], 95.00th=[ 28181],
| 99.00th=[446694], 99.50th=[446694], 99.90th=[446694], 99.95th=[446694],
| 99.99th=[446694]
bw ( KiB/s): min=114878, max=114878, per=100.00%, avg=114878.00, stdev= 0.00, samples=1
iops : min= 112, max= 112, avg=112.00, stdev= 0.00, samples=1
lat (usec) : 500=1.45%, 750=44.93%, 1000=27.54%
lat (msec) : 2=4.35%, 50=18.84%, 250=1.45%, 500=1.45%
cpu : usr=2.20%, sys=4.60%, ctx=138, majf=0, minf=36
IO depths : 1=100.0%, 2=0.0%, 4=0.0%, 8=0.0%, 16=0.0%, 32=0.0%, >=64=0.0%
submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
issued rwts: total=69,0,0,0 short=0,0,0,0 dropped=0,0,0,0
latency : target=0, window=0, percentile=100.00%, depth=1
执行sudo rmmod metax && sudo modprobe metax mas=1,日志显示不符合预期
root@switch-PURICO:/home/switch# dmesg -T | grep "metax pro feature"
[Tue Sep 22 02:58:21 2026] METAX.BASE.INFO metax pro feature device page support 0
root@switch-PURICO:/home/switch# rmmod metax
root@switch-PURICO:/home/switch# modprobe metax mas=1
root@switch-PURICO:/home/switch# dmesg -T | grep "metax pro feature"
[Tue Sep 22 02:58:21 2026] METAX.BASE.INFO metax pro feature device page support 0
[Tue Sep 22 03:01:58 2026] METAX.BASE.INFO metax pro feature device page support 0
查看dmesg 日志,会有[Tue Sep 22 03:03:58 2026] METAX.B6300.D0.BASE.ERROR devm_memremap_pages failed, -22 报错
[Tue Sep 22 03:03:57 2026] METAX.PCI.INFO found 1 metax pcie devices.
[Tue Sep 22 03:03:57 2026] MXCD.TOPO.INFO Update layout SYSFS entries
[Tue Sep 22 03:03:57 2026] MXCD.TOPO.INFO layout: add 2 CPU node
[Tue Sep 22 03:03:57 2026] MXCD.TOPO.INFO finish initializing layout
[Tue Sep 22 03:03:57 2026] MXCD.PEER.INFO Try to initialize peer memory client support
[Tue Sep 22 03:03:57 2026] failing symbol_get of non-GPLONLY symbol ib_register_peer_memory_client.
[Tue Sep 22 03:03:57 2026] failing symbol_get of non-GPLONLY symbol ib_register_peer_memory_client.
[Tue Sep 22 03:03:57 2026] failing symbol_get of non-GPLONLY symbol ib_unregister_peer_memory_client.
[Tue Sep 22 03:03:57 2026] failing symbol_get of non-GPLONLY symbol ib_unregister_peer_memory_client.
