Masao Uebayashi <uebayasi%gmail.com@localhost> writes: > On Mon, Jun 17, 2013 at 4:33 PM, Manuel Bouyer > <bouyer%antioche.eu.org@localhost> wrote: >> On Mon, Jun 17, 2013 at 11:41:24AM +0900, Masao Uebayashi wrote: >>> [...] >>> I also saw docs mentioning that some Intel chipsets have WDT. But >>> info is very limited. Not sure if it's real. >> >> They do; some of them even have several: >> Available watchdog timers: >> ipmi0, 0 second period >> ichlpcib0, 367 second period >> >> ichlpcib0 at pci0 dev 31 function 0: Intel 63xxESB LPC Interface >> Bridge (rev. 0x09) >> timecounter: Timecounter "ichlpcib0" frequency 3579545 Hz quality 1000 >> ichlpcib0: 24-bit timer >> ichlpcib0: TCO (watchdog) timer configured. >> ipmi0 at mainbus0 >> ipmi0: version 2.0 interface KCS iobase 0xca2/2 spacing 1 >> >> >> Available watchdog timers: >> ichlpcib0, 38 second period [armed, user tickle, pid 11] >> timecounter: Timecounter "ichlpcib0" frequency 3579545 Hz quality 1000 >> ichlpcib0: 24-bit timer >> ichlpcib0: TCO (watchdog) timer configured. >> >> >> Basically, all hardware I have which has a IPMI BMC includes a ipmi watchdog; >> and most intel P4 or newer desktop or server I have includes a ichlpcib >> watchdog. > > Nice. This would be one of the reasons real H/W machines are better > than VMs; hardware reset effectively resets H/W. > > Unfortunately, my customer runs VMware, which have no H/W watchdog > emulation. (It has heartbeat as a host-guest IPC (used via vmware > tools), but I can't use it for administrative reasons...) It would be cool if xen had a watchdog that was code in the dom0 that the domU had to poke, so from the domU point of view, it had a real (virtual) hardware watchdog.
Description: PGP signature