Bug 20071 - Kernel 4.9.x kernel stack overflow booting
Summary: Kernel 4.9.x kernel stack overflow booting
Status: RESOLVED FIXED
Alias: None
Product: Mageia
Classification: Unclassified
Component: RPM Packages (show other bugs)
Version: Cauldron
Hardware: x86_64 Linux
Priority: Normal critical
Target Milestone: ---
Assignee: Kernel and Drivers maintainers
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2017-01-03 21:10 CET by Morgan Leijström
Modified: 2017-01-13 08:16 CET (History)
1 user (show)

See Also:
Source RPM: kernel-4.9.1-1.mga6.src.rpm
CVE:
Status comment:


Attachments
output of journalctl -xb (24.39 KB, application/gzip)
2017-01-03 21:12 CET, Morgan Leijström
Details

Description Morgan Leijström 2017-01-03 21:10:45 CET
This system have been running following cauldron successfully a half year including kernel 4.8.15-1 but now it fails using kernel 4.9.0-3; in syslog:

jan 03 20:16:59 kajsas kernel: BUG: stack guard page was hit at ffffb5b9008f0000 (stack is ffffb5b9008ec000..ffffb5b9008effff)
jan 03 20:16:59 kajsas kernel: kernel stack overflow (page fault): 0000 [#1] SMP

I tried uninstalling and resinstalling kernel.
(it still run OK when booting 4.8.15-1)

Other details: Lenovo ThinkPad T60, model 2007FVG; ATI graphics, SSD with LVM but separate /boot.
Comment 1 Morgan Leijström 2017-01-03 21:12:16 CET
Created attachment 8832 [details]
output of journalctl -xb
Morgan Leijström 2017-01-03 21:14:41 CET

Assignee: bugsquad => kernel
Source RPM: (none) => kernel-4.9.0-3.mga6.src.rpm

Comment 2 Thomas Backlund 2017-01-04 13:30:35 CET
seem to have been reported upstream at 4.9-rc1 times but no reaction so far..

If you blacklist the modules: tpm_tis tpm_tis_core tpm

does it boot up then ?

CC: (none) => tmb

Comment 3 Morgan Leijström 2017-01-04 14:43:10 CET
YES :)  by adding to kernel boot line manually at grub: 
 modprobe.blacklist=tpm_tis,tpm_tis_core,tpm

However now i tried by creating /etc/modprobe.d/blacklist-tpm, containing

# mga bug 20071, workaround f kernel 4.9.0-3 @ tmb
blacklist tpm
blacklist tpm_tis
blacklist tpm_tis_core

And it do not work, even after reinstalling kernel.
What am i doing wrong?



Sidenote:
on this machine when booting other kernels i have long seen the messages:

tpm tpm0: tpm_transmit: tpm_send: error -5
tpm_tis: could not get TPM timeouts and durations
Comment 4 Morgan Leijström 2017-01-04 17:54:27 CET
For upstream you mean this report ? : https://lkml.org/lkml/2016/10/24/150
Interesting: That is also a Lenovo Thinkpad T6x... i believe that one have Intel X3100 GPU

I have two other thinkpads T61 (type 7665-CTO) and T61p (type 6457-5KG but upgraded with T9500 CPU) that run well with this kernel, some differences against the one in my comment 0 (type 2007FVG) is that they use Nvidia GPU (instead of ATI) and no disk encryption (the one above use ext4 partitions on LVM in one encrypted partition, all set up by diskdrake.)

Then there are probably some difference in chips on main board like network, USB, 1394, sound...

Oh and all three have a modified (Middleton 2.29) BIOS for faster SATA and some other tweaks; all use SSD.
Comment 5 Morgan Leijström 2017-01-05 09:15:52 CET
Another difference is that the laptop where stack overflow happen have a sticker "Duo" while the working laptops stickers say "vPro", which i know implies a more efficient memory buffering etc, so other chipset version.
Comment 6 Morgan Leijström 2017-01-07 22:59:22 CET
Still same problem and workaround on 4.9.1-1

( PS: workaround is to add
   modprobe.blacklist=tpm_tis,tpm_tis_core,tpm
to kernel start parameters using MCC -> start -> configure start system 
  DS )

What are possible implications of blacklisting tpm?

Summary: Kernel 4.9.0-3 kernel stack overflow booting => Kernel 4.9.x kernel stack overflow booting
Source RPM: kernel-4.9.0-3.mga6.src.rpm => kernel-4.9.1-1.mga6.src.rpm

Comment 7 Morgan Leijström 2017-01-11 09:55:17 CET
Problem is no more on 4.9.2 :)

During booting, at 0:06 i still see the messages - as i have done for long time  (when not blacklisting tpm) :

  tpm tpm0: tpm_transmit: tpm_send: error -5
  tpm_tis: could not get TPM timeouts and durations

I do not know if that is a sign of some trouble or not.



Thomas, tell me if you want me to test something, or you can just close this bug.

I will reopen if i see this problem again.
Comment 8 Morgan Leijström 2017-01-13 08:16:52 CET
kernel-4.9.3-1 OK too.  Closing.

Status: NEW => RESOLVED
Resolution: (none) => FIXED


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