Patrick Rudolph has posted comments on this change. ( https://review.coreboot.org/c/coreboot/+/34570 )
Change subject: cpu/x86/mp_init: Wait longer with serial enabled
......................................................................
Patch Set 2:
Patch Set 2:
Patch Set 1: Code-Review-1
While a bigger timeout fixes the issue I'm seeing and it's related to slow serial, the fix is incorrect.
The main issue is that `mp_run_on_all_cpus` doesn't wait for CPUs to finish execution.
That means consecutive calls to mp_run_on_all_cpus expect all CPUs to pick-up the work, but they are still busy with the old job (due to high core count and console spinlock).
I guess that's the reason mp_park_aps has this big timeout.
An idea how to fix it:
Wait it mp_run_on_aps for all APs to signal idle state before trying to launch a new job.
How long would this wait be? If a timed wait is added for APs to signal idle, wouldn't it mean that the timeout is just moved from "AP accepting work" to "AP finishing old work"?
Yes, true the timeout would just be moved within this function.
There's no benefit to current code.
--
To view, visit
https://review.coreboot.org/c/coreboot/+/34570
To unsubscribe, or for help writing mail filters, visit
https://review.coreboot.org/settings
Gerrit-Project: coreboot
Gerrit-Branch: master
Gerrit-Change-Id: Id47df02a9238e66f2b628b9d6805858724bf30a9
Gerrit-Change-Number: 34570
Gerrit-PatchSet: 2
Gerrit-Owner: Patrick Rudolph
patrick.rudolph@9elements.com
Gerrit-Reviewer: Aaron Durbin
adurbin@chromium.org
Gerrit-Reviewer: Alexandru Gagniuc
alexandrux.gagniuc@intel.com
Gerrit-Reviewer: Christian Walter
christian.walter@9elements.com
Gerrit-Reviewer: Furquan Shaikh
furquan@google.com
Gerrit-Reviewer: Julius Werner
jwerner@chromium.org
Gerrit-Reviewer: Kyösti Mälkki
kyosti.malkki@gmail.com
Gerrit-Reviewer: Lee Leahy
leroy.p.leahy@intel.com
Gerrit-Reviewer: Marshall Dawson
marshalldawson3rd@gmail.com
Gerrit-Reviewer: Naresh Solanki
naresh.solanki@intel.com
Gerrit-Reviewer: Patrick Rudolph
patrick.rudolph@9elements.com
Gerrit-Reviewer: Richard Spiegel
richard.spiegel@silverbackltd.com
Gerrit-Reviewer: Subrata Banik
subrata.banik@intel.com
Gerrit-Reviewer: build bot (Jenkins)
no-reply@coreboot.org
Gerrit-CC: Paul Menzel
paulepanter@users.sourceforge.net
Gerrit-Comment-Date: Sun, 28 Jul 2019 08:09:17 +0000
Gerrit-HasComments: No
Gerrit-Has-Labels: No
Gerrit-MessageType: comment