Bug 4737 - cardbus fault to start
Summary: cardbus fault to start
Status: CLOSED CODE_FIX
Alias: None
Product: Drivers
Classification: Unclassified
Component: PCI (show other bugs)
Hardware: i386 Linux
: P2 high
Assignee: Dominik Brodowski
URL:
Keywords:
: 4766 4775 4783 (view as bug list)
Depends on:
Blocks: 4766
  Show dependency tree
 
Reported: 2005-06-12 01:19 UTC by Oleksij Rempel (fishor)
Modified: 2005-06-30 13:52 UTC (History)
3 users (show)

See Also:
Kernel Version: 2.6.12-rc6-mm1
Subsystem:
Regression: ---
Bisected commit-id:


Attachments
dmesg kernel 2.6.12-rc6 (11.93 KB, text/plain)
2005-06-12 01:20 UTC, Oleksij Rempel (fishor)
Details
dmesg kernel 2.6.12-rc6 (12.50 KB, text/plain)
2005-06-12 01:27 UTC, Oleksij Rempel (fishor)
Details
dmesg-2.6.12-mm2+patch (13.03 KB, text/plain)
2005-06-26 10:35 UTC, Oleksij Rempel (fishor)
Details

Description Oleksij Rempel (fishor) 2005-06-12 01:19:32 UTC
Distribution: Debian Sarge  
Hardware Environment: laptop samsung nv5000 ; CardBus bridge: Texas Instruments PCI1410 
Software Environment: kernel 2.6.12-rc6-mm1  
Problem Description:  
 
 syslog output kernel 2.6.12-rc6: 
Jun 12 10:08:36 lopi cardmgr[3333]: watching 1 socket 
Jun 12 10:08:36 lopi kernel: cs: IO port probe 0x100-0x4ff: excluding 0x200-0x207 0x398-0x39f 
0x4d0-0x4d7 
Jun 12 10:08:36 lopi kernel: cs: IO port probe 0x800-0x8ff: clean. 
Jun 12 10:08:36 lopi kernel: cs: IO port probe 0xc00-0xcff: clean. 
Jun 12 10:08:36 lopi kernel: cs: IO port probe 0xa00-0xaff: clean. 
Jun 12 10:08:36 lopi cardmgr[3334]: starting, version is 3.2.5 
Jun 12 10:08:37 lopi authdaemond.plain: modules="authpam", daemons=5 
 
  
 syslog output kernel 2.6.12-rc6-mm1: 
Jun 12 10:03:39 lopi cardmgr[3374]: watching 1 socket 
Jun 12 10:03:39 lopi cardmgr[3374]: could not adjust resource: IO ports 0x100-0x4ff: Input/output 
error 
Jun 12 10:03:39 lopi cardmgr[3374]: could not adjust resource: IO ports 0x800-0x8ff: Input/output 
error 
Jun 12 10:03:39 lopi cardmgr[3374]: could not adjust resource: IO ports 0xc00-0xcff: Input/output 
error 
Jun 12 10:03:39 lopi cardmgr[3374]: could not adjust resource: memory 0xc0000-0xfffff: 
Input/output error 
Jun 12 10:03:39 lopi cardmgr[3374]: could not adjust resource: memory 0xa0000000-0xa0ffffff: 
Input/output error 
Jun 12 10:03:39 lopi cardmgr[3374]: could not adjust resource: memory 0x60000000-0x60ffffff: 
Input/output error 
Jun 12 10:03:39 lopi cardmgr[3374]: could not adjust resource: IO ports 0xa00-0xaff: Input/output 
error 
Jun 12 10:03:39 lopi cardmgr[3375]: starting, version is 3.2.5 
Jun 12 10:03:39 lopi authdaemond.plain: modules="authpam", daemons=5
Comment 1 Oleksij Rempel (fishor) 2005-06-12 01:20:34 UTC
Created attachment 5150 [details]
dmesg kernel 2.6.12-rc6
Comment 2 Dominik Brodowski 2005-06-12 01:25:26 UTC
There's a bug in 2.6.12-rc6-mm1 PCI code which should be the cause of it. Do I
assume correctly that 2.6.12-rc6 works and 2.6.12-rc5-mm also worked?
Comment 3 Oleksij Rempel (fishor) 2005-06-12 01:27:21 UTC
Created attachment 5151 [details]
dmesg kernel 2.6.12-rc6
Comment 4 Dominik Brodowski 2005-06-12 14:02:43 UTC
Just reproduced it on my own notebook -- some changes to the PCI resource
handling break allocation of such resources outside a very specific window. Even
for "transparent" parent bridges.
Comment 5 Andrew Morton 2005-06-17 00:55:54 UTC
Greg, do we have post-2.6.12-rc6-mm1 fixes which might
make this go away?
Comment 6 Greg Kroah-Hartman 2005-06-17 15:13:26 UTC
Dominik, what patch in the -mm tree do you think caused this problem?

