Date: Wed, 15 Sep 2021 15:49:01 -0400 From: Alexander Motin <mav@FreeBSD.org> To: mike tancsa <mike@sentex.net>, FreeBSD-STABLE Mailing List <freebsd-stable@freebsd.org> Subject: Re: ipmi0: Watchdog set returned 0xc0 (releng_13) Message-ID: <8fcb399d-8f26-b7a2-bf60-414d4ea1006e@FreeBSD.org> In-Reply-To: <586b04b4-d9b7-1597-4c8f-9bbcb9255c5a@sentex.net> References: <c065ff32-719f-ef2e-85b9-d5504a36810f@sentex.net> <51e58d7a-1b33-f7e3-13aa-3be5fee5c826@FreeBSD.org> <2c1ad9db-b92a-d0a8-b2ae-b3ffafc04604@sentex.net> <fc5f879f-91da-c45d-366c-29c721de948e@FreeBSD.org> <f93369e5-51b6-1a0c-cc14-e16a71be314b@sentex.net> <1f0d32e1-e907-080f-1669-65148c245dd8@sentex.net> <b6f97a7d-73ce-c3d3-ee56-5a7ae5bfed05@FreeBSD.org> <b5d6dbbc-92b8-52db-0aab-d2aa04288f8d@sentex.net> <586b04b4-d9b7-1597-4c8f-9bbcb9255c5a@sentex.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On 15.09.2021 14:49, mike tancsa wrote:
> On 9/15/2021 11:23 AM, mike tancsa wrote:
>>
>> I also tried on a X11SSL-F
>>
>> ipmi0: IPMI device rev. 1, firmware rev. 1.60, version 2.0, device
>> support mask 0xbf
>> ipmi0: Number of channels 2
>> ipmi0: Attached watchdog
>> ipmi0: Establishing power cycle handler
>>
>> # ipmitool sel list | tail -3
>> 6 | 08/20/2021 | 20:45:38 | Fan #0x45 | Lower Non-recoverable going
>> low | Asserted
>> 7 | 09/15/2021 | 11:15:28 | Watchdog2 #0xca | Timer interrupt () |
>> Asserted
>> 8 | 09/15/2021 | 11:15:38 | Watchdog2 #0xca | Power cycle () | Asserted
>> #
>>
>> I have a RELENG_12 box in production I will try as well later, but so
>> far so good. Thanks for fixing!
>
> On RELENG_12 as of today, I still have the issue on a X11SCL-F board.
What issues do you have there exactly?
> Not sure if thats the board, RELENG_12 or other missing changes from
> HEAD or a combo there of ?
>
> --- ipmi.c.prev 2021-09-15 14:39:28.943161000 -0400
> +++ ipmi.c 2021-09-15 14:39:50.997836000 -0400
> @@ -662,7 +662,8 @@
> req->ir_request[0] = IPMI_SET_WD_TIMER_DONT_STOP
> | IPMI_SET_WD_TIMER_SMS_OS;
> req->ir_request[1] = (wd_timer_actions & 0xff);
> - req->ir_request[2] = (wd_pretimeout_countdown & 0xff);
> + req->ir_request[2] = min(0xff,
> + min(wd_pretimeout_countdown, (sec + 2) / 4));
> req->ir_request[3] = 0; /* Timer use */
> req->ir_request[4] = (sec * 10) & 0xff;
> req->ir_request[5] = (sec * 10) >> 8;
I suppose this is what you merged down? It by itself should fix the
specific issue that affected at least me.
> I did notice, there are other changes in the version in HEAD
>
> @@ -27,16 +27,19 @@
> */
>
> #include <sys/cdefs.h>
> -__FBSDID("$FreeBSD: stable/12/sys/dev/ipmi/ipmi.c 370303 2021-08-13
> 01:20:59Z git2svn $");
> +__FBSDID("$FreeBSD$");
>
> #include <sys/param.h>
> #include <sys/systm.h>
> #include <sys/bus.h>
> #include <sys/condvar.h>
> #include <sys/conf.h>
> +#include <sys/eventhandler.h>
> #include <sys/kernel.h>
> +#include <sys/lock.h>
> #include <sys/malloc.h>
> #include <sys/module.h>
> +#include <sys/mutex.h>
> #include <sys/poll.h>
> #include <sys/reboot.h>
> #include <sys/rman.h>
> @@ -93,11 +96,14 @@
> static int wd_startup_countdown = 0; /* sec */
> static int wd_pretimeout_countdown = 120; /* sec */
> static int cycle_wait = 10; /* sec */
> +static int wd_init_enable = 1;
>
> -static SYSCTL_NODE(_hw, OID_AUTO, ipmi, CTLFLAG_RD, 0,
> +static SYSCTL_NODE(_hw, OID_AUTO, ipmi, CTLFLAG_RD | CTLFLAG_MPSAFE, 0,
> "IPMI driver parameters");
> SYSCTL_INT(_hw_ipmi, OID_AUTO, on, CTLFLAG_RWTUN,
> &on, 0, "");
> +SYSCTL_INT(_hw_ipmi, OID_AUTO, wd_init_enable, CTLFLAG_RWTUN,
> + &wd_init_enable, 1, "Enable watchdog initialization");
> SYSCTL_INT(_hw_ipmi, OID_AUTO, wd_timer_actions, CTLFLAG_RW,
> &wd_timer_actions, 0,
> "IPMI watchdog timer actions (including pre-timeout interrupt)");
> @@ -662,7 +668,8 @@
> req->ir_request[0] = IPMI_SET_WD_TIMER_DONT_STOP
> | IPMI_SET_WD_TIMER_SMS_OS;
> req->ir_request[1] = (wd_timer_actions & 0xff);
> - req->ir_request[2] = (wd_pretimeout_countdown & 0xff);
> + req->ir_request[2] = min(0xff,
> + min(wd_pretimeout_countdown, (sec + 2) / 4));
> req->ir_request[3] = 0; /* Timer use */
> req->ir_request[4] = (sec * 10) & 0xff;
> req->ir_request[5] = (sec * 10) >> 8;
> @@ -906,7 +913,7 @@
> * Probe for watchdog, but only for backends which support
> * polled driver requests.
> */
> - if (sc->ipmi_driver_requests_polled) {
> + if (wd_init_enable && sc->ipmi_driver_requests_polled) {
> IPMI_INIT_DRIVER_REQUEST(req,
> IPMI_ADDR(IPMI_APP_REQUEST, 0),
> IPMI_GET_WDOG, 0, 0);
>
> @@ -949,14 +956,14 @@
> } else if (!on)
> (void)ipmi_set_watchdog(sc, 0);
> /*
> - * Power cycle the system off using IPMI. We use last - 1 since
> we don't
> + * Power cycle the system off using IPMI. We use last - 2 since
> we don't
> * handle all the other kinds of reboots. We'll let others
> handle them.
> * We only try to do this if the BMC supports the Chassis device.
> */
> if (sc->ipmi_dev_support & IPMI_ADS_CHASSIS) {
> device_printf(dev, "Establishing power cycle handler\n");
> sc->ipmi_power_cycle_tag =
> EVENTHANDLER_REGISTER(shutdown_final,
> - ipmi_power_cycle, sc, SHUTDOWN_PRI_LAST - 1);
> + ipmi_power_cycle, sc, SHUTDOWN_PRI_LAST - 2);
> }
> }
This seems to be irrelevant.
--
Alexander Motin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?8fcb399d-8f26-b7a2-bf60-414d4ea1006e>