On Mon, Dec 09, 2024 at 09:05:46AM +0100, Magnus Lindholm wrote: > > Maybe, maybe not. Building 6.11.11 with alpha-linux-gnu-gcc-9 > > (Debian 9.3.0-22) fixed the null pointer access in the scsi > > subsystem, and finally I have a bootable system! (Tested on ES45.) > > But I now get a null pointer access when network driver loaded: > > Very interesting! Can you provide a full stack dump? interested to > see if this is the same issue as on my system. [ 33.144514] e100 0001:02:04.0 enP1p2s4: NIC Link is Up 100 Mbps Full Duplex [ 36.878887] Unable to handle kernel paging request at virtual address 0000000000000000 [ 36.879863] CPU 0 [ 36.879863] kworker/0:1(10): Oops -1 [ 36.881817] pc = [<0000000000000000>] ra = [<0000000000000001>] ps = 0000 Not tainted [ 36.882793] pc is at 0x0 [ 36.883770] ra is at 0x1 [ 36.884746] v0 = 0000000000000007 t0 = 0000000000000001 t1 = fffffc000091cc80 [ 36.884746] t2 = 0000000000000000 t3 = fffffc00040cca40 t4 = 0000000000000001 [ 36.884746] t5 = 0000000000000001 t6 = 0000000000000000 t7 = fffffc0004124000 [ 36.884746] s0 = fffffc000a037e00 s1 = fffffc000a037f28 s2 = fffffffffffffed8 [ 36.884746] s3 = fffffc0004020b40 s4 = fffffc0001c8d688 s5 = fffffc0004020bc0 [ 36.884746] s6 = fffffc0001c8d680 [ 36.884746] a0 = fffffc0001c5b320 a1 = fffffc000ac14200 a2 = fffffc00040ccac0 [ 36.884746] a3 = fffffc0003df36c0 a4 = fffffc000dfb3d38 a5 = ffffffffffffffff [ 36.884746] t8 = 0000000000000000 t9 = 0000000000000000 t10= 0000000000000001 [ 36.884746] t11= 0000000000000001 pv = fffffc00018b7180 at = 0000000000000001 [ 36.884746] gp = fffffc0001cb08c8 sp = 00000000ab1ca88d [ 36.884746] Disabling lock debugging due to kernel taint [ 36.896465] Trace: [ 36.896465] [<fffffc0001049c8c>] process_scheduled_works+0xdc/0x420 [ 36.898418] [<fffffc000104a740>] worker_thread+0x0/0x3d0 [ 36.899395] [<fffffc000104a8f0>] worker_thread+0x1b0/0x3d0 [ 36.900371] [<fffffc000104a740>] worker_thread+0x0/0x3d0 [ 36.901348] [<fffffc000105705c>] kthread+0x17c/0x1c0 [ 36.902324] [<fffffc000104a740>] worker_thread+0x0/0x3d0 [ 36.903301] [<fffffc0001011198>] ret_from_kernel_thread+0x18/0x20 [ 36.904277] [<fffffc0001056ee0>] kthread+0x0/0x1c0 [ 36.905254] [<fffffc0001011180>] ret_from_kernel_thread+0x0/0x20 [ 36.907207] Code: [ 36.907207] 00000000 [ 36.908184] 00000000 [ 36.909160] 00063301 [ 36.910137] 00000e90 [ 36.911113] 00001111 [ 36.912090] 0000795c > Have you tried passing rcupdate.rcu_normal=1 as boot parameter > to the kernel? That allowed me to bypass the network interface > problem on my system. Haven't as yet, but backtrace above does not go through rcu routines. Maybe I am seeing a different issue. Cheers, Michael.