Comment 7 Dominik Brodowski 2005-06-18 06:14:36 UTC
AFAIK it's one of
+gregkh-pci-pci-fix-routing-in-parent-bridge.patch
+gregkh-pci-pci-dma-bursting-advice.patch
+gregkh-pci-pci-collect-host-bridge-resources-01.patch
+gregkh-pci-pci-collect-host-bridge-resources-02.patch
Comment 8 Oleksij Rempel (fishor) 2005-06-18 10:53:53 UTC
tested! 
 AFAIK is 
+gregkh-pci-pci-collect-host-bridge-resources-01.patch  
 
 
cs: IO port probe 0x100-0x4ff: excluding 0x200-0x207 0x398-0x39f 0x4d0-0x4d7 
cs: IO port probe 0x800-0x8ff: clean. 
cs: IO port probe 0xc00-0xcff: clean. 
cs: IO port probe 0xa00-0xaff: clean. 
 
 
Comment 9 Greg Kroah-Hartman 2005-06-21 16:20:01 UTC
The 01 patch?  Are you sure it's not this one instead:
http://www.kernel.org/pub/linux/kernel/people/gregkh/gregkh-2.6/patches/pci/pci-collect-host-bridge-resources-02.patch

That one looks like it could mess people up, not the 01 patch, which only
increases the number of busses we look at.
Comment 10 Greg Kroah-Hartman 2005-06-21 16:31:21 UTC
*** Bug 4775 has been marked as a duplicate of this bug. ***
Comment 11 Greg Kroah-Hartman 2005-06-21 17:27:05 UTC
*** Bug 4766 has been marked as a duplicate of this bug. ***
Comment 12 Oleksij Rempel (fishor) 2005-06-22 05:20:34 UTC
I'm not sure but, if i set "PCI_BUS_NUM_RESOURCES" to 4 then ewery thing working. I don't know 
how about kernel 2.6.12-mm1 wile this one crash on my laptop. 
Comment 13 Grant Coady 2005-06-22 16:55:28 UTC
Applied patch from:
http://www.uwsg.iu.edu/hypermail/linux/kernel/0506.2/1704.html
And it works for CardBus 32-bit NIC

16-bit PCMCIA still broken :(
Several MB of 
Jun 23 09:43:14 tosh kernel:  0xae900000-0xaf0fffff<4>iounmap: bad address ccaa0000
Jun 23 09:43:14 tosh kernel: iounmap: bad address ccaa0000
in syslog.
Comment 14 Oleksij Rempel (fishor) 2005-06-24 00:01:43 UTC
I tryd kernel 2.6.12-rc6-mm1 with acpi=off , and ist working! 
Comment 15 Dominik Brodowski 2005-06-26 07:17:59 UTC
Please test out -mm2 with http://lkml.org/lkml/diff/2005/6/26/47/1 on top of it.
Comment 16 Oleksij Rempel (fishor) 2005-06-26 10:34:31 UTC
*** Bug 4783 has been marked as a duplicate of this bug. ***
Comment 17 Oleksij Rempel (fishor) 2005-06-26 10:35:58 UTC
Created attachment 5216 [details]
dmesg-2.6.12-mm2+patch
Comment 18 Oleksij Rempel (fishor) 2005-06-26 10:39:25 UTC
kernel 2.6.12-mm2 and 2.6.12-mm2+patch are tested. Patched kernel working properly. Row -mm2  
kernel hang on bootup.  
Comment 19 Oleksij Rempel (fishor) 2005-06-30 08:52:30 UTC
With kernel 2.6.13-rc1 pcmcia working properly. Thenk you  

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