Bug 5428

Summary: echi_hcd hangs machine
Product: Drivers Reporter: Borislav Petkov (bp)
Component: USBAssignee: David Brownell (dbrownell)
Status: REJECTED UNREPRODUCIBLE    
Severity: high CC: akpm, betelgeuse, greg, stern
Priority: P2    
Hardware: i386   
OS: Linux   
Kernel Version: 2.6.14-rc4 Subsystem:
Regression: --- Bisected commit-id:
Bug Depends on:    
Bug Blocks: 5089    
Attachments: early ehci irq disable

Description Borislav Petkov 2005-10-12 11:44:33 UTC 
Most recent kernel where this bug did not occur:2.6.14-rc4
Distribution:Debian unstable
Hardware Environment:i386
Software Environment:
Linux zmei 2.6.13 #1 SMP Mon Oct 3 14:58:00 CEST 2005 i686 GNU/Linux

Gnu C                  4.0.2
Gnu make               3.80
binutils               2.16.91
util-linux             2.12p
mount                  2.12p
module-init-tools      3.2-pre9
e2fsprogs              1.38
reiserfsprogs          line
reiser4progs           line
PPP                    2.4.3
nfs-utils              1.0.7
Linux C Library        2.3.5
Dynamic linker (ldd)   2.3.5
Procps                 3.2.5
Net-tools              1.60
Console-tools          0.2.3
Sh-utils               5.2.1
udev                   070
Modules Loaded         isofs nfsd exportfs nfs lockd sunrpc parport_pc parport
pcspkr tuner tvaudio bttv video_buf firmware_class i2c_algo_bit v4l2_common
btcx_risc tveeprom i2c_core videodev rtc

Problem Description:
When I boot into 2.6.14-rc[3,4] (rc2 not tested yet) the console boot messages
end with:
   [4294687.294000] usb usb3: Manufacturer: Linux 2.6.14-rc4 uhci_hcd
   [4294687.300000] usb usb3: SerialNumber: 0000:00:1d.2
   [4294687.321000] hub 3-0:1.0: USB hub found
   [4294687.326000] hub 3-0:1.0: 2 ports detected
   [4294689.363000] ACPI: PCI Interrupt 0000:00:1d.7[D] -> GSI 23 (level, low) 
-> IRQ 20
   [4294689.372000] ehci_hcd 0000:00:1d.7: EHCI Host Controller
   [4294689.378000] ehci_hcd 0000:00:1d.7: debug port 1

and the machine hangs. USB debugging is enabled but it spits no more than that.
Steps to reproduce:
Comment 1 Andrew Morton 2005-10-12 12:55:38 UTC 
bugme-daemon@kernel-bugs.osdl.org wrote:
>
> Most recent kernel where this bug did not occur:2.6.14-rc4

Could you please correct this?
Comment 2 Borislav Petkov 2005-10-12 21:55:09 UTC 
Most recent kernel where this bug did not occur:None
Comment 3 Greg Kroah-Hartman 2005-10-13 15:49:01 UTC 
*** Bug 5433 has been marked as a duplicate of this bug. ***
Comment 4 Alan Stern 2005-10-13 18:46:22 UTC 
Just out of curiosity, does this happen if you boot with "acpi=off" on the boot
command line?
Comment 5 David Brownell 2005-10-13 19:42:12 UTC 
Created attachment 6294 [details]
early ehci irq disable

This patch tweaks some BIOS-related behavior that can appear during
initialization or with kexec(); it's in the 2.6.15 queue.  It might
explain the problem you're seeing.

You might also try booting with "usb-handoff" set on the kernel
command line.
Comment 6 Borislav Petkov 2005-10-13 22:56:56 UTC 
Test results:

1. booting 14-rc4 with acpi=off doesn't help.
2. patching the kernel so that the ehci controller is halted before reboot
doesn't help alone but...
3. the patched kernel with the "usb-handoff" boot option booted just fine.

Tell me if you need more info.
Comment 7 David Brownell 2005-10-14 03:10:16 UTC 
Looks like your BIOS just demands the usb-handoff to be done 
early; there are some systems that need that.  In fact, enough 
that we'll probably make that the default at some point. 
 
Since you now know about that flag, I'll close this bug.
Comment 8 Borislav Petkov 2005-10-14 04:41:00 UTC 
is this patch going to enter 2.6.14 or it remains in the 2.6.15 queue?
Comment 9 Petteri Räty 2005-10-14 07:31:45 UTC 
Bug 5433 was marked as a duplicate of this bug, but I know for certain that my
machine has worked without problems in the past and as such should not require
any special kernel options. Please consider reopening the bug or making
usb-handoff the default asap. I will be testing if it fixes my problem.
Comment 10 Petteri Räty 2005-10-14 08:42:34 UTC 
I applied the patch and added usb-handoff in the grub menu but it did not solve
the problem.
Comment 11 David Brownell 2005-10-14 18:31:14 UTC 
To Petteri R
Comment 12 David Brownell 2005-10-16 08:08:51 UTC 
To Borislav Petkov ... the patch shouldn't be needed 
if you're using usb-handoff and the only issue you're 
seeing is at boot time.  So it should be no problem 
for you that it won't be merged quite yet.
Comment 13 Borislav Petkov 2005-10-17 01:13:24 UTC 
... however, I booted this morning with usb-handoff and it crashed at the same
point again... maybe useb-handoff with the patch doesn't work OK. Should I
remove the patch and use _only_ usb-handoff as a boot option?
Comment 14 Borislav Petkov 2005-10-22 02:45:09 UTC 
2.6.14-rc5 doesn't boot with 'usb-handoff' and the machine halts at the same
point in the boot process.
Comment 15 Borislav Petkov 2005-10-26 21:30:33 UTC 
correction: the usb-handoff boot option helps only when you reboot. Otherwise,
on initial boot, the machine hangs. For 2.6.14-rc5-mm1 this is also the case.
Comment 16 David Brownell 2005-11-02 11:09:41 UTC 
Closing this again.  (a) 2.6.13 worked, contrary to comment #2,
but most importantly (b) after a BIOS update, 2.6.14 works too.