Bug 43014 - Kernel freeze on Acer Aspire 1510
Summary: Kernel freeze on Acer Aspire 1510
Status: NEW
Alias: None
Product: Other
Classification: Unclassified
Component: Other (show other bugs)
Hardware: All Linux
: P1 normal
Assignee: other_other
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2012-03-30 17:44 UTC by Karl-Felix Glatzer
Modified: 2014-08-13 15:56 UTC (History)
2 users (show)

See Also:
Kernel Version: 3.3.0
Subsystem:
Regression: Yes
Bisected commit-id:


Attachments
Freeze with the 3.3.0 kernel (602.24 KB, image/jpeg)
2012-03-30 17:44 UTC, Karl-Felix Glatzer
Details
Output of dmesg run on Ubuntu Lucid using the 2.6.32-40-generic kernel (40.35 KB, text/plain)
2012-03-30 17:53 UTC, Karl-Felix Glatzer
Details
Output of lspci -vvnn run on Ubuntu Lucid using the 2.6.32-40-generic kernel (11.29 KB, text/plain)
2012-03-30 17:54 UTC, Karl-Felix Glatzer
Details
dmesg log with pci=earlydump,realloc (56.76 KB, text/plain)
2012-09-03 17:29 UTC, Karl-Felix Glatzer
Details
lspci -t (389 bytes, text/plain)
2012-09-03 17:30 UTC, Karl-Felix Glatzer
Details
new dmesg log with corrected pci=earlydump,realloc (79.72 KB, text/plain)
2012-09-04 11:38 UTC, Karl-Felix Glatzer
Details
lspci -t on v2.6.32 (389 bytes, text/plain)
2012-09-12 16:47 UTC, Karl-Felix Glatzer
Details
lspci -vvnn on v3.6 (11.37 KB, text/plain)
2012-09-12 16:48 UTC, Karl-Felix Glatzer
Details
Kernel v3.6-rc3 with override_bios=1 (674.41 KB, image/jpeg)
2012-09-12 17:01 UTC, Karl-Felix Glatzer
Details
Freeze on v3.15.8 (1016.12 KB, image/jpeg)
2014-08-11 18:45 UTC, Karl-Felix Glatzer
Details
Freeze on v3.16 (957.87 KB, image/jpeg)
2014-08-13 15:56 UTC, Karl-Felix Glatzer
Details

Description Karl-Felix Glatzer 2012-03-30 17:44:37 UTC
Created attachment 72765 [details]
Freeze with the 3.3.0 kernel

On my Acer Aspire 1510 notebook all kernels newer than 2.6.33 freeze while booting. When freezing the notebook completely ignores keyboard input and the fan speed turns up noticeably.

As I am using Ubuntu I've filed a bug report on launchpad which you can find here:

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/911210

I've been able to reproduce the bug on builds of upstream mainline kernels provided for Ubuntu and thus was asked to file an upstream bug report here.

I've attached a photo of the output printed before the freeze when using the 3.3.0 kernel.
Comment 1 Karl-Felix Glatzer 2012-03-30 17:53:56 UTC
Created attachment 72766 [details]
Output of dmesg run on Ubuntu Lucid using the 2.6.32-40-generic kernel
Comment 2 Karl-Felix Glatzer 2012-03-30 17:54:54 UTC
Created attachment 72767 [details]
Output of lspci -vvnn run on Ubuntu Lucid using the 2.6.32-40-generic kernel
Comment 3 Lee, Chun-Yi 2012-08-24 22:47:53 UTC
Hi, 

Did you try 3.5 or 3.6-rc1 kernel can still reproduce this issue?

Thanks
Comment 4 Karl-Felix Glatzer 2012-08-25 15:30:41 UTC
I just tried the latest ubuntu mainline kernel (3.6-rc3) and can reproduce the issue with it. However when blacklisting yenta_socket the kernel boots fine.
Comment 5 Lee, Chun-Yi 2012-08-27 06:28:20 UTC
(In reply to comment #4)
> I just tried the latest ubuntu mainline kernel (3.6-rc3) and can reproduce
> the
> issue with it. However when blacklisting yenta_socket the kernel boots fine.

There have BAR allocation fail on your machine with VGA and CardBus bridge. 

Please try 'pci=realloc' kernel parameter.
Comment 6 Karl-Felix Glatzer 2012-09-02 07:40:40 UTC
i've tried booting the 3.6-rc3 kernel with pci=realloc but it still freezes at
"pcmcia_socket pcmcia_socket0: cs: IO port probe 0x3000-0x3fff".
Comment 7 Lee, Chun-Yi 2012-09-03 02:11:11 UTC
(In reply to comment #6)
> i've tried booting the 3.6-rc3 kernel with pci=realloc but it still freezes
> at
> "pcmcia_socket pcmcia_socket0: cs: IO port probe 0x3000-0x3fff".

hmm...

