ARM Versatile defconfig PCI io out of range?

[robherring2@gmail.com (Rob Herring)]

On 05/30/2012 01:02 PM, Richard Maw wrote:
> When I build from f9369910a6225b8d4892c3f20ae740a711cd5ace
> with versatile_defconfig, after making the changes described in
> http://fedoraproject.org/wiki/Architectures/ARM/HowToQemu#Build_Kernel_Image_From_Source
> Enabling PCI and the SYM53C8XX scsi disk driver, which aren't in the
> defconfig.
> 
> QEMU is run with the command
>   qemu-system-arm -M versatilepb \
>   -drive file=$diskimage,if=scsi \
>   -kernel linux/arch/arm/boot/zImage -append root=/dev/sda1
>   
> The system fails to find the scsi disk, due to the controller failing to
> request the IO resource.
> 
> It gives the following on the console
> Uncompressing Linux... done, booting the kernel.
> Booting Linux on physical CPU 0
> Linux version 3.4.0+ (richardmaw at aegir) (gcc version 4.6.1
> (Ubuntu/Linaro 4.6.1-9ubuntu3) ) #14 Wed May 30 16:54:55 BST 2012
> CPU: ARM926EJ-S [41069265] revision 5 (ARMv5TEJ), cr=00093177
> CPU: VIVT data cache, VIVT instruction cache
> Machine: ARM-Versatile PB
> Memory policy: ECC disabled, Data cache writeback
> --SNIP--
> PCI core found (slot 11)
> PCI host bridge to bus 0000:00
> pci_bus 0000:00: root bus resource [io  0x44000000-0x4fffffff]
> pci_bus 0000:00: root bus resource [mem 0x50000000-0x5fffffff]
> pci_bus 0000:00: root bus resource [mem 0x60000000-0x6fffffff pref]
> PCI: bus0: Fast back to back transfers disabled
> pci 0000:00:0c.0: BAR 2: assigned [mem 0x50000000-0x50001fff]
> pci 0000:00:0c.0: BAR 1: assigned [mem 0x50002000-0x500023ff]
> pci 0000:00:0c.0: BAR 0: assigned [io  0x44000000-0x440000ff]
> bio: create slab <bio-0> at 0
> --SNIP--
> brd: module loaded
> PCI: enabling device 0000:00:0c.0 (0100 -> 0103)
> sym53c8xx 0000:00:0c.0: pci_request_regions called
> sym53c8xx 0000:00:0c.0: BAR 0: can't reserve [io  0x44000000-0x440000ff]
> --SNIP--
> VFS: Cannot open root device "(null)" or unknown-block(0,0): error -6
> Please append a correct "root=" boot option; here are the available
> partitions:
> Kernel panic - not syncing: VFS: Unable to mount root fs on
> unknown-block(0,0)
> [<c0019360>] (unwind_backtrace+0x0/0xf0) from [<c02876e8>]
> (panic+0x74/0x1a8)
> [<c02876e8>] (panic+0x74/0x1a8) from [<c035fc50>]
> (mount_block_root+0x178/0x22c)
> [<c035fc50>] (mount_block_root+0x178/0x22c) from [<c035fee4>]
> (mount_root+0xe4/0x10c)
> [<c035fee4>] (mount_root+0xe4/0x10c) from [<c036006c>]
> (prepare_namespace+0x160/0x1b8)
> [<c036006c>] (prepare_namespace+0x160/0x1b8) from [<c035f908>]
> (kernel_init+0x198/0x1d8)
> [<c035f908>] (kernel_init+0x198/0x1d8) from [<c0014510>]
> (kernel_thread_exit+0x0/0x8)
> QEMU: Terminated
> 
> We've tried to track down the problem, but we couldn't get much further
> than finding out that the IO_SPACE_LIMIT in kernel/resource.c is 0xffff,
> which is lower than the required 0x44000000.
> 
> We're not sure whether this is a problem with it not being allocated
> enough space, or that it is not being translated properly.
> 
> Is this a known issue, I did a quick google for the "sym53c8xx
>  can't reserve io" error, but nothing turned up.
> If not, does anyone have any pointers to how this could be fixed?
> 

I believe this is the same issue:

http://www.gossamer-threads.com/lists/linux/kernel/1543569

In looking at various PCI platforms and for a qemu platform with PCI to
test PCI changes, I had concluded (wrongly I guess) that Versatile PCI
was broken. This patch which has never gone upstream also played a part
in my conclusion:

http://lists.infradead.org/pipermail/linux-arm-kernel/2010-July/021035.html

The simple fix is to restore io.h for Versatile, but I'm looking for a
better way.

Rob