Login
[x]
Log in using an account from:
Fedora Account System
Red Hat Associate
Red Hat Customer
Or login using a Red Hat Bugzilla account
Forgot Password
Login:
Hide Forgot
Create an Account
Red Hat Bugzilla – Attachment 910105 Details for
Bug 1109967
kernel BUG at kernel/smpboot.c:134! after moving all tasks to a given cpuset cgroup
[?]
New
Simple Search
Advanced Search
My Links
Browse
Requests
Reports
Current State
Search
Tabular reports
Graphical reports
Duplicates
Other Reports
User Changes
Plotly Reports
Bug Status
Bug Severity
Non-Defaults
|
Product Dashboard
Help
Page Help!
Bug Writing Guidelines
What's new
Browser Support Policy
5.0.4.rh83 Release notes
FAQ
Guides index
User guide
Web Services
Contact
Legal
This site requires JavaScript to be enabled to function correctly, please enable it.
Why flag PF_NO_SETAFFINITY is cleared on select_fallback_rq when the fixed CPU is not on NUMA node 0
why_the_flag_is_cleared_on_numa_node_non_0.txt (text/plain), 1.68 KB, created by
Daniel Bristot de Oliveira
on 2014-06-18 17:14:22 UTC
(
hide
)
Description:
Why flag PF_NO_SETAFFINITY is cleared on select_fallback_rq when the fixed CPU is not on NUMA node 0
Filename:
MIME Type:
Creator:
Daniel Bristot de Oliveira
Created:
2014-06-18 17:14:22 UTC
Size:
1.68 KB
patch
obsolete
>select_task_rq() >the cpu is always initialized as 0. >and then it enters in the if (because the tasks are not allowed to run on cpu 0, unless it is for the CPU 0 :-)) > > call the select_fallback_rq(task_cpu(p), p); > the first argument (cpu) is always 0 because set_task_cpu(fixed CPU) was not called yet. > > int nid = cpu_to_node(cpu); <----- here we have a logical error, because it will always return node 0, because cpu is always = 0; > > /* > * If the node that the cpu is on has been offlined, cpu_to_node() > * will return -1. There is no cpu on the node, and we should > * select the cpu on the other node. > */ > if (nid != -1) { <----- it will always enter here because nid = 0. > nodemask = cpumask_of_node(nid); <----- here we have another error, if the desired CPU is not on the same NUMA node that CPU0. > > /* Look for allowed, online CPU in same node. */ > for_each_cpu(dest_cpu, nodemask) { > if (!cpu_online(dest_cpu)) > continue; > if (!cpu_active(dest_cpu)) > continue; > if (cpumask_test_cpu(dest_cpu, tsk_cpus_allowed(p))) <---- if tsk_cpus_allowed(p) returns a CPU on the numa node 0, it will match. > return dest_cpu; <---- when it returns here, the flag is not cleared. > } > } > > if the allowed cpu is not on the numa node 0, it will continue and the flag will be cleared. >
You cannot view the attachment while viewing its details because your browser does not support IFRAMEs.
View the attachment on a separate page
.
View Attachment As Raw
Actions:
View
Attachments on
bug 1109967
:
909211
| 910105 |
911799