Please attach the following log files:

 + dmesg log with pci=earlydump,realloc

 + 'lspci -t' result

Thanks
Comment 8 Karl-Felix Glatzer 2012-09-03 17:29:50 UTC
Created attachment 79131 [details]
dmesg log with pci=earlydump,realloc
Comment 9 Karl-Felix Glatzer 2012-09-03 17:30:22 UTC
Created attachment 79141 [details]
lspci -t
Comment 10 Lee, Chun-Yi 2012-09-03 21:13:42 UTC
(In reply to comment #8)
> Created an attachment (id=79131) [details]
> dmesg log with pci=earlydump,realloc

Hi, 

Per dmesg log, you give a wrong kernel parameter:

pci-earlzdump,realloc

Please use:

pci=earlydump,realloc


thanks
Comment 11 Karl-Felix Glatzer 2012-09-04 11:38:05 UTC
Created attachment 79201 [details]
new dmesg log with corrected pci=earlydump,realloc

Oh I'm sorry. Here's a new dmesg log with the correct parameter.
Comment 12 Lee, Chun-Yi 2012-09-12 06:40:37 UTC
(In reply to comment #11)
> Created an attachment (id=79201) [details]
> new dmesg log with corrected pci=earlydump,realloc
> 
> Oh I'm sorry. Here's a new dmesg log with the correct parameter.

hm... that's interesting the BAR 16 on 0000:02:07.0 cardbus device still reassign fail:

[    0.179976] pci 0000:02:07.0: BAR 16: can't assign mem (size 0x8000000) <=== need 128M?

I am aware in bug description, you said this issue doesn't happen before v2.6.33 kernel. And I saw the same BAR 16 assign fail in dmesg on comment#1, that means on v2.6.32, yenta_socket driver has no problem with BAR assign fail.
I simply check the yenta_socket patches bewteen v2.6.32 to v2.6.33, only 2 coding style changed.

Could you please provide the following information for compare difference?
 # lspci -vvnn on v3.6 kernel
 # lspci -t on v2.6.32 kernel

And, please try 'pci=cbmemsize=128M' kernel parameter, then attach on dmesg.

If cdmemsize doesn't work, please try override_bios=1 module parameter:
 # vi /etc/modprobe.d/yenta_socket.conf
add 'options yenta_socket override_bios=1'


Thanks a lot!
Comment 13 Lee, Chun-Yi 2012-09-12 08:15:49 UTC
Hi Karl-Felix, 

Please ignore my suggestion on Comment#12 to try 'pci=cbmemsize=128M', it's stupid for use bigger memsize for cardbus.

Besides this, please still provide the lspci result and try override_bios parameter.

Thanks
Comment 14 Karl-Felix Glatzer 2012-09-12 16:47:07 UTC
Created attachment 79911 [details]
lspci -t on v2.6.32
Comment 15 Karl-Felix Glatzer 2012-09-12 16:48:21 UTC
Created attachment 79921 [details]
lspci -vvnn on v3.6
Comment 16 Karl-Felix Glatzer 2012-09-12 17:01:00 UTC
Created attachment 79931 [details]
Kernel v3.6-rc3 with override_bios=1

I've tried booting the v3.6-rc3 kernel with the override_bios=1 parameter of yenta_socket and now it freezes with this output.
Comment 17 Karl-Felix Glatzer 2014-08-11 18:45:30 UTC
Created attachment 146251 [details]
Freeze on v3.15.8

The problem seems to be partially fixed upstream (tested with v3.15.8 and v3.16). The laptop boots and runs fine but prints this error message when loading the yenta_cardbus module:

yenta_cardbus 0000:02:07.0: no bus associated! (try 'pci=assign-busses')

Booting with pci=assign-busses triggers the freeze. Attached is a picture of the freeze on v3.15.8.

When booting without pci=assign-busses the card reader works though. It detects my SD-Card and i can read from it and write to it.
Comment 18 Andrey Utkin 2014-08-12 10:14:49 UTC
Please try again with 3.15.9, or 3.16, or linux-next git.
Comment 19 Karl-Felix Glatzer 2014-08-13 15:56:53 UTC
Created attachment 146481 [details]
Freeze on v3.16

Output of booting v3.16 with pci=assign-busses.

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