Re: NetBSD on crashes on Radxa Rock 5B+

[Robert Swindells <rjs%fdy2.co.uk@localhost>]

Ted Spradley <tsprad%talent-free-studios.com@localhost> wrote:
> On Tue, 10 Nov 2026 08:50:18 +0000 (UTC)
> Benny Siegert <bsiegert%netbsd.org@localhost> wrote:
>
>> We are getting somewhere!
>> 
>> My rk3588 has an NVMe 1.1 disk that works perfectly. Putting in a newer 
>> NVME 1.4 makes it crash on startup like you describe.
>
> For what it's worth, I have a OrangePi 5+ RK3588:
>
> Last login: Tue Nov 10 12:02:47 2026 from 10.0.0.186
> NetBSD 11.1 (GENERIC64) #0: Mon Nov  2 16:34:32 UTC 2026
>
> Welcome to NetBSD!
>
> netop$ dmesg | grep -i nvme
> [     1.000004] nvme0 at pci0 dev 0 function 0: vendor 126f product
> 2263 (rev. 0x03) [     1.000004] nvme0: NVMe 1.3
> [     1.000004] nvme0: interrupting at irq 292
> [     1.000004] nvme0: Timetec 35TTFP6PCIE-1TB, firmware V1103A0,
> serial YS240104B1T2341 [     1.000004] ld4 at nvme0 nsid 1
> netop$ 
>
> NVMe 1.3 works for me

The common factor for people having problems seemed to be NVMe 1.4.

I have an Orange Pi 5 MAX with an NVMe 1.4 SSD that is working fine
but it is using a different PCIe driver to you (and to the VisionFive
2).

rk3568pcie1 at simplebus0: RK3568 PCIe pcie@fe150000
rk3588pciephy0: acquire len 0
rk3588pciephy0: enable 128 1
rk3568pcie1: atu viewport ffffffff
rk3568pcie1: IO f0100000 f0100000 100000
rk3568pcie1: MEM f0200000 f0200000 e00000
rk3568pcie1: PMEM 40000000 900000000 40000000
rk3568pcie1: IO: 0xf0100000+0x100000@0xf0100000
rk3568pcie1: MMIO (32-bit non-prefetchable): 0xf0200000+0xe00000@0xf0200000
rk3568pcie1: MMIO (64-bit prefetchable): 0x40000000+0x40000000@0x900000000
pci2 at rk3568pcie1 bus 0
pci2: i/o space, memory space enabled, rd/line, rd/mult, wr/inv ok
ppb1 at pci2 dev 0 function 0: vendor 1d87 product 3588 (rev. 0x00)
ppb1: PCI Express capability version 2 <Root Port of PCI-E Root Complex> x4 @ 8.
0GT/s
pci3 at ppb1 bus 1
pci3: i/o space, memory space enabled, rd/line, wr/inv ok
nvme0 at pci3 dev 0 function 0: vendor 2646 product 501d (rev. 0x00)
nvme0: NVMe 1.4
rk3568pcie1: intx_establish nvme0 ipl 5 flags 1
nvme0: interrupting at GICv3 irq 294
nvme0: KINGSTON SNV2S250G, firmware CRTP3011, serial 50026B738359A34A
ld4 at nvme0 nsid 1
ld4: 232 GB, 30401 cyl, 255 head, 63 sec, 512 bytes/sect x 488397168 sectors

Has anyone checked that the proposed patch from the port-riscv thread is
ok to commit?