drivers/ieee1394/sbp2.c:734: error: `host' undeclared (first use in this function) 2.6.3-bk3

drivers/ieee1394/sbp2.c:734: error: `host' undeclared (first use in this function) 2.6.3-bk3

[Bob Gill]

Hi.  The whole error message is (when building 2.6.3-bk3):
drivers/ieee1394/sbp2.c: In function `sbp2_alloc_device':
drivers/ieee1394/sbp2.c:734: error: `host' undeclared (first use in this
function)
drivers/ieee1394/sbp2.c:734: error: (Each undeclared identifier is
reported only once
drivers/ieee1394/sbp2.c:734: error: for each function it appears in.)
make[2]: *** [drivers/ieee1394/sbp2.o] Error 1
make[1]: *** [drivers/ieee1394] Error 2
make: *** [drivers] Error 2

It gives the same message when building with the source from the 1394
subversion tree.  Apparently the declaration for 'host' is not where it
used to be. 

#ifdef CONFIG_IEEE1394_SBP2_PHYS_DMA
                /* Handle data movement if physical dma is not
                 * enabled/supportedon host controller */
                hpsb_register_addrspace(&sbp2_highlevel, host,
&sbp2_physdma_ops,
                                        0x0ULL, 0xfffffffcULL);
#endif

**--------------------------------------------------------------
highlevel.c (which isn't #included in sbp2.c) lists:
struct hl_host_info {
        struct list_head list;
        struct hpsb_host *host;
        size_t size;
        unsigned long key;
        void *data;
};

**(highlevel.h is #included in sbp2.c, highlevel.c isn't, and 
highlevel.h doesn't mention struct hl_host_info). 
**---------------------------------------------------------------
** The above information is intended merely for recreational value
   and may provide no practical information with respect to 
   resolving the problem as I have never been accused of being a
   systems programmer!  :)

If you need more info, please mail me as I'm not on the list.  TIA,

Bob

Re: drivers/ieee1394/sbp2.c:734: error: `host' undeclared (first use in this function) 2.6.3-bk3

[Ben Collins]

On Sat, Feb 21, 2004 at 02:36:43PM -0700, Bob Gill wrote:
> Hi.  The whole error message is (when building 2.6.3-bk3):
> drivers/ieee1394/sbp2.c: In function `sbp2_alloc_device':
> drivers/ieee1394/sbp2.c:734: error: `host' undeclared (first use in this
> function)
> drivers/ieee1394/sbp2.c:734: error: (Each undeclared identifier is
> reported only once
> drivers/ieee1394/sbp2.c:734: error: for each function it appears in.)
> make[2]: *** [drivers/ieee1394/sbp2.o] Error 1
> make[1]: *** [drivers/ieee1394] Error 2
> make: *** [drivers] Error 2

Disable CONFIG_IEEE1394_SBP2_PHYS_DMA. I'll have this fixed next time I
sync with Linus. Really, CONFIG_IEEE1394_SBP2_PHYS_DMA is for debugging
only, and you shouldn't have it enabled.

-- 
Debian     - http://www.debian.org/
Linux 1394 - http://www.linux1394.org/
Subversion - http://subversion.tigris.org/
WatchGuard - http://www.watchguard.com/

Re: drivers/ieee1394/sbp2.c:734: error: `host' undeclared (first use in this function) 2.6.3-bk3

[Dave Jones]

On Sat, Feb 21, 2004 at 02:36:43PM -0700, Bob Gill wrote:
 > Hi.  The whole error message is (when building 2.6.3-bk3):
 > drivers/ieee1394/sbp2.c: In function `sbp2_alloc_device':
 > drivers/ieee1394/sbp2.c:734: error: `host' undeclared (first use in this
 > function)
 > drivers/ieee1394/sbp2.c:734: error: (Each undeclared identifier is
 > reported only once
 > drivers/ieee1394/sbp2.c:734: error: for each function it appears in.)
 > make[2]: *** [drivers/ieee1394/sbp2.o] Error 1
 > make[1]: *** [drivers/ieee1394] Error 2
 > make: *** [drivers] Error 2

I've no hardware to test this, but does this do the right thing for you ?

