talos-hostboot/src/usr/initservice, branch master

istepdispatcher: Reset the watchog to enable it

2018-05-19T22:16:25+00:00

Performing a watchdog set command alone is not enough to re-enable the watchdog. In any instance where we want the watchdog to be running we need to issue a reset immediately after the set action to guarantee it was started. Resolves #137 Signed-off-by: William A. Kennington III Change-Id: I16b617cf26c9e6af8151ac415689a8c6956c6e11 Reviewed-on: http://ralgit01.raleigh.ibm.com/gerrit1/58547 Reviewed-by: Daniel M. Crowell Tested-by: Daniel M. Crowell

istepdispatcher: When trying to set the watchdog, clear DONT_STOP

2018-05-19T22:14:25+00:00

We intend to actually stop the watchdog so allow the set command stop it as defined in the IPMI spec. Resolves #137 Signed-off-by: William A. Kennington III Change-Id: I07152602dd0c03c0224a20e5bea3d4ff804fc093 Reviewed-on: http://ralgit01.raleigh.ibm.com/gerrit1/58546 Tested-by: Jenkins Server Tested-by: Jenkins OP Build CI Tested-by: Jenkins OP HW Tested-by: FSP CI Jenkins Reviewed-by: Daniel M. Crowell

Check if Console Library is Loaded Before Printing to Console

2018-05-11T02:31:09+00:00

When a failure occurs early in IPL (for instance, due to corrupt secure headers), the doShutdown path attempts to print a trace to the console, which, if the console has not been initialized yet, locks up the shutdown path, and we don't shut down gracefully. This change adds a check to make sure the console library has been loaded prior to flushing the trace in doShutdown path. Change-Id: I74942476de0a8c3f4cba41b92b05369687441f68 CQ: SW426722 Reviewed-on: http://ralgit01.raleigh.ibm.com/gerrit1/58346 Tested-by: Jenkins Server Tested-by: Jenkins OP Build CI Tested-by: Jenkins OP HW Tested-by: FSP CI Jenkins Reviewed-by: Michael Baiocchi Reviewed-by: Nicholas E. Bofferding Reviewed-by: Daniel M. Crowell

Log Hostboot state information in all error logs

2018-04-13T16:33:21+00:00

Add the Istep number and substep number to Hostboot Error Logs to assist in problem debug. These items will be added as a user data record to all Hostboot Error Logs prior to Runtime. Change-Id: Ib5d269f3c1853d6d6856669744630dc130fa7094 Reviewed-on: http://ralgit01.raleigh.ibm.com/gerrit1/54764 Reviewed-by: Martin Gloff Reviewed-by: Matt Derksen Tested-by: Jenkins Server Tested-by: Jenkins OP Build CI Tested-by: Jenkins OP HW Tested-by: FSP CI Jenkins Reviewed-by: Daniel M. Crowell

Get Hostboot Dump after Hostboot Crash

2018-04-09T20:38:00+00:00

When Hostboot has a TI with a PLID rather than a Reason Code (RC), HWSV does not attempt the Hostboot Dump. Thought is that happens per the design as the error log related to the PLID sent by Hostboot should already have the required information. There are instances where the error log does not have sufficient data and a dump would be beneficial. A path is being added so an error log can be flagged as needing a Hostboot dump. Change-Id: I97972308c70e7210f578fb818563bb9b0cd940b4 CQ: SW420219 Reviewed-on: http://ralgit01.raleigh.ibm.com/gerrit1/56600 CI-Ready: Daniel M. Crowell CI-Ready: Martin Gloff Reviewed-by: Matt Derksen Tested-by: Jenkins Server Tested-by: Jenkins OP Build CI Reviewed-by: Corey V. Swenson Tested-by: Jenkins OP HW Tested-by: FSP CI Jenkins Reviewed-by: Daniel M. Crowell

