Hi, Here are some observations from running linux smp kernels on sparc32 hardware: linux-2.6.31.9 seems stable on both hypersparc and supersparc, runs fine for long periods of time even under quite heavy load. linux-2.32.x also runs fine some instability notices with lookups on both hypersparc and supersparc but I've had a ss10 dual superparc-II running for over 100 days without problems so its fairly reliable. Sometimes. With kernel 2.6.32.46 I see things like: Jan 20 00:46:23 ss10 kernel: BUG: rwlock wrong CPU on CPU#2, tr/7032, f0234020 Jan 20 00:46:23 ss10 kernel: [f0038420 : do_exit+0x5b0/0x64c ] [f0038768 : do_group_exit+0x88/0xb8 ] [f00387a8 : sys_exit_group+0x10/0x1c ] [f001229c : syscall_is_too_hard+0x34/0x40 ] [f0011fa0 : srmmu_fault+0x58/0x68 ] f0234020=tasklist_lock More recent kernels will run smp but freezes fairly quick under heavy load (like building the kernel), The more recent kernel the less uptime I get under heavy load. On linux 3.4.x I get messages like: "BUG: scheduling while atomic: swapper/2/0/0x00000000" These messages are frequent and quickly fills my log files whit the same repeated message. The exception seems to bee 3.5-rc1 (modified for hypersparc) which on hypersparc runs fairly stable even i smp. Building a kernel with "make -j2" is no problem and I can do it over and over again. However building a kernel with "make -j4" will cause the system too freeze after like 30min. Not a complete lockup though, I can ping the machine, I can toggle numlock and if I hit ctlr-alt-del it displays a message that the system is about to reboot, but then nothing happens. On supersparc-II the modified 3.5-rc1 kernel will freeze alot faster than on hypersparc, not sure why but smp on hypersparc seems much more stable than with supersparc-II at least on 3.5-rc1. Magnus -- To unsubscribe from this list: send the line "unsubscribe sparclinux" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
