Re: Recent 'Make if_rge MPSAFE.' breaks by Radxa ROCK 5 Model B Plus

On 23/03/2026 10:27, Markus Kilbinger wrote:
> Hi!
> 
> After recent -current change
> 
>    https://mail-index.netbsd.org/source-changes/2026/03/22/msg161130.html
> 
> my Radxa ROCK 5 Model B Plus panics during net config:
> 
> [   1.0000040] acpipchb2 at acpi0 (PCI4, PNP0A08-4): PCI Express Host Bridge
> [   1.0000040] pci2 at acpipchb2 bus 101
> [   1.0000040] rge0 at pci2 dev 0 function 0: vendor 10ec product 8125
> (rev. 0x05)
> [   1.0000040] rge0: interrupting at irq 282
> [   1.0000040] rge0: HW rev. B
> [   1.0000040] rge0: Ethernet address 00:48:54:xx:xx:xx
> [...]
> IPv6 mode: host
> Configuring network interfaces: rge0[   4.1822413] Reader / writer
> lock error: rw_vector_enter,298: assertion failed: !cpu_intr_p()
> 
> [   4.1822413] lock address : netbsd:kernel_map_store+0x8
> [   4.1822413] current cpu  :                  0
> [   4.1822413] current lwp  : 0xffff0003fd21a580
> [   4.1822413] owner/count  : 000000000000000000 flags    : 000000000000000000
> 
> [   4.1822413] panic: lock error: Reader / writer lock:
> rw_vector_enter,298: assertion failed: !cpu_intr_p(): lock
> 0xffffc0000111a6d8 cpu 0 lwp 0xffff0003fd21a580
> [   4.1822413] cpu0: Begin traceback...
> [   4.1822413] trace fp ffffc004bc8cef80
> [   4.1822413] fp ffffc004bc8cefb0 vpanic() at ffffc000005eeee4
> netbsd:vpanic+0x1bc
> [   4.1822413] fp ffffc004bc8cf010 panic() at ffffc000005eefa4 netbsd:panic+0x44
> [   4.1922437] fp ffffc004bc8cf0a0 lockdebug_abort() at
> ffffc000005e2e60 netbsd:lockdebug_abort+0xdc
> [   4.1922437] fp ffffc004bc8cf170 rw_enter() at ffffc000005a7bb4
> netbsd:rw_enter+0x394
> [   4.1922437] fp ffffc004bc8cf210 uvm_fault_internal() at
> ffffc0000053b894 netbsd:uvm_fault_internal+0xe4
> [   4.1922437] fp ffffc004bc8cf420 data_abort_handler() at
> ffffc000000c9848 netbsd:data_abort_handler+0xc8
> [   4.1922437] tf ffffc004bc8cf490 el1_trap() at ffffc000000caf84
> netbsd:el1_vectors+0x784
> [   4.1922437] ---- Instruction Abort (EL1): trapframe
> 0xffffc004bc8cf490 (304 bytes) ----
> [...]
> 
> Reverting this change makes everything working fine again.
> 
> Provide more infos?
> send-pr?

I think I fixed this with sys/dev/pci/if_rge.c:1.53

Nick