Do not elevate severity of reconfig error log

2018-03-12T16:39:37+00:00

The code as currently written will elevate the severity of any log that is used to terminate the hostboot ipl. In most cases this is the correct behavior as we always want a visible log. However, there are cases where hostboot terminates for the sole purpose of triggering a reconfig loop via the FSP in order to recover from a recoverable hardware issue. In this case the log shouldn't be visible since we don't want the customer to take any actions. The fix is to add a little bit of logic to differentiate these two scenarios to control the severity setting. Change-Id: I7253aec8c28a40c5cdebf4933ceccbecd119b9f4 CQ: SW420495 Backport: yes Reviewed-on: http://ralgit01.raleigh.ibm.com/gerrit1/55379 CI-Ready: Nicholas E. Bofferding CI-Ready: Daniel M. Crowell Tested-by: Jenkins Server Tested-by: Jenkins OP Build CI Tested-by: Jenkins OP HW Tested-by: FSP CI Jenkins Reviewed-by: Nicholas E. Bofferding Reviewed-by: Marshall J. Wilks Reviewed-by: William G. Hoffa

Remove all explicit calls to special_wakeup from hostboot code

2018-02-17T01:22:29+00:00

The only time that special wakeup should be called on the hostboot platform is when it is called by other HWPs. PM_RESET will only enable special wakeup if the SKIP_WAKEUP attribute is set to 0 Change-Id: Ib9939eae857fa5113a876a717f8169e150b78318 Reviewed-on: http://ralgit01.raleigh.ibm.com/gerrit1/54148 Tested-by: Jenkins Server Tested-by: Jenkins OP Build CI Tested-by: Jenkins OP HW Tested-by: FSP CI Jenkins Reviewed-by: Daniel M. Crowell

Send additional FSP progress messages during long isteps (e.g. memdiags)

2018-02-16T02:52:28+00:00

Updated sendProgressCode() so that it would only output trace or console messages once per step/substep. Replaced most calls to IPMIWATCHDOG::resetWatchDogTimer() with calls to sendProgressCode() which calls resetWatchDogTimer() if communicating with a BMC, or updates the FSP with its progress on FSP-based systems. Change-Id: I29beb7ce5cdae467d26a0b2c5fee9e3cc4629161 RTC: 169682 Reviewed-on: http://ralgit01.raleigh.ibm.com/gerrit1/53995 Reviewed-by: William G. Hoffa Reviewed-by: Prachi Gupta Tested-by: Jenkins Server Tested-by: Jenkins OP Build CI Tested-by: Jenkins OP HW Tested-by: FSP CI Jenkins Reviewed-by: Daniel M. Crowell

Switch polarity of return data for perst commands

2018-02-12T15:11:47+00:00

HWSV and Hostboot had backwards interpretations of fail/success. Changing Hostboot to return zero for success. Change-Id: Ia21b2b7c78eb9c64f06a690a50740357441b23c1 CQ: SW415796 Reviewed-on: http://ralgit01.raleigh.ibm.com/gerrit1/53653 Tested-by: Jenkins Server Tested-by: Jenkins OP Build CI Tested-by: Jenkins OP HW Reviewed-by: Brian E. Bakke Tested-by: FSP CI Jenkins Reviewed-by: Daniel M. Crowell

host_coalesce_host: bup updates

2018-02-06T20:33:40+00:00

- move p9_fab_iovalid to fab_iovalid common lib to avoid conflicts with hwp lib - doorbell_send to wake up the cores in istep18 - move block wakeup interrupts set later in the istep 18 flow after we have suspended the mbox and drained the interrupt queue Change-Id: I241240ca1d1787182c5baaf3bdd10283878d5798 Reviewed-on: http://ralgit01.raleigh.ibm.com/gerrit1/52701 Tested-by: Jenkins Server Tested-by: Jenkins OP Build CI Tested-by: Jenkins OP HW Reviewed-by: SWATHI M. BHATTIPROLU Reviewed-by: Sakethan R. Kotta Tested-by: FSP CI Jenkins Reviewed-by: Martin Gloff Reviewed-by: William G. Hoffa