[BUG] skge 0000:02:05: read data parity error

Previous thread: [oops report analysis] hfs_bnode_split() one (Arjan's #2753) by Al Viro on Monday, January 14, 2008 - 12:41 pm. (1 message)

Next thread: [GIT PULL] more i2c fixes for 2.6.24 by Jean Delvare on Monday, January 14, 2008 - 2:01 pm. (1 message)
From: Oliver Pinter (Pintér Olivér)
Date: Monday, January 14, 2008 - 12:57 pm

Hi All!

It is fully reproductable under 2.6.22.15, 2.6.23.13 (all tainted and
not tainted [4 different kernel] ) and 2 different PC:

[BUG] skge 0000:02:05: read data parity error
[BUG] skge 0000:02:05: read data parity error

steps:
1. login as root
2. start mc
3. cd /sys/bus/pci/drivers/skge/0000:02:05.0
4. press F3 (mcview) on resource0
5. the system hang up, without panic or bug ... only this message
printed 2x: [BUG] skge 0000:02:05: read data parity error

when I used cat for show what is in file , then bocome this message:
 cat: resource0: Input/output error
but the permissions is:
 -rw------- 1 root root 16384 2008-01-14 20:36 resource0

when I used mcview for show whats in file, then the system hang up,
and not reagiert neither for SysRQ-s, only for hard reset.

-----------

PC1.:
00:00.0 Host bridge [0600]: Intel Corporation 945G/GZ/P/PL Express
Memory Controller Hub [8086:2770] (rev 02)
00:01.0 PCI bridge [0604]: Intel Corporation 945G/GZ/P/PL Express PCI
Express Root Port [8086:2771] (rev 02)
00:1b.0 Audio device [0403]: Intel Corporation 82801G (ICH7 Family)
High Definition Audio Controller [8086:27d8] (rev 01)
00:1d.0 USB Controller [0c03]: Intel Corporation 82801G (ICH7 Family)
USB UHCI #1 [8086:27c8] (rev 01)
00:1d.1 USB Controller [0c03]: Intel Corporation 82801G (ICH7 Family)
USB UHCI #2 [8086:27c9] (rev 01)
00:1d.2 USB Controller [0c03]: Intel Corporation 82801G (ICH7 Family)
USB UHCI #3 [8086:27ca] (rev 01)
00:1d.3 USB Controller [0c03]: Intel Corporation 82801G (ICH7 Family)
USB UHCI #4 [8086:27cb] (rev 01)
00:1d.7 USB Controller [0c03]: Intel Corporation 82801G (ICH7 Family)
USB2 EHCI Controller [8086:27cc] (rev 01)
00:1e.0 PCI bridge [0604]: Intel Corporation 82801 PCI Bridge
[8086:244e] (rev e1)
00:1f.0 ISA bridge [0601]: Intel Corporation 82801GB/GR (ICH7 Family)
LPC Interface Bridge [8086:27b8] (rev 01)
00:1f.2 IDE interface [0101]: Intel Corporation 82801GB/GR/GH (ICH7
Family) Serial ATA Storage Controller IDE [8086:27c0] (rev ...
From: Stephen Hemminger
Date: Monday, January 14, 2008 - 1:52 pm

On Mon, 14 Jan 2008 20:57:49 +0100

This is not a bug.

The hardware has some debug registers that if accessed cause a read
back to the host. Since this can point anywhere, it will cause errors
or system hang.

The point is don't do it.

-- 
Stephen Hemminger <stephen.hemminger@vyatta.com>
--

From: Adrian Bunk
Date: Monday, January 14, 2008 - 2:01 pm

Is it really a good idea that _reading_ files under /sys can kill your 
machine?

That sounds like a huge trap for people debugging their machine (or e.g. 
forgetting to exclude /sys from their backup).

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

--

From: Greg KH
Date: Monday, January 14, 2008 - 2:30 pm

No, but mmapping them as root and then doing bad things with that data
isn't recommended :)

Note that this is a special file, just like the ones in /proc, that
provide a mmap interface into the pci card's resource.  This is nothing
new...

thanks,

greg k-h
--

From: Oliver Pinter (Pintér Olivér)
Date: Monday, January 14, 2008 - 2:31 pm

I think, it is a potential security breakpoint, when applications with
root permission its read, then a machine is freezed, or only i thin
it's?


-- 
Thanks,
Oliver
--

From: Adrian Bunk
Date: Monday, January 14, 2008 - 2:34 pm

When you are root there are infinite ways to kill your machine, so 

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

--

From: Oliver Pinter (Pintér Olivér)
Date: Monday, January 14, 2008 - 2:41 pm

Yes, i know, but when some application or daemons read some file with
running root privileges, then ...




-- 
Thanks,
Oliver
--

From: Greg KH
Date: Monday, January 14, 2008 - 2:48 pm

It's always been that way, this is nothing new.

thanks,

greg k-h
--

From: Adrian Bunk
Date: Monday, January 14, 2008 - 2:52 pm

It might be a bug in the application.

But there are worse things than crashing your machine (e.g. getting your 
/etc/shadow) that can happen when someone with bad intentions can read 

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

--

From: Oliver Pinter (Pintér Olivér)
Date: Monday, January 14, 2008 - 2:58 pm

I "tested" other devices resources file, and only with skge freezed
the system. from this think, that is skge driver bug

-- 
Thanks,
Oliver
--

From: Stephen Hemminger
Date: Monday, January 14, 2008 - 3:37 pm

On Mon, 14 Jan 2008 22:58:51 +0100

Its a property of the hardware, and the current device model has
no way to stop it.  Other devices are worse and can die if you access
pci config space.




-- 
Stephen Hemminger <stephen.hemminger@vyatta.com>
--

From: Greg KH
Date: Monday, January 14, 2008 - 2:38 pm

I'm sorry, I don't quite understand what you are trying to say here.
--

From: Oliver Pinter (Pintér Olivér)
Date: Monday, January 14, 2008 - 2:44 pm

huh, sorry, it's typo and i'm not learn english,  only myself ... and
my spelling is very bad, sorry

-- 
Thanks,
Oliver
--

Previous thread: [oops report analysis] hfs_bnode_split() one (Arjan's #2753) by Al Viro on Monday, January 14, 2008 - 12:41 pm. (1 message)

Next thread: [GIT PULL] more i2c fixes for 2.6.24 by Jean Delvare on Monday, January 14, 2008 - 2:01 pm. (1 message)