linux-mips
[Top] [All Lists]

Re: CardBus on DBAu1500 -- IRQ storm?

To: linux-mips <linux-mips@linux-mips.org>
Subject: Re: CardBus on DBAu1500 -- IRQ storm?
From: Jeff Baitis <baitisj@evolution.com>
Date: Fri, 7 Mar 2003 18:08:04 -0800
In-reply-to: <20030303181421.C20129@luca.pas.lab>; from baitisj@evolution.com on Mon, Mar 03, 2003 at 06:14:21PM -0800
Original-recipient: rfc822;linux-mips@linux-mips.org
References: <20030228194820.Z20129@luca.pas.lab> <1046499358.12356.2.camel@adsl.pacbell.net> <20030303181421.C20129@luca.pas.lab>
Reply-to: baitisj@evolution.com
Sender: linux-mips-bounce@linux-mips.org
User-agent: Mutt/1.2.5i
I belive I've made some progress tracking down this problem. The yenta_socket
driver launches the yenta_open_bh thread. Earlier, all CardBus and EXCA
interrupts were masked out.  Near line 650, yenta_open_bh tries to allocate an
interrupt line:

    if (!(socket->cb_irq) || request_irq(socket->cb_irq, yenta_interrupt,
                             SA_SHIRQ, socket->dev->name, socket)) {

request_irq, from linux/arch/mips/kernel/irq.c, enters a critical
section around line 873. startup() is invoked on the IRQ descriptor,
and then the critical section is exited with a call to 
spin_unlock_irqrestore(&desc->lock,flags);

As soon as this call is made, the kernel calls yenta_interrupt over and over.
Pardon my inexperience, but is this called an 'IRQ storm?'

yenta_interrupt: Entering    
yenta_events: Entering   
cb_readl: c0113eb0 0000 00000000
cb_writel: c0113eb0 0000 00000000
exca_readb: c0113eb0 0004 00     
exca_readb: c0113eb0 0003 00
yenta_events: Exiting (retval is 0)
yenta_interrupt: Exiting           
yenta_interrupt: Entering

etc etc

I've verified using our PCI analyzer that no interrupt events are transmitted
by the CardBus bridge, so this seems to be a software problem.

I'm going to continue to hunt this down, on the assumption that the kernel's
interrupt table is getting incorrect / corrupt data.  Suggestions are always
appreciated.

May all your CardBuses be happy!

