Hey everybody, I'm the arch lead for Gentoo Linux on alpha (yes, the arch still exists). Starting with 4.20-rc*, I get SCSI reset loops shortly after boot:
aboot: starting kernel boot/vmlinuz-4.20.0-rc3 with arguments ro root=/dev/sda2 console=ttyS0 loglevel=6 [ 0.000000] Linux version 4.20.0-rc3+ (klausman@monolith) (gcc version 8.2.0 (Gentoo 8.2.0-r4 p1.5)) #49 SMP Sun Nov 25 20:23:08 CET 2018 [ 0.000000] Booting on Tsunami variation Clipper using machine vector Clipper from SRM [ 0.000000] Major Options: SMP EV67 LEGACY_START VERBOSE_MCHECK DISCONTIGMEM MAGIC_SYSRQ [ 0.000000] Command line: ro root=/dev/sda2 console=ttyS0 loglevel=6 [ 0.000000] Raw memory layout: [ 0.000000] memcluster 0, usage 1, start 0, end 256 [ 0.000000] memcluster 1, usage 0, start 256, end 130985 [ 0.000000] memcluster 2, usage 1, start 130985, end 131072 [ 0.000000] memcluster 3, usage 0, start 131072, end 1048562 [ 0.000000] memcluster 4, usage 1, start 1048562, end 1048576 [ 0.000000] Initializing bootmem allocator on Node ID 0 [ 0.000000] memcluster 1, usage 0, start 256, end 130985 [ 0.000000] memcluster 3, usage 0, start 131072, end 1048562 [ 0.000000] Detected node memory: start 256, end 1048562 [ 0.000000] 8192K Bcache detected; load hit latency 21 cycles, load miss latency 175 cycles [ 0.000000] Kernel command line: ro root=/dev/sda2 console=ttyS0 loglevel=6 [ 0.000000] Sorting __ex_table... [ 0.000000] random: get_random_u64 called from cache_random_seq_create+0x98/0x1b0 with crng_init=0 [ 0.000000] random: get_random_u32 called from new_slab+0x210/0x700 with crng_init=0 [ 0.000000] Turning on RTC interrupts. [ 0.069335] random: get_random_u32 called from bucket_table_alloc+0xbc/0x230 with crng_init=0 [ 0.069335] random: fast init done [ 0.218749] SCSI subsystem initialized [ 0.244140] pci 0001:01:03.0: Firmware left e100 interrupts enabled; disabling [ 0.247070] Initialise system trusted keyrings [ 0.272460] Key type asymmetric registered [ 0.273437] Asymmetric key parser 'x509' registered [ 0.441406] sym0: Symbios NVRAM, ID 7, Fast-40, LVD, parity checking [ 0.442382] sym0: open drain IRQ line driver, using on-chip SRAM [ 0.443359] sym0: using LOAD/STORE-based firmware. [ 0.444335] sym0: SCSI BUS has been reset. [ 0.445312] sym0:0: ERROR (81:0) (0-0-0) (0/7/0) @ (scripta 38:f31c0004). [ 0.445312] sym0: script cmd = e21c0004 [ 0.445312] sym0: regdump: [ 0.445312] ca [ 0.445312] 00 [ 0.445312] 00 [ 0.445312] 07 [ 0.445312] 47 [ 0.445312] 00 [ 0.445312] 00 [ 0.445312] 0f [ 0.445312] 00 [ 0.445312] 08 [ 0.445312] 00 [ 0.445312] 00 [ 0.445312] 80 [ 0.445312] 00 [ 0.445312] 08 [ 0.445312] 0a [ 0.445312] 07 [ 0.445312] ff [ 0.445312] ff [ 0.445312] ff [ 0.445312] 00 [ 0.445312] ff [ 0.445312] ff [ 0.445312] ff [ 0.445312] . (repeats here, forever) The machine is a Compaq ES40 (4-cpu, 8G of RAM) with two 53c895s, which have six SCA disks each. Some of these disks are quite old, over a decade in some cases, but have never had problems (and booting 4.19.5 works just fine). I haven't tested 4.20-rc5 yet, but I doubt the behaviour will be different. As for the rest, it's a pretty normal setup aside from the exotic arch, no fancy extra buses, no AGP or memory holes. I'll try to bisect this when the machine can be taken out of testing and image building, but that might not happen before Christmas or even the new year. Any ideas for something to try? Best, Tobias PS: Not subscribed to the list, so please CC me on replies, thanks!