[Tue Sep 22 03:03:57 2026] MXCD.PEER.INFO Peer memory client interface was not detected
[Tue Sep 22 03:03:57 2026] MXCD.TOPO.INFO Create sysfs manager folder
[Tue Sep 22 03:03:57 2026] METAX.BASE.INFO load by process kworker/72:7
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.BASE.INFO initializing device (0x9999:0x4001 0x9999:0x4001 0x01) cap 0x0000000c000006cf
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.PCI.INFO pci bar5 mmio base 0x9a200000 size 0x100000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.PCI.INFO register mmio CPU va 0xff68a383f3b00000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.VIRT.INFO virt caps 0x0
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.INT.INFO MSI vectors 1 MSI-X vectors 3
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.INT.INFO using MSI
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.XCORE.INFO xcore_version 0
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.PCI.INFO MX_IOV pos e90
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.BASE.INFO PCIe atomic ops is supported
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_mc]
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_atu]
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_int]
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_sdma]
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_ccx]
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_smp]
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.BASE.INFO register ip block [mxc500_xcore]
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.SMP.INFO wait SMP0 change to be idle start 760194 wait 30 secs
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.SMP.INFO SMP0 idle now
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.SMP.INFO wait SMP1 change to be idle start 745136 wait 30 secs
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.SMP.INFO SMP1 idle now
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MGPU.INFO dnoc 0x100 topo_id 0 socket_id 0 die_id 0 die_num 1
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MGPU.INFO total die 1 available die 1 int hop base 0x0
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MGPU.INFO mgpu topo id 0 socket id 0 die_id 0
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MGPU.INFO topo_sw_id 0 sockets 1 sriov[0]
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.BASE.INFO available die num 1 activated die num 1
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.SOC.INFO vbios version 1.33.5.0, driver version 3.8.30
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MON.INFO the active status of this card is changed, new status:1
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.RAS.INFO XSC Proxy Supported
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.PCI.INFO pci bar2 doorbell base 0x1f4000000000 size 0x800000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.PCI.INFO doorbell ptr 0xff68a383dc304000 doorbell num 1024
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO XVM PT block size is fixed at 9-bit, page size is 2048 KB
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO get vram size 0x1000000000 from fuse
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO mc vram size 0x1000000000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO pcie bar0 vram original base 0x1f3000000000 size 0x1000000000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO pcie bar0 vram final base 0x1f3000000000 size 0x1000000000 cpu visible 0x1000000000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO unmapped one 0x0-0xfffffffff, off 0x0 unmapped two 0x0-0x0
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.BASE.INFO detected vram size 65536M, bar size 65536M
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.BASE.INFO vram width 4096bits HBM
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO buffer object driver init succeeded, dma32 0
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO unmapped_aperture_kaddr[0] 0xff68a38400000000-0xff68a393ffffffff
[Tue Sep 22 03:03:57 2026] METAX.MC.INFO 481945M of (XTT + Userptr) memory available
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO 65536M of vram memory ready
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO 0M of MXME memory ready
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.RAS.INFO sync bad pages from SMP
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO reserved memory index 5: offset 0x0
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO reserved memory index 6: offset 0x400000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO reserved memory index 7: offset 0xff3800000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO reserved memory index 8: offset 0x1400000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO reserved memory index 9: offset 0x1800000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO reserved memory index 10: offset 0x1c00000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO reserved memory index 11: offset 0x2000000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO reserved memory index 12: offset 0x2400000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO reserved memory index 13: offset 0x7400000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO reserved memory index 14: offset 0xc400000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO reserved memory index 15: offset 0xc800000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO reserved xtt share memory index 20: cpu_va 0xff68a383f1059000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_mc> init done!
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.ATU.INFO system memory size is 0x9ce2159000, GPA starts from 0x9d00000000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.ATU.INFO ATUL2_CTRL 0x10004 APTURE_0 BASE 0x0 TOP 0xfffffffff OFFSET 9d00000000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.ATU.INFO ATUL2_CTRL 0x10004 APTURE_0 BASE 0x0 TOP 0xfffffffff OFFSET 9d00000000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.HAG.INFO hag ATUL1 CTRL1 0x400080
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.HAG.INFO hag init done, ATUL1 CTRL3 0x82
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.HAG.INFO hag init done, ATUL1 XPID 0x0
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.HAG.INFO hag init done, ATUL1 SPACE 0x4
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO metax_kmem_init: pd_base 0x200000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.ATU.INFO KMEM PAGE BASE ENABLE (table at 0x200000) pd_pa 0x200000.
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_atu> init done!
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.INT.INFO ring-wptr addr 0x4000200000 0xff2dcbe08fff5000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.INT.INFO gpu va 0x4000400000 cpu va 0xff68a383f109a000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.INT.INFO page index 0 pa 0x63056700000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.INT.INFO wptr_addr 0x4000200000 pa 0x6308fff5000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_int> init done!
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_sdma> init done!