-Jeff



On Mon, Mar 03, 2003 at 06:14:21PM -0800, Jeff Baitis wrote:
> Pete and others:
> 
> Thanks! I can access CardBus-enabled cards... but only if I plug the CardBus
> card into a PCI bridge device. 
> 
> For those who might be interested in hearing, here's my current status:
> 
> I tested the PCI->PCI bridge code by copying pb1500/pci_ops.c into
> db1x00/pci_ops.c. It seemed to work. Great!
> 
> After noticing that yenta_socket driver still locked up, I plugged a PCI-PCI
> bridge board in between the Au1500 dev board and the CardBus bridge. CardBus
> didn't work quite right, so I checked the PCI configuration registers on all
> devices on my PCI bus.
> 
> I noticed the PCI bridge device's SUBORDINATE_BUS value was set to 01, which
> is completely incorrect. It should have been at *least* 11, given that
> the CardBus bridge's SECONDARY_BUS register was set to 10. 
> 
> So, I used setpci to change the PCI bridge's SUBORDINATE_BUS to 1f. I started
> the CardBus services, and it seems that everything worked with CardBus cards.
> Eject, insert, and status all good -- and I can look at all the PCI
> configuration registers on the CardBus card.
> 
> ***
> 
> It seems like the issue with plugging in the CardBus bridge directly into
> PCI bus 0 (no PCI-PCI bridge) has to do with 16-bit CardBus support.
> I enabled debugging on yenta_socket, and here's some results. First, I present
> information *without* the PCI bridge, and then *with* PCI bridge.
> 
> 
> Without PCI bridge:
> 
> dmesg gives:
> 
>   Autoconfig PCI channel 0x8029fc38°°
>     Scanning bus 00, I/O 0x00000300:0x00100000, Mem 0x40000000:0x44000000
>     00:0c.0 Class 0104: 1103:0007 (rev 01)°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°     
>                                                            
>             I/O at 0x00000300 [size=0x8]°°                         
>             I/O at 0x00000308 [size=0x4]
>             I/O at 0x00000310 [size=0x8]
>             I/O at 0x00000318 [size=0x4]
>             I/O at 0x00000400 [size=0x100]
>     00:0d.0 Class 0607: 104c:ac56°°°°°°°°°                         
>             Mem at 0x40000000 [size=0x1000]                        
>             Mem unavailable -- skipping°°°°                        
>             I/O at 0x00000500 [size=0x4]                           
>             Mem at 0x40001000 [size=0x1000]                        
>             Mem at 0x40002000 [size=0x1000]                        
>             Mem at 0x40003000 [size=0x1000]                        
>   
> lspci -v gives:
>   
>   00:0c.0 RAID bus controller: Triones Technologies, Inc.: Unknown device 0007
>   (rev 01)
>           Subsystem: Triones Technologies, Inc.: Unknown device 0001
>           Flags: bus master, 66Mhz, medium devsel, latency 128, IRQ 1
>           I/O ports at 0300 [size=8]
>           I/O ports at 0308 [size=4]
>           I/O ports at 0310 [size=8]
>           I/O ports at 0318 [size=4]
>           I/O ports at 0400 [size=256]
>           Expansion ROM at <unassigned> [disabled] [size=128K]   
>           Capabilities: [60] Power Management version 2          
>   
>   00:0d.0 CardBus bridge: Texas Instruments: Unknown device ac56 
>           Subsystem: Unknown device 5678:1234                    
>           Flags: bus master, medium devsel, latency 128, IRQ 1   
>           Memory at 40000000 (32-bit, non-prefetchable) [size=4K]
>           Bus: primary=00, secondary=05, subordinate=00, sec-latency=0      
>           Memory window 0: 40001000-40002000 (prefetchable)      
>           I/O window 0: 00000000-00000003                        
>           I/O window 1: 00000000-00000003                        
>           16-bit legacy interface ports at 0001                  
> 
> 
> root@10.1.1.154:~# modprobe yenta_socket                       
> 
>   Linux Kernel Card Services 3.1.22
>     options:  [pci] [cardbus]°°°°°°
>   config_writel: c0112bd0 0044 00000000
>   config_writel: c0112bd0 0010 40000000
>   config_writew: c0112bd0 0004 0087°°°°
>   config_writeb: c0112bd0 000c 08°°
>   config_writeb: c0112bd0 000d a8