		Dave

--- linux-2.6.3/drivers/ieee1394/sbp2.c~	2004-02-22 00:54:51.000000000 +0000
+++ linux-2.6.3/drivers/ieee1394/sbp2.c	2004-02-22 00:55:11.000000000 +0000
@@ -727,7 +727,7 @@
 #ifdef CONFIG_IEEE1394_SBP2_PHYS_DMA
 		/* Handle data movement if physical dma is not
 		 * enabled/supportedon host controller */
-		hpsb_register_addrspace(&sbp2_highlevel, host, &sbp2_physdma_ops,
+		hpsb_register_addrspace(&sbp2_highlevel, hi->host, &sbp2_physdma_ops,
 					0x0ULL, 0xfffffffcULL);
 #endif
 	}

Re: drivers/ieee1394/sbp2.c:734: error: `host' undeclared (first use in this function) 2.6.3-bk3

[Bob Gill]

Ok.  I rebuilt the kernel 3 times.  I built it twice with the following
change in sbp2.c:727
#ifdef CONFIG_IEEE1394_SBP2_PHYS_DMA
                /* Handle data movement if physical dma is not
                 * enabled/supportedon host controller */
                        hpsb_register_addrspace(&sbp2_highlevel,
hi->host, &sbp2_physdma_ops,
                                        0x0ULL, 0xfffffffcULL);
#endif

(twice to confirm that the kernel builds without throwing any errors). 
It builds sbp2.  I then removed the patch and built with my build script
set as:
# CONFIG_IEEE1394_SBP2_PHYS_DMA is not set

----------------------------------------------
both build all of the 1394 modules.  Unfortunately, neither change gives
me devices in /proc/scsi/scsi (or /proc/partitions) with 2.6,3-bk3.  
/var/log/messages doesn't show much about what is going on, but dmesg
shows:
kudzu: numerical sysctl 1 23 is obsolete.
SCSI subsystem initialized
ohci1394: $Rev: 1131 $ Ben Collins <bcollins@debian.org>
ohci1394: fw-host0: OHCI-1394 1.0 (PCI): IRQ=[9] 
MMIO=[eb005000-eb0057ff]  Max
Packet=[2048]
kudzu: numerical sysctl 1 49 is obsolete.
ip_conntrack version 2.1 (8191 buckets, 65528 max) - 296 bytes per
conntrack
ieee1394: Node added: ID:BUS[0-00:1023]  GUID[080046010203ab38]
ieee1394: Node added: ID:BUS[0-01:1023]  GUID[0030e000e0000b39]
ieee1394: Node added: ID:BUS[0-02:1023]  GUID[0030e000e0000ae3]
ieee1394: Host added: ID:BUS[0-03:1023]  GUID[001106001a250051]
eth0: Media Link On 10mbps half-duplex
raw1394: /dev/raw1394 device initialized
video1394: Installed video1394 module

------------a cut and paste of what I get---------------
[root@localhost root]# cat /proc/scsi/scsi
Attached devices:
[root@localhost root]#
[root@localhost root]# lsmod | grep ieee
ieee1394               88628  4 video1394,dv1394,raw1394,ohci1394
[root@localhost root]#
[root@localhost root]# cat /proc/partitions
major minor  #blocks  name
                                                                                
   3     0   20010312 hda
   3     1      30208 hda1
   3     2    1050840 hda2
   3     3   18922680 hda3
   3    64   80043264 hdb
   3    65   80035798 hdb1
[root@localhost root]#
[root@localhost udev]# ls /udev/sd*
/udev/sda  /udev/sda1  /udev/sdb  /udev/sdb1  /udev/sdc  /udev/sdc1
[root@localhost udev]#
[root@localhost udev]# od /udev/sda
od: /udev/sda: No such device or address
[root@localhost udev]#
[root@localhost devices]# ls /sys/bus/ieee1394/devices
001106001a250051  0030e000e0000b39  080046010203ab38-0
0030e000e0000ae3  080046010203ab38  fw-host0
[root@localhost devices]#
[root@localhost fw-host0]# cat
/sys/bus/ieee1394/devices/fw-host0/node_count
4
[root@localhost fw-host0]#
[root@localhost fw-host0]# cat
/sys/bus/ieee1394/devices/fw-host0/nodes_active
4
[root@localhost fw-host0]#
[root@localhost fw-host0]# cat
/sys/bus/ieee1394/devices/fw-host0/is_root
1
[root@localhost fw-host0]#
[root@localhost fw-host0]# cat
/sys/bus/ieee1394/devices/fw-host0/is_busmgr
0
[root@localhost fw-host0]#
[root@localhost fw-host0]# cat
/sys/bus/ieee1394/devices/fw-host0/in_bus_reset
0
[root@localhost fw-host0]#