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.FW.INFO load /lib/firmware/metax/mxc500/vpud0_fw.bin firmware
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.FW.INFO load /lib/firmware/metax/mxc500/vpue_fw.bin firmware
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.FW.INFO load /lib/firmware/metax/mxc500/vpud1_fw.bin firmware
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_ccx> init done!
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.SMP.INFO SMP0, log_pa:c400000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.SMP.INFO SMP1, log_pa:c800000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_smp> init done!
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.SMP.INFO wait XSC change to be idle start 1227208 wait 10 secs
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.SMP.INFO XSC idle now
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.BASE.INFO IP block <mxc500_xcore> init done!
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.RAS.INFO RAS CE Monitor is on
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.SMI.INFO smi init ok
[Tue Sep 22 03:03:57 2026] MXCD.BASE.INFO allocated 1572864 bytes for tba memory in xtt.
[Tue Sep 22 03:03:57 2026] MXCD.B6300.D0.DOORBELL.INFO doorbell initialization:
[Tue Sep 22 03:03:57 2026] MXCD.B6300.D0.DOORBELL.INFO doorbell base = 0x1f4000000000
[Tue Sep 22 03:03:57 2026] MXCD.B6300.D0.DOORBELL.INFO doorbell_id_offset = 0x00000000
[Tue Sep 22 03:03:57 2026] MXCD.B6300.D0.DOORBELL.INFO db_process_ida_end = 0x00000400
[Tue Sep 22 03:03:57 2026] MXCD.B6300.D0.DOORBELL.INFO doorbell vmrange size = 0x00800000
[Tue Sep 22 03:03:57 2026] METAX.B6300.D0.MC.INFO aper: base 1f3000000000 limit 1f4000000000 pub size 1000000000 pri size 0 addr mask ffff000000000000
[Tue Sep 22 03:03:57 2026] MXCD.TOPO.INFO Adding GPU node 2(ID: 58793) to layout
[Tue Sep 22 03:03:58 2026] MXCD.B6300.D0.TOPO.INFO Added [157] GPU cache entries
[Tue Sep 22 03:03:58 2026] MXCD.TOPO.INFO Update layout SYSFS entries
[Tue Sep 22 03:03:58 2026] MXCD.B6300.D0.RINGBUF.INFO alloc private memory size 1c000000
[Tue Sep 22 03:03:58 2026] MXCD.B6300.D0.RINGBUF.INFO alloc GPU reserved memory, size=0x1000000
[Tue Sep 22 03:03:58 2026] METAX.XCORE.INFO set xcore page size to 2MB.
[Tue Sep 22 03:03:58 2026] MXCD.BASE.INFO added device 9999:4001
[Tue Sep 22 03:03:58 2026] Hotplug memory [0x1f3000000000-0x1f4000000000] exceeds maximum addressable range [0x0-0x10ffffffffff]
[Tue Sep 22 03:03:58 2026] METAX.B6300.D0.BASE.ERROR devm_memremap_pages failed, -22
[Tue Sep 22 03:03:58 2026] METAX.BASE.INFO metax pro feature device page support 0
[Tue Sep 22 03:03:58 2026] METAX.B6300.D0.BASE.INFO pcba sn: AENA2510000498, gpuid 58793, socket_id 0, topo_id 0
[Tue Sep 22 03:03:58 2026] METAX.B6300.D0.BASE.INFO native device init ok
[Tue Sep 22 03:03:58 2026] METAX.B6300.D0.BASE.INFO device pm enabled.