>   config_writel: c0112bd0 0018 b0000500
>   config_readw: c0112bd0 003e 0340°°°°°
>   config_writew: c0112bd0 003e 0580
>   exca_writeb: c0112bd0 001e 00°°°°
>   exca_writeb: c0112bd0 0016 00
>   cb_writel: c0112bd0 000c 00004000
>   cb_writel: c0112bd0 0004 00000000
>   config_readl: c0112bd0 001c 40001000
>   config_readl: c0112bd0 0020 40002000
>   config_readl: c0112bd0 0024 40003000
>   config_readl: c0112bd0 0028 00000000
>   config_writel: c0112bd0 0024 40400000
>   config_writel: c0112bd0 0028 407fffff
>   config_readl: c0112bd0 002c 00000000°
>   config_readl: c0112bd0 0030 00000000
>   config_writel: c0112bd0 002c 00004000
>   config_writel: c0112bd0 0030 000040ff
>   config_readl: c0112bd0 0034 00000000°
>   config_readl: c0112bd0 0038 00000000
>   config_writel: c0112bd0 0034 00004400
>   config_writel: c0112bd0 0038 000044ff
>   cb_readl: c0112bd0 0000 00000006°°°°°
>   cb_writel: c0112bd0 0000 00000006
>   exca_readb: c0112bd0 0004 00°°°°°
>   exca_readb: c0112bd0 0003 00
>   cb_readl: c0112bd0 0000 00000000
>   cb_writel: c0112bd0 0000 00000000
>   exca_readb: c0112bd0 0004 00°°°°°
>   exca_readb: c0112bd0 0003 00
>   cb_readl: c0112bd0 0000 00000000
>   cb_writel: c0112bd0 0000 00000000
>   exca_readb: c0112bd0 0004 00°°°°°
>   exca_readb: c0112bd0 0003 00
>   cb_readl: c0112bd0 0000 00000000
>   cb_writel: c0112bd0 0000 00000000
>   exca_readb: c0112bd0 0004 00°°°°°
>   exca_readb: c0112bd0 0003 00
>   cb_readl: c0112bd0 0000 00000000
>   cb_writel: c0112bd0 0000 00000000
>   exca_readb: c0112bd0 0004 00°°°°°
>   exca_readb: c0112bd0 0003 00
>   cb_readl: c0112bd0 0000 00000000
>   cb_writel: c0112bd0 0000 00000000
>   exca_readb: c0112bd0 0004 00°°°°°
>   exca_readb: c0112bd0 0003 00
>     ** following portion loops indefinitely **
>   cb_readl: c0112bd0 0000 00000000
>   cb_writel: c0112bd0 0000 00000000
>   exca_readb: c0112bd0 0004 00°°°°°
>   exca_readb: c0112bd0 0003 00
> 
> With PCI bridge:
> 
>   Autoconfig PCI channel 0x8029fc38°°
>   Scanning bus 00, I/O 0x00000300:0x00100000, Mem 0x40000000:0x44000000
>   00:0c.0 Class 0104: 1103:0007 (rev 01)°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°
>           I/O at 0x00000300 [size=0x8]°°
>           I/O at 0x00000308 [size=0x4]
>           I/O at 0x00000310 [size=0x8]
>           I/O at 0x00000318 [size=0x4]
>           I/O at 0x00000400 [size=0x100]
>   00:0d.0 Class 0604: 1011:0022 (rev 02)
>           Bridge: primary=00, secondary=01
>   Scanning sub bus 01, I/O 0x00001000, Mem 0x40000000
>   01:07.0 Class 0607: 104c:ac56°°°°°°°°°°°°°°°°°°°°°°
>           Mem at 0x40000000 [size=0x1000]
>           Mem unavailable -- skipping°°°°
>           I/O at 0x00001000 [size=0x4]
>           Mem at 0x40001000 [size=0x1000]
>           Mem at 0x40002000 [size=0x1000]
>           Mem at 0x40003000 [size=0x1000]
>   Back to bus 00°°°°°°°°°°°°°°°°°°°°°°°°°
> 
> lspci -v gives:
>   
>   00:0c.0 RAID bus controller: Triones Technologies, Inc.: Unknown device 0007
>   (rev 01)
>           Subsystem: Triones Technologies, Inc.: Unknown device 0001
>           Flags: bus master, 66Mhz, medium devsel, latency 128, IRQ 1
>           I/O ports at 0300 [size=8]
>           I/O ports at 0308 [size=4]
>           I/O ports at 0310 [size=8]
>           I/O ports at 0318 [size=4]
>           I/O ports at 0400 [size=256]
>           Expansion ROM at <unassigned> [disabled] [size=128K]
>           Capabilities: [60] Power Management version 2
>   
>   00:0d.0 PCI bridge: Digital Equipment Corporation DECchip 21150 (rev 02)
>   (prog-if 00 [Normal decode])
>           Flags: bus master, medium devsel, latency 0
>           Bus: primary=00, secondary=01, subordinate=1f, sec-latency=0
>           I/O behind bridge: 00001000-00001fff     
>           Memory behind bridge: 40000000-400fffff  
>           Prefetchable memory behind bridge: 0000000000000000-0000000000000000
>   
>   01:07.0 CardBus bridge: Texas Instruments: Unknown device ac56
>           Subsystem: Unknown device 5678:1234      
>           Flags: bus master, medium devsel, latency 128, IRQ 255
>           Memory at 40000000 (32-bit, non-prefetchable) [size=4K]
