Based on kernel version 4.16.1. Page generated on 2018-04-09 11:53 EST.
1 Last reviewed: 05/20/2016 2 3 HPE iLO NMI Watchdog Driver 4 NMI sourcing for iLO based ProLiant Servers 5 Documentation and Driver by 6 Thomas Mingarelli 7 8 The HPE iLO NMI Watchdog driver is a kernel module that provides basic 9 watchdog functionality and the added benefit of NMI sourcing. Both the 10 watchdog functionality and the NMI sourcing capability need to be enabled 11 by the user. Remember that the two modes are not dependent on one another. 12 A user can have the NMI sourcing without the watchdog timer and vice-versa. 13 All references to iLO in this document imply it also works on iLO2 and all 14 subsequent generations. 15 16 Watchdog functionality is enabled like any other common watchdog driver. That 17 is, an application needs to be started that kicks off the watchdog timer. A 18 basic application exists in tools/testing/selftests/watchdog/ named 19 watchdog-test.c. Simply compile the C file and kick it off. If the system 20 gets into a bad state and hangs, the HPE ProLiant iLO timer register will 21 not be updated in a timely fashion and a hardware system reset (also known as 22 an Automatic Server Recovery (ASR)) event will occur. 23 24 The hpwdt driver also has three (3) module parameters. They are the following: 25 26 soft_margin - allows the user to set the watchdog timer value. 27 Default value is 30 seconds. 28 allow_kdump - allows the user to save off a kernel dump image after an NMI. 29 Default value is 1/ON 30 nowayout - basic watchdog parameter that does not allow the timer to 31 be restarted or an impending ASR to be escaped. 32 Default value is set when compiling the kernel. If it is set 33 to "Y", then there is no way of disabling the watchdog once 34 it has been started. 35 36 NOTE: More information about watchdog drivers in general, including the ioctl 37 interface to /dev/watchdog can be found in 38 Documentation/watchdog/watchdog-api.txt and Documentation/IPMI.txt. 39 40 The NMI sourcing capability is disabled by default due to the inability to 41 distinguish between "NMI Watchdog Ticks" and "HW generated NMI events" in the 42 Linux kernel. What this means is that the hpwdt nmi handler code is called 43 each time the NMI signal fires off. This could amount to several thousands of 44 NMIs in a matter of seconds. If a user sees the Linux kernel's "dazed and 45 confused" message in the logs or if the system gets into a hung state, then 46 the hpwdt driver can be reloaded. 47 48 1. If the kernel has not been booted with nmi_watchdog turned off then 49 edit and place the nmi_watchdog=0 at the end of the currently booting 50 kernel line. Depending on your Linux distribution and platform setup: 51 For non-UEFI systems 52 /boot/grub/grub.conf or 53 /boot/grub/menu.lst 54 For UEFI systems 55 /boot/efi/EFI/distroname/grub.conf or 56 /boot/efi/efi/distroname/elilo.conf 57 2. reboot the sever 58 3. Once the system comes up perform a modprobe -r hpwdt 59 4. modprobe /lib/modules/`uname -r`/kernel/drivers/watchdog/hpwdt.ko 60 61 Now, the hpwdt can successfully receive and source the NMI and provide a log 62 message that details the reason for the NMI (as determined by the HPE BIOS). 63 64 Below is a list of NMIs the HPE BIOS understands along with the associated 65 code (reason): 66 67 No source found 00h 68 69 Uncorrectable Memory Error 01h 70 71 ASR NMI 1Bh 72 73 PCI Parity Error 20h 74 75 NMI Button Press 27h 76 77 SB_BUS_NMI 28h 78 79 ILO Doorbell NMI 29h 80 81 ILO IOP NMI 2Ah 82 83 ILO Watchdog NMI 2Bh 84 85 Proc Throt NMI 2Ch 86 87 Front Side Bus NMI 2Dh 88 89 PCI Express Error 2Fh 90 91 DMA controller NMI 30h 92 93 Hypertransport/CSI Error 31h 94 95 96 97 -- Tom Mingarelli