Bug 5152

Summary: 2.6.13 hangs during boot
Product: Drivers Reporter: Larry Finger (Larry.Finger)
Component: PCIAssignee: Greg Kroah-Hartman (greg)
Status: REJECTED INVALID    
Severity: blocking CC: bunk, kernelbugs
Priority: P2    
Hardware: i386   
OS: Linux   
Kernel Version: 2.6.13 Subsystem:
Regression: --- Bisected commit-id:

Description Larry Finger 2005-08-29 20:42:09 UTC
Most recent kernel where this bug did not occur: 2.6.13-rc7
Distribution: SuSE 9.2
Hardware Environment: HP Pavilion ze1115 notebook
Software Environment:
Problem Description: 2.6.13 fails to boot. It prints the "loading ACPI .... 
done" message, then hangs. On other kernels, the next message is "starting
powersaved ..." The configuration is identical with the one used for 2.6.13-rc7.

Steps to reproduce:
Comment 1 Larry Finger 2005-08-29 21:15:37 UTC
Since the first filing, I have more information. The hang only occurs when my
Linksys WPC54G PCMCIA card is installed. The card uses ndiswrapper and a Windows
driver; however, the hang happened before the ndiswrapper module had been built
for 2.6.13. The wifi card connects with my network and gets its DHCP IP number -
it looks as if the card and PCMCIA system are both working until ACPI is
started. Then the machine hangs and must be powered off.
Comment 2 Adrian Bunk 2005-08-30 07:14:11 UTC
Does this problem also occur if ndiswrapper was _never_ loaded since booting?

If it doesn't, this bug should be closed because external drivers are off-topic
in this Bugzilla and binary-only drivers undebuggable.
Comment 3 Larry Finger 2005-08-30 07:39:23 UTC
Yes - the problem occurs when the ndiswrapper module had never been loaded.
Today, I plan to selectively remove the patches between 2.6.13-rc7 and 2.6.13 to
find which causes the problem.
Comment 4 Larry Finger 2005-08-30 10:07:21 UTC
Removing the patch below fixes the problem. Obviously it should be reassigned to
the PCI group.

diff -aur linux-2.6.13-rc7/drivers/pci/setup-bus.c
/home/finger/linux-2.6.13/drivers/pci/setup-bus.c
--- linux-2.6.13-rc7/drivers/pci/setup-bus.c    2005-08-24 19:19:32.000000000 -0500
+++ /home/finger/linux-2.6.13/drivers/pci/setup-bus.c   2005-08-28
18:41:01.000000000 -0500
@@ -40,7 +40,7 @@
  * FIXME: IO should be max 256 bytes.  However, since we may
  * have a P2P bridge below a cardbus bridge, we need 4K.
  */
-#define CARDBUS_IO_SIZE                (4096)
+#define CARDBUS_IO_SIZE                (256)
 #define CARDBUS_MEM_SIZE       (32*1024*1024)

 static void __devinit
Comment 5 Larry Finger 2005-08-30 19:15:22 UTC
Further experimentation shows that it works with a value of 512 or bigger for
CARDBUS_IO_SIZE, just not with 256.
Comment 6 Markus schaber 2005-09-08 12:57:55 UTC
I experience what I believe to be the same problem, and have put all diagnostics 
on http://schabi.de/temp/kernelbug/

It boots fine with 2.6.8-2-386 (debian) and 2.6.12.2 and 2.6.12.6 stock kernels, 
but not with 2.6.13.

As the machine does not boot, I can only provide screenshots with a digicam, as 
well as diagnostics output from 2.6.12.6.

You may notice the ipw2200 driver tainting the kernel when studying the dmesg 
output, however it is loaded as a module, and the 2.6.13 seems to hang well 
before booting into the initrd, despite loading any modules.
Comment 7 Markus schaber 2005-09-08 13:24:50 UTC
Sorry, my bug seems to be a different one, as changing the CARDBUS_IO_SIZE does
not seem to help.

Should I try to test all the 2.6.3-rc kernels, or some configure options?

Btw, should I file this as a new bug report?
Comment 8 Larry Finger 2005-09-08 15:02:12 UTC
At 03:24 PM 9/8/2005, you wrote:



>------- Additional Comments From kernelbugs@schabi.de  2005-09-08 
>13:24 -------
>Sorry, my bug seems to be a different one, as changing the 
>CARDBUS_IO_SIZE does
>not seem to help.
>
>Should I try to test all the 2.6.3-rc kernels, or some configure options?
>
>Btw, should I file this as a new bug report?
>
>------- You are receiving this mail because: -------
>You reported the bug, or are watching the reporter.

You should definitely file a separate bug. Before you do, try 
2.6.13-rc7 to see if it makes a difference. 

Comment 9 Markus schaber 2005-09-09 02:47:51 UTC
Thanks, I'll do so this night when I have access to that machine again.

Comment 10 Markus schaber 2005-09-10 03:11:11 UTC
Did so: http://bugzilla.kernel.org/show_bug.cgi?id=5216
Thanks,
Markus