>           Bus: primary=00, secondary=10, subordinate=00, sec-latency=0
>           Memory window 0: 40001000-40002000 (prefetchable)
>           I/O window 0: 00000000-00000003
>           I/O window 1: 00000000-00000003
>           16-bit legacy interface ports at 0001    
> 
> root@10.1.1.154:~# modprobe yenta_socket                       
> 
>   Linux Kernel Card Services 3.1.22
>     options:  [pci] [cardbus]°°°°°°
>   config_writel: c0112bd0 0044 00000000
>   config_writel: c0112bd0 0010 40000000
>   config_writew: c0112bd0 0004 0087°°°°
>   config_writeb: c0112bd0 000c 08°°
>   config_writeb: c0112bd0 000d a8
>   config_writel: c0112bd0 0018 b0001000
>   config_readw: c0112bd0 003e 0340°°°°°
>   config_writew: c0112bd0 003e 0580
>   exca_writeb: c0112bd0 001e 00°°°°
>   exca_writeb: c0112bd0 0016 00
>   cb_writel: c0112bd0 000c 00004000
>   cb_writel: c0112bd0 0004 00000000
>   config_readl: c0112bd0 001c 40001000
>   config_readl: c0112bd0 0020 40002000
>   config_readl: c0112bd0 0024 40003000
>   config_readl: c0112bd0 0028 00000000
>   config_writel: c0112bd0 0024 10000000
>   config_writel: c0112bd0 0028 103fffff
>   config_readl: c0112bd0 002c 00000000°
>   config_readl: c0112bd0 0030 00000000
>   config_writel: c0112bd0 002c 00004000
>   config_writel: c0112bd0 0030 000040ff
>   config_readl: c0112bd0 0034 00000000°
>   config_readl: c0112bd0 0038 00000000
>   config_writel: c0112bd0 0034 00004400
>   config_writel: c0112bd0 0038 000044ff
>   config_readw: c0112bd0 003e 05c0°°°°°
>   cb_writel: c0112bd0 0000 ffffffff
>   cb_writel: c0112bd0 0004 00000001
>   exca_writeb: c0112bd0 0005 00°°°°
>   warning: end_irq 60 did not enable (6)
>   warning: end_irq 59 did not enable (6)
>   warning: end_irq 58 did not enable (6)
>   warning: end_irq 57 did not enable (6)
>   warning: end_irq 55 did not enable (6)
>   warning: end_irq 52 did not enable (6)
>   warning: end_irq 48 did not enable (6)
>   warning: end_irq 46 did not enable (6)
>   warning: end_irq 45 did not enable (6)
>   warning: end_irq 44 did not enable (6)
>   warning: end_irq 43 did not enable (6)
>   warning: end_irq 42 did not enable (6)
>   warning: end_irq 41 did not enable (6)
>   warning: end_irq 40 did not enable (6)
>   warning: end_irq 39 did not enable (6)
>   warning: end_irq 38 did not enable (6)
>   warning: end_irq 31 did not enable (6)
>   warning: end_irq 27 did not enable (6)
>   warning: end_irq 18 did not enable (6)
>   warning: end_irq 14 did not enable (6)
>   warning: end_irq 5 did not enable (6)°
>   warning: end_irq 4 did not enable (6)
>   warning: end_irq 2 did not enable (6)
>   warning: end_irq 1 did not enable (6)
>   warning: end_irq 60 did not enable (16)
>   warning: end_irq 59 did not enable (16)
>   warning: end_irq 58 did not enable (16)
>   warning: end_irq 57 did not enable (16)
>   warning: end_irq 55 did not enable (16)
>   warning: end_irq 52 did not enable (16)
>   warning: end_irq 48 did not enable (16)
>   warning: end_irq 46 did not enable (16)
>   warning: end_irq 45 did not enable (16)
>   warning: end_irq 44 did not enable (16)
>   warning: end_irq 43 did not enable (16)
>   warning: end_irq 42 did not enable (16)
>   warning: end_irq 41 did not enable (16)
>   warning: end_irq 40 did not enable (16)
>   warning: end_irq 39 did not enable (16)
>   warning: end_irq 38 did not enable (16)
>   warning: end_irq 27 did not enable (16)
>   warning: end_irq 18 did not enable (16)
>   warning: end_irq 5 did not enable (16)°
>   warning: end_irq 4 did not enable (16)
>   warning: end_irq 2 did not enable (16)
>   warning: end_irq 1 did not enable (16)
>   exca_writeb: c0112bd0 0005 31°°°°°°°°°
>   cb_writel: c0112bd0 000c 00000001
>   cb_writel: c0112bd0 0000 ffffffff
>   exca_writeb: c0112bd0 0005 91°°°°
>   cb_writel: c0112bd0 000c 00000001
>   cb_writel: c0112bd0 0000 ffffffff
>   exca_writeb: c0112bd0 0005 a1°°°°
>   cb_writel: c0112bd0 000c 00000001
>   cb_writel: c0112bd0 0000 ffffffff
>   exca_writeb: c0112bd0 0005 b1°°°°
>   cb_writel: c0112bd0 000c 00000001
>   cb_writel: c0112bd0 0000 ffffffff
