[PATCH 3/6] powerpc/fsl-pci: Determine primary bus by looking for ISA node

[Scott Wood]

On 07/24/2012 09:42 PM, Jia Hongtao-B38951 wrote:
> 
> 
>> -----Original Message-----
>> From: Wood Scott-B07421
>> Sent: Wednesday, July 25, 2012 2:48 AM
>> To: Jia Hongtao-B38951
>> Cc: linuxppc-dev at lists.ozlabs.org; galak at kernel.crashing.org; Wood Scott-
>> B07421; Li Yang-R58472
>> Subject: Re: [PATCH 3/6] powerpc/fsl-pci: Determine primary bus by
>> looking for ISA node
>>
>> On 07/24/2012 05:20 AM, Jia Hongtao wrote:
>>> PCI host bridge is primary bus if it contains an ISA node. But not all
>> boards
>>> fit this rule. Device tree should be updated for all these boards.
>>>
>>> Signed-off-by: Jia Hongtao <B38951 at freescale.com>
>>> Signed-off-by: Li Yang <leoli at freescale.com>
>>> ---
>>>  arch/powerpc/include/asm/pci-bridge.h |    1 +
>>>  arch/powerpc/sysdev/fsl_pci.c         |   31 ++++++++++++++++++++++++-
>> ------
>>>  arch/powerpc/sysdev/fsl_pci.h         |   12 +++++++++++-
>>>  3 files changed, 36 insertions(+), 8 deletions(-)
>>>
>>> diff --git a/arch/powerpc/include/asm/pci-bridge.h
>> b/arch/powerpc/include/asm/pci-bridge.h
>>> index ac39e6a..b48fa7f 100644
>>> --- a/arch/powerpc/include/asm/pci-bridge.h
>>> +++ b/arch/powerpc/include/asm/pci-bridge.h
>>> @@ -20,6 +20,7 @@ struct device_node;
>>>  struct pci_controller {
>>>  	struct pci_bus *bus;
>>>  	char is_dynamic;
>>> +	int is_primary;
>>>  #ifdef CONFIG_PPC64
>>>  	int node;
>>>  #endif
>>> diff --git a/arch/powerpc/sysdev/fsl_pci.c
>> b/arch/powerpc/sysdev/fsl_pci.c
>>> index 99a3e78..2a369be 100644
>>> --- a/arch/powerpc/sysdev/fsl_pci.c
>>> +++ b/arch/powerpc/sysdev/fsl_pci.c
>>> @@ -453,6 +453,7 @@ int __init fsl_add_bridge(struct device_node *dev,
>> int is_primary)
>>>
>>>  	hose->first_busno = bus_range ? bus_range[0] : 0x0;
>>>  	hose->last_busno = bus_range ? bus_range[1] : 0xff;
>>> +	hose->is_primary = is_primary;
>>>
>>>  	setup_indirect_pci(hose, rsrc.start, rsrc.start + 0x4,
>>>  		PPC_INDIRECT_TYPE_BIG_ENDIAN);
>>> @@ -932,18 +933,34 @@ void pci_check_swiotlb(void)
>>>  }
>>>  #endif
>>>
>>> -int primary_phb_addr;
>>> +/*
>>> + * Recursively scan all the children nodes of parent and find out if
>> there
>>> + * is "isa" node. Return 1 if parent has isa node otherwise return 0.
>>> + */
>>> +int has_isa_node(struct device_node *parent)
>>> +{
>>> +	static int result;
>>> +	struct device_node *cur_child;
>>> +
>>> +	cur_child = NULL;
>>> +	result = 0;
>>> +	while (!result && (cur_child = of_get_next_child(parent,
>> cur_child))) {
>>> +		/* Get "isa" node and return 1 */
>>> +		if (of_node_cmp(cur_child->type, "isa") == 0)
>>> +			return result = 1;
>>> +		has_isa_node(cur_child);
>>> +	}
>>> +
>>> +	return result;
>>> +}
>>
>> Why are you reimplementing this?  It's already in Linus's tree.  See
>> fsl_pci_init().
>>
>> Plus, your version is recursive which is unacceptable in kernel code
>> with a small stack (outside of a few rare examples where the depth has a
>> small fixed upper bound), and once it finds an ISA node, it returns 1
>> forever, regardless of what node you pass in in the future.
>>
>> -Scott
> 
> About recursion I will do some more investigation.
> If it finds ISA it returns 1. But for next PCI node it will return 0 if
> no ISA is found (note that I set result to 0 at the beginning).

Sorry, I misread that as an initializer.  Still, it's awkward (why not
just use a return value?) and non-thread-safe (may not matter here, but
it's a bad habit), and it's still an unacceptable use of recursion, and
still a reimplementation of functionality that already exists. :-)

-Scott