[Tue Sep 22 03:03:58 2026] [drm] Initialized METAX 3.8.30 for 0000:63:00.0 on minor 1
[Tue Sep 22 03:03:58 2026] METAX.BASE.INFO MetaX driver startup! built by jenkins_sw_bot, git commit ff4d0144
[Tue Sep 22 03:03:58 2026] METAX.B6300.D0.BASE.INFO systemd-logind Open kms should open mxcd first
libmcfile.so 日志输出为下文
./fio --name=gds-test --directory=/mnt/gds/ext4 --ioengine=libcufile --cuda_io=cufile --gpu_dev_ids=0 --direct=1 --rw=write --bs=1M --size=4G --numjobs=1 --group_reporting --time_based --runtime=60 --ramp_time=4 --end_fsync=1
gds-test: (g=0): rw=write, bs=(R) 1024KiB-1024KiB, (W) 1024KiB-1024KiB, (T) 1024KiB-1024KiB, ioengine=libcufile, iodepth=1
fio-3.40
Starting 1 process
switch-PURICO:22386:22386 [MCFILE DEBUG] (mcFileDriverOpen:105) API_CALL
switch-PURICO:22386:22386 [MCFILE DEBUG] (mcFileBufRegister:83) API_CALL devPtr_base = 0x7b5292200000, length = 1056766, flags = 0
switch-PURICO:22386:22386 [MCFILE DEBUG] (mcFileHandleRegister:45) API_CALL descr->handle.fd = 11, descr->type = 1
switch-PURICO:22386:22386 [MCFILE DEBUG] (mcFileWrite:217) API_CALL devPtr_base = 0x7b5292200000, size = 1048576, file_offset = 0, devPtr_offset = 0
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:150) ioctl<MXCD_IOC_VRAM_IO> failed, ret = -1, errno = 22, args.size = 1048576
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:161) direct copy failed, fallback to cpu copy
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:182) size = 1048576, dur = 3204usec(s), speed = 312.11MB/sec
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:184) Host
switch-PURICO:22386:22386 [MCFILE DEBUG] (mcFileWrite:217) API_CALL devPtr_base = 0x7b5292200000, size = 1048576, file_offset = 1048576, devPtr_offset = 0
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:150) ioctl<MXCD_IOC_VRAM_IO> failed, ret = -1, errno = 22, args.size = 1048576
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:161) direct copy failed, fallback to cpu copy
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:182) size = 1048576, dur = 1057usec(s), speed = 946.07MB/sec
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:184) Host
switch-PURICO:22386:22386 [MCFILE DEBUG] (mcFileWrite:217) API_CALL devPtr_base = 0x7b5292200000, size = 1048576, file_offset = 2097152, devPtr_offset = 0
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:150) ioctl<MXCD_IOC_VRAM_IO> failed, ret = -1, errno = 22, args.size = 1048576
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:161) direct copy failed, fallback to cpu copy
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:182) size = 1048576, dur = 1131usec(s), speed = 884.17MB/sec
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:184) Host
switch-PURICO:22386:22386 [MCFILE DEBUG] (mcFileWrite:217) API_CALL devPtr_base = 0x7b5292200000, size = 1048576, file_offset = 3145728, devPtr_offset = 0
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:150) ioctl<MXCD_IOC_VRAM_IO> failed, ret = -1, errno = 22, args.size = 1048576
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:161) direct copy failed, fallback to cpu copy
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:182) size = 1048576, dur = 920usec(s), speed = 1.06GB/sec
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:184) Host
switch-PURICO:22386:22386 [MCFILE DEBUG] (mcFileWrite:217) API_CALL devPtr_base = 0x7b5292200000, size = 1048576, file_offset = 4194304, devPtr_offset = 0
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:150) ioctl<MXCD_IOC_VRAM_IO> failed, ret = -1, errno = 22, args.size = 1048576
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:161) direct copy failed, fallback to cpu copy
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:182) size = 1048576, dur = 1088usec(s), speed = 919.12MB/sec