Both of the changes allow the kernel to build, but neither gives me
devices.  2.6.3-bk2 gives me:
[bob@localhost bob]$ cat /proc/partitions
major minor  #blocks  name
 
   3     0   20010312 hda
   3     1      30208 hda1
   3     2    1050840 hda2
   3     3   18922680 hda3
   3    64   80043264 hdb
   3    65   80035798 hdb1
   8     0  134217727 sda
   8     1  134215011 sda1
   8    16  134217727 sdb
   8    17  134215011 sdb1
[bob@localhost bob]$

Thanks for your replies.  
Please email me off the list if you need more information or need me to
do any testing.  Also, my PCI hardware is:
00:0d.0 FireWire (IEEE 1394): VIA Technologies, Inc. IEEE 1394 Host
Controller (rev 43)
(and the controller in the devices are Oxford Semi FW900's.)

Thanks,
Bob

Re: drivers/ieee1394/sbp2.c:734: error: `host' undeclared (first use in this function) 2.6.3-bk3

[Brad Davidson]

All,

This is the same change I made in my local tree against 2.6.3-bk4. It 
was pretty much a blind change, so I have no idea if it's the right 
thing to do or not, but it certainly compiles now. I've connected a few 
firewire devices and they appear to work as expected without anything 
going belly-up, so it's good enought for me. I guess only Bob really 
knows if that's what he meant.

I'm not subscribed to the list, reply to me personally, etc etc etc.

 > I've no hardware to test this, but does this do the right thing for you ?
 > Dave

 > < patch converting host to hi->host >

Re: drivers/ieee1394/sbp2.c:734: error: `host' undeclared (first use in this function) 2.6.3-bk3

[Dave Jones]

On Fri, Feb 27, 2004 at 07:07:38AM -0800, Brad Davidson wrote:
 > All,
 > 
 > This is the same change I made in my local tree against 2.6.3-bk4. It 
 > was pretty much a blind change, so I have no idea if it's the right 
 > thing to do or not, but it certainly compiles now. I've connected a few 
 > firewire devices and they appear to work as expected without anything 
 > going belly-up, so it's good enought for me. I guess only Bob really 
 > knows if that's what he meant.

This got fixed in -bk5 iirc.

		Dave

Re: drivers/ieee1394/sbp2.c:734: error: `host' undeclared (first use in this function) 2.6.3-bk3

[Bob Gill]

Hi.  I just saw your questions on LKML about this problem.  The current
status is that the problem (for me) still exists (but a solution is in
progress).  The initial problem was that the kernel build failed.  Both
Ben Collins' and Dave Jones' solutions fixed that problem and the build
process is successful, but my external firewire drives weren't working. 
Ben suspected IEEE1212 detection wasn't working and sent me csr_dump.  I
ran it and it confirmed that not all of my hardware rom is being read
(the rom in the device, not in the 1394 controller) and so devices
aren't being detected.  
Ben contacted Steve Kinneberg (who contacted me).  I sent Steve a
(really ugly hack) of csr_dump to try and display all of the rom, but
there are still quadlets not being detected.  I don't have IEEE1212 spec
sheets and am not familiar with the protocol, but Steve said that a CSR
(configuration and status rom) node is missing from the ieee1212 root
directory for my hard drives, and there is a gap in the rom data.  Steve
is working on a updated csr_dump and hoped to have it finished this
weekend.  I would suspect that anyone with Oxford Semi FW900 controllers
would have the same problem (certainly anyone with same rom).  I have a
sony camera connected  to the 1394 bus also, and it gets detected
perfectly.  

Bob