Re: [PATCH v3 04/26] PCI: endpoint: Use PCI_STD_NUM_BARS

[Denis Efremov]

On 9/18/19 12:19 PM, Andrew Murray wrote:
> On Mon, Sep 16, 2019 at 11:41:36PM +0300, Denis Efremov wrote:
>> To iterate through all possible BARs, loop conditions refactored to the
>> *number* of BARs "i < PCI_STD_NUM_BARS", instead of the index of the last
>> valid BAR "i <= BAR_5". This is more idiomatic C style and allows to avoid
>> the fencepost error. Array definitions changed to PCI_STD_NUM_BARS where
>> appropriate.
>>
>> Cc: Kishon Vijay Abraham I <kishon@xxxxxx>
>> Cc: Lorenzo Pieralisi <lorenzo.pieralisi@xxxxxxx>
>> Signed-off-by: Denis Efremov <efremov@xxxxxxxxx>
>> ---
>>  drivers/pci/endpoint/functions/pci-epf-test.c | 10 +++++-----
>>  include/linux/pci-epc.h                       |  2 +-
>>  2 files changed, 6 insertions(+), 6 deletions(-)
>>
>> diff --git a/drivers/pci/endpoint/functions/pci-epf-test.c b/drivers/pci/endpoint/functions/pci-epf-test.c
>> index 1cfe3687a211..5d74f81ddfe4 100644
>> --- a/drivers/pci/endpoint/functions/pci-epf-test.c
>> +++ b/drivers/pci/endpoint/functions/pci-epf-test.c
>> @@ -44,7 +44,7 @@
>>  static struct workqueue_struct *kpcitest_workqueue;
>>  
>>  struct pci_epf_test {
>> -	void			*reg[6];
>> +	void			*reg[PCI_STD_NUM_BARS];
>>  	struct pci_epf		*epf;
>>  	enum pci_barno		test_reg_bar;
>>  	struct delayed_work	cmd_handler;
>> @@ -377,7 +377,7 @@ static void pci_epf_test_unbind(struct pci_epf *epf)
>>  
>>  	cancel_delayed_work(&epf_test->cmd_handler);
>>  	pci_epc_stop(epc);
>> -	for (bar = BAR_0; bar <= BAR_5; bar++) {
>> +	for (bar = 0; bar < PCI_STD_NUM_BARS; bar++) {
>>  		epf_bar = &epf->bar[bar];
>>  
>>  		if (epf_test->reg[bar]) {
>> @@ -400,7 +400,7 @@ static int pci_epf_test_set_bar(struct pci_epf *epf)
>>  
>>  	epc_features = epf_test->epc_features;
>>  
>> -	for (bar = BAR_0; bar <= BAR_5; bar += add) {
>> +	for (bar = 0; bar < PCI_STD_NUM_BARS; bar += add) {
> 
> Is it possible to completely remove the BAR_x macros, or are there exsiting
> users after this patchset?

They are still used in other parts of the code. So, I've decided to preserve
the defines in this case.

pci-epc-core.c
400:        (epf_bar->barno == BAR_5 &&
429:        (epf_bar->barno == BAR_5 &&
functions/pci-epf-test.c
497:    enum pci_barno test_reg_bar = BAR_0;

> 
> As your patchset replaces BAR_0 with 0 and BAR_1 with 1, does this suggest
> that other users of BAR_x should be removed and also replaced with a number?

I changed BAR_0 to 0 in order to not mix different notions, i.e. the number
of bars and the concrete bar.

> 
> Apologies if you this doesn't fall in the remit of this patchset.

I don't know what is better here. It's simple enough to remove these defines.
However, I would prefer to wait for the endpoint developers opinion.

Thanks for the review!
Denis

> 
> Thanks,
> 
> Andrew Murray
> 
>>  		epf_bar = &epf->bar[bar];
>>  		/*
>>  		 * pci_epc_set_bar() sets PCI_BASE_ADDRESS_MEM_TYPE_64
>> @@ -450,7 +450,7 @@ static int pci_epf_test_alloc_space(struct pci_epf *epf)
>>  	}
>>  	epf_test->reg[test_reg_bar] = base;
>>  
>> -	for (bar = BAR_0; bar <= BAR_5; bar += add) {
>> +	for (bar = 0; bar < PCI_STD_NUM_BARS; bar += add) {
>>  		epf_bar = &epf->bar[bar];
>>  		add = (epf_bar->flags & PCI_BASE_ADDRESS_MEM_TYPE_64) ? 2 : 1;
>>  
>> @@ -478,7 +478,7 @@ static void pci_epf_configure_bar(struct pci_epf *epf,
>>  	bool bar_fixed_64bit;
>>  	int i;
>>  
>> -	for (i = BAR_0; i <= BAR_5; i++) {
>> +	for (i = 0; i < PCI_STD_NUM_BARS; i++) {
>>  		epf_bar = &epf->bar[i];
>>  		bar_fixed_64bit = !!(epc_features->bar_fixed_64bit & (1 << i));
>>  		if (bar_fixed_64bit)
>> diff --git a/include/linux/pci-epc.h b/include/linux/pci-epc.h
>> index f641badc2c61..56f1846b9d39 100644
>> --- a/include/linux/pci-epc.h
>> +++ b/include/linux/pci-epc.h
>> @@ -117,7 +117,7 @@ struct pci_epc_features {
>>  	unsigned int	msix_capable : 1;
>>  	u8	reserved_bar;
>>  	u8	bar_fixed_64bit;
>> -	u64	bar_fixed_size[BAR_5 + 1];
>> +	u64	bar_fixed_size[PCI_STD_NUM_BARS];
>>  	size_t	align;
>>  };
>>  
>> -- 
>> 2.21.0
>>