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:184) Host
switch-PURICO:22386:22386 [MCFILE DEBUG] (mcFileWrite:217) API_CALL devPtr_base = 0x7b5292200000, size = 1048576, file_offset = 5242880, devPtr_offset = 0
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:150) ioctl<MXCD_IOC_VRAM_IO> failed, ret = -1, errno = 22, args.size = 1048576
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:161) direct copy failed, fallback to cpu copy
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:182) size = 1048576, dur = 880usec(s), speed = 1.11GB/sec
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:184) Host
switch-PURICO:22386:22386 [MCFILE DEBUG] (mcFileWrite:217) API_CALL devPtr_base = 0x7b5292200000, size = 1048576, file_offset = 6291456, devPtr_offset = 0
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:150) ioctl<MXCD_IOC_VRAM_IO> failed, ret = -1, errno = 22, args.size = 1048576
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:161) direct copy failed, fallback to cpu copy
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:182) size = 1048576, dur = 916usec(s), speed = 1.07GB/sec
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:184) Host
switch-PURICO:22386:22386 [MCFILE DEBUG] (mcFileWrite:217) API_CALL devPtr_base = 0x7b5292200000, size = 1048576, file_offset = 7340032, devPtr_offset = 0
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:150) ioctl<MXCD_IOC_VRAM_IO> failed, ret = -1, errno = 22, args.size = 1048576
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:161) direct copy failed, fallback to cpu copy
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:182) size = 1048576, dur = 899usec(s), speed = 1.09GB/sec
switch-PURICO:22386:22386 [MCFILE DEBUG] (write:184) Host
操作系统:
PRETTY_NAME="Ubuntu 24.04.3 LTS"
NAME="Ubuntu"
VERSION_ID="24.04"
VERSION="24.04.3 LTS (Noble Numbat)"
VERSION_CODENAME=noble
ID=ubuntu
ID_LIKE=debian
HOME_URL="https://www.ubuntu.com/"
SUPPORT_URL="https://help.ubuntu.com/"
BUG_REPORT_URL="https://bugs.launchpad.net/ubuntu/"
PRIVACY_POLICY_URL="https://www.ubuntu.com/legal/terms-and-policies/privacy-policy"
UBUNTU_CODENAME=noble
LOGO=ubuntu-logo
内核版本:
Linux switch-PURICO 6.14.0-27-generic #27~24.04.1-Ubuntu SMP PREEMPT_DYNAMIC Tue Jul 22 17:38:49 UTC 2 x86_64 x86_64 x86_64 GNU/Linux
未使用容器镜像环境,已关闭iommu,未开启虚拟化
根据github.com/MetaX-MACA/mxFIO 在c500 上测试gds 发现实际路径并没有实现 gpu 到storage
测试命令如下:
./fio --name=gds-test --directory=/mnt/gds/ext4 --ioengine=libcufile --cuda_io=cufile --gpu_dev_ids=0 --direct=1 --rw=write --bs=1M --size=4G --numjobs=1 --group_reporting --time_based --runtime=60 --ramp_time=4 --end_fsync=1
mcFile 日志发现回退到cpu copy ,是什么原因呢?
switch-PURICO:21057:21057 [MCFILE DEBUG] (mcFileWrite:217) API_CALL devPtr_base = 0x74da76600000, size = 1048576, file_offset = 1115684864, devPtr_offset = 0
switch-PURICO:21057:21057 [MCFILE DEBUG] (write:150) ioctl<MXCD_IOC_VRAM_IO> failed, ret = -1, errno = 22, args.size = 1048576
switch-PURICO:21057:21057 [MCFILE DEBUG] (write:161) direct copy failed, fallback to cpu copy
switch-PURICO:21057:21057 [MCFILE DEBUG] (write:182) size = 1048576, dur = 465usec(s), speed = 2.10GB/sec
switch-PURICO:21057:21057 [MCFILE DEBUG] (write:184) Host
switch-PURICO:21057:21057 [MCFILE DEBUG] (mcFileWrite:217) API_CALL devPtr_base = 0x74da76600000, size = 1048576, file_offset = 1116733440, devPtr_offset = 0
switch-PURICO:21057:21057 [MCFILE DEBUG] (write:150) ioctl<MXCD_IOC_VRAM_IO> failed, ret = -1, errno = 22, args.size = 1048576
switch-PURICO:21057:21057 [MCFILE DEBUG] (write:161) direct copy failed, fallback to cpu copy
switch-PURICO:21057:21057 [MCFILE DEBUG] (write:182) size = 1048576, dur = 457usec(s), speed = 2.14GB/sec
switch-PURICO:21057:21057 [MCFILE DEBUG] (write:184) Host