>   cb_writel: c0112bd0 0004 00000000
>   exca_writeb: c0112bd0 0005 00°°°°
>   config_writew: c0112bd0 003e 0540
>   Yenta IRQ list 0000, PCI irq0°°°°
>   cb_readl: c0112bd0 0008 30000020
>   Socket status: 30000020°°°°°°°°°
>   config_writel: c0112bd0 0044 00000000
>   config_writel: c0112bd0 0010 40000000
>   config_writew: c0112bd0 0004 0087°°°°
>   config_writeb: c0112bd0 000c 08°°
>   config_writeb: c0112bd0 000d a8
>   config_writel: c0112bd0 0018 b0001000
>   config_readw: c0112bd0 003e 0540°°°°°
>   config_writew: c0112bd0 003e 0580
>   exca_writeb: c0112bd0 001e 00°°°°
>   exca_writeb: c0112bd0 0016 00
>   cb_writel: c0112bd0 000c 00004000
>   cb_readl: c0112bd0 0010 00000400°
>   cb_writel: c0112bd0 0010 00000000
>   config_readw: c0112bd0 003e 05c0°
>   cb_readl: c0112bd0 0008 30000820
>   exca_readb: c0112bd0 0003 00°°°°
>   exca_writeb: c0112bd0 0003 00
>   config_writew: c0112bd0 003e 0580
>   cb_writel: c0112bd0 0000 ffffffff
>   cb_writel: c0112bd0 0004 00000006
>   exca_readb: c0112bd0 0006 00°°°°°
>   exca_writew: c0112bd0 0008 0000
>   exca_writew: c0112bd0 000a 0001
>   exca_readb: c0112bd0 0007 00°°°
>   exca_writeb: c0112bd0 0007 00
>   exca_readb: c0112bd0 0006 00°
>   exca_writew: c0112bd0 000c 0000
>   exca_writew: c0112bd0 000e 0001
>   exca_readb: c0112bd0 0007 00°°°
>   exca_writeb: c0112bd0 0007 00
>   exca_readb: c0112bd0 0006 00°
>   exca_writeb: c0112bd0 0040 00
>   exca_writew: c0112bd0 0010 0000
>   exca_writew: c0112bd0 0012 0000
>   exca_writew: c0112bd0 0014 0000
>   ..... etc etc 
>   
> Thanks, all, for the information. I hope someone finds this useful!
> I'm gonna get remote kgdb goin' here, so I can hopefully fix the obnoxious
> infinite loop...
> 
> -Jeff
> 
> 
> On Fri, Feb 28, 2003 at 10:15:58PM -0800, Pete Popov wrote:
> > On Fri, 2003-02-28 at 19:48, Jeff Baitis wrote:
> > > Hey Pete and others!
> > > 
> > > I'm finally working on CardBus support on the DBAu1500. Just got 
> > > acquainted
> > > with PCI today. :)
> > > 
> > > I decided that the first step is to plug in a PCI->PCI bridge, and try to 
> > > see
> > > if it would work, which it did not. After winding around inside of the 
> > > kernel,
> > > I finally arrived in arch/mips/au1000/db1x00/pci_ops.c:
> > > 
> > > Inside of config_access(unsigned char access_type, struct pci_dev *dev,
> > > unsigned char where, u32 * data), on line 97, a little surprise:
> > > 
> > >     if (bus != 0) {
> > >         *data = 0xffffffff;
> > >         return -1;
> > >     }
> > > 
> > > At this point, I concluded that I cannot traverse a PCI-PCI or CardBus 
> > > bridge,
> > > since any devices behind the bridge will require Type 1 Configuration 
> > > Cycles,
> > > and it seems that only Type 0 is currently supported.
> > > 
> > > I assume that I should add code to handle the case where I need to 
> > > generate
> > > Type 1 Configuration Cycles inside of config_access. Pete, since you 
> > > authored
> > > this code, I thought I'd quickly run this by you to make sure that I'm on
> > > track.
> > > 
> > > Thanks for your suggestions!
> > 
> > Take a look at arch/mips/au1000/pb1500/pci_ops.c for type 1 config
> > access. The patch was courtesy of David Gathright and apparently I
> > missed adding it in the db1500. Actually, we need to combine that code
> > because it's the same. Let me get through my eternal struggle of getting
> > the 36 bit patch applied (with Ralf's help I think a modified patch
> > should be ready this weekend) and then I'll worry about clean ups :)
> > 
> > Pete
> > 
> > 
> 
> -- 
>          Jeffrey Baitis - Associate Software Engineer
> 
>                     Evolution Robotics, Inc.
>                      130 West Union Street
>                        Pasadena CA 91103
> 
>  tel: 626.535.2776  |  fax: 626.535.2777  |  baitisj@evolution.com 
> 
> 

-- 
         Jeffrey Baitis - Associate Software Engineer

                    Evolution Robotics, Inc.
                     130 West Union Street
                       Pasadena CA 91103

 tel: 626.535.2776  |  fax: 626.535.2777  |  baitisj@evolution.com 


<Prev in Thread] Current Thread [Next in Thread>