Note: This is a public test instance of Red Hat Bugzilla. The data contained within is a snapshot of the live data so any changes you make will not be reflected in the production Bugzilla. Email is disabled so feel free to test any aspect of the site that you want. File any problems you find or give feedback at bugzilla.redhat.com.
Bug 1661568 - Tons of BUG: non-zero pgtables_bytes on freeing mm: -16384" in the 4.19 kernel of f29
Summary: Tons of BUG: non-zero pgtables_bytes on freeing mm: -16384" in the 4.19 ker...
Keywords:
Status: CLOSED ERRATA
Alias: None
Product: Fedora
Classification: Fedora
Component: kernel
Version: 29
Hardware: s390x
OS: Linux
unspecified
high
Target Milestone: ---
Assignee: Kernel Maintainer List
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks: ZedoraTracker
TreeView+ depends on / blocked
 
Reported: 2018-12-21 15:33 UTC by Christian Borntraeger
Modified: 2019-01-07 12:47 UTC (History)
21 users (show)

Fixed In Version: kernel-4.19.13-200.fc28 kernel-4.19.13-300.fc29
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2019-01-04 12:28:26 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)


Links
System ID Private Priority Status Summary Last Updated
IBM Linux Technology Center 174269 0 None None None 2019-04-01 10:41:33 UTC

Description Christian Borntraeger 2018-12-21 15:33:28 UTC
User-Agent:       Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101 Firefox/60.0
Build Identifier: 

I see (for almost every process)
the following message
BUG: non-zero pgtables_bytes on freeing mm: -16384"
after upgrading from the 4.18 to 4.19 kernel (via yum).

Reproducible: Always

Steps to Reproduce:
update your fedora 29 on s390x to latest greatest kernel 4.19.9-300.fc29.s390x
Actual Results:  
BUG: non-zero pgtables_bytes on freeing mm: -16384"

Expected Results:  
no bug messages in dmesg

This seems to have been fixed upstream with 

e12e4044aede9 s390/mm: fix mis-accounting of pgtable_bytes

dependencies are probably the following commits:

6d212db11947a mm: add mm_pxd_folded checks to pgtable_bytes accounting functions
1071fc5779d98 mm: introduce mm_[p4d|pud|pmd]_folded
a8874e7e8a889 mm: make the __PAGETABLE_PxD_FOLDED defines non-empty

Comment 1 Steve 2018-12-23 04:45:34 UTC
(In reply to Christian Borntraeger from comment #0)
...
> This seems to have been fixed upstream with 
> 
> e12e4044aede9 s390/mm: fix mis-accounting of pgtable_bytes
...

That appears to have gone into v4.20-rc2, so you might try a rawhide kernel:
https://mirrors.kernel.org/fedora/development/rawhide/Everything/x86_64/os/Packages/k/

s390/mm: fix mis-accounting of pgtable_bytes
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v4.20-rc2&id=e12e4044aede97974f2222eb7f0ed726a5179a32

There was also a later merge commit with the same patch:

Merge tag 's390-4.20-2' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v4.20-rc2&id=3541833fd1f264e7579e573a6586a1b665da37db

Comment 2 Steve 2018-12-23 05:47:45 UTC
(In reply to Steve from comment #1)

> ... so you might try a rawhide kernel:
> https://mirrors.kernel.org/fedora/development/rawhide/Everything/x86_64/os/
> Packages/k/

Link to s390x architecture:
https://mirrors.kernel.org/fedora-secondary/development/rawhide/Everything/s390x/os/Packages/k/

Comment 3 Christian Borntraeger 2018-12-27 08:36:50 UTC
FWIW, 4.19.2-301.fc29.s390x does not have the issue, so the warning must have sneaked in between 4.19.2 and 4.19.10

The 4.20 from rawhide does not have the pgtable_bytes issue but it is crashing very early in random places when running on f29. No idea yet whats wrong (the same kernel runs fine when booting that via virt-install from
pub/fedora-secondary/development/rawhide/Server/s390x/os/

Comment 4 Christian Borntraeger 2018-12-27 09:13:41 UTC
Seems that this is caused by a partial backport of e12e4044aede9 s390/mm: fix mis-accounting of pgtable_bytes in 4.19.5

See https://lore.kernel.org/lkml/4e58571e-1991-4a70-7012-ff9dd76afee3@de.ibm.com/T/#u

Comment 5 Dan Horák 2019-01-04 12:26:02 UTC
Reading the thread mentioned in comment #4 there will be a fix in the stable 4.19 tree. And looking into the 4.19 log it's already in 4.19.13.

Comment 6 Christian Borntraeger 2019-01-07 12:47:26 UTC
I can confirm that 4.19.13-300.fc29.s390x fixes the issue.


Note You need to log in before you can comment on or make changes to this bug.