Re: [PATCH v2 00/10] Add definition for the number of standard PCI BARs

From: Denis Efremov <efremov@linux.com>
To: Andrew Murray <andrew.murray@arm.com>
Cc: Bjorn Helgaas <bhelgaas@google.com>,
	linux-kernel@vger.kernel.org, linux-pci@vger.kernel.org,
	Sebastian Ott <sebott@linux.ibm.com>,
	Gerald Schaefer <gerald.schaefer@de.ibm.com>,
	"H. Peter Anvin" <hpa@zytor.com>,
	Giuseppe Cavallaro <peppe.cavallaro@st.com>,
	Alexandre Torgue <alexandre.torgue@st.com>,
	Matt Porter <mporter@kernel.crashing.org>,
	Alexandre Bounine <alex.bou9@gmail.com>,
	Peter Jones <pjones@redhat.com>,
	Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>,
	Cornelia Huck <cohuck@redhat.com>,
	Alex Williamson <alex.williamson@redhat.com>,
	Jose Abreu <Jose.Abreu@synopsys.com>,
	kvm@vger.kernel.org, linux-fbdev@vger.kernel.org,
	netdev@vger.kernel.org, x86@kernel.org,
	linux-s390@vger.kernel.org
Subject: Re: [PATCH v2 00/10] Add definition for the number of standard PCI BARs
Date: Thu, 5 Sep 2019 22:02:13 +0300	[thread overview]
Message-ID: <5d8cdc6b-f8c9-c3f1-e11d-13b3a7eb5b26@linux.com> (raw)
In-Reply-To: <20190816105128.GD14111@e119886-lin.cambridge.arm.com>

On 16.08.2019 13:51, Andrew Murray wrote:
> On Fri, Aug 16, 2019 at 12:24:27PM +0300, Denis Efremov wrote:
>> Code that iterates over all standard PCI BARs typically uses
>> PCI_STD_RESOURCE_END, but this is error-prone because it requires
>> "i <= PCI_STD_RESOURCE_END" rather than something like
>> "i < PCI_STD_NUM_BARS". We could add such a definition and use it the same
>> way PCI_SRIOV_NUM_BARS is used. There is already the definition
>> PCI_BAR_COUNT for s390 only. Thus, this patchset introduces it globally.
>>
>> Changes in v2:
>>   - Reverse checks in pci_iomap_range,pci_iomap_wc_range.
>>   - Refactor loops in vfio_pci to keep PCI_STD_RESOURCES.
>>   - Add 2 new patches to replace the magic constant with new define.
>>   - Split net patch in v1 to separate stmmac and dwc-xlgmac patches.
>>
>> Denis Efremov (10):
>>   PCI: Add define for the number of standard PCI BARs
>>   s390/pci: Loop using PCI_STD_NUM_BARS
>>   x86/PCI: Loop using PCI_STD_NUM_BARS
>>   stmmac: pci: Loop using PCI_STD_NUM_BARS
>>   net: dwc-xlgmac: Loop using PCI_STD_NUM_BARS
>>   rapidio/tsi721: Loop using PCI_STD_NUM_BARS
>>   efifb: Loop using PCI_STD_NUM_BARS
>>   vfio_pci: Loop using PCI_STD_NUM_BARS
>>   PCI: hv: Use PCI_STD_NUM_BARS
>>   PCI: Use PCI_STD_NUM_BARS
>>
>>  arch/s390/include/asm/pci.h                      |  5 +----
>>  arch/s390/include/asm/pci_clp.h                  |  6 +++---
>>  arch/s390/pci/pci.c                              | 16 ++++++++--------
>>  arch/s390/pci/pci_clp.c                          |  6 +++---
>>  arch/x86/pci/common.c                            |  2 +-
>>  drivers/net/ethernet/stmicro/stmmac/stmmac_pci.c |  4 ++--
>>  drivers/net/ethernet/synopsys/dwc-xlgmac-pci.c   |  2 +-
>>  drivers/pci/controller/pci-hyperv.c              | 10 +++++-----
>>  drivers/pci/pci.c                                | 11 ++++++-----
>>  drivers/pci/quirks.c                             |  4 ++--
>>  drivers/rapidio/devices/tsi721.c                 |  2 +-
>>  drivers/vfio/pci/vfio_pci.c                      | 11 +++++++----
>>  drivers/vfio/pci/vfio_pci_config.c               | 10 ++++++----
>>  drivers/vfio/pci/vfio_pci_private.h              |  4 ++--
>>  drivers/video/fbdev/efifb.c                      |  2 +-
>>  include/linux/pci.h                              |  2 +-
>>  include/uapi/linux/pci_regs.h                    |  1 +
>>  17 files changed, 51 insertions(+), 47 deletions(-)
> 
> I've come across a few more places where this change can be made. There
> may be multiple instances in the same file, but only the first is shown
> below:
> 
> drivers/misc/pci_endpoint_test.c:       for (bar = BAR_0; bar <= BAR_5; bar++) {
> drivers/net/ethernet/intel/e1000/e1000_main.c:          for (i = BAR_1; i <= BAR_5; i++) {
> drivers/net/ethernet/intel/ixgb/ixgb_main.c:    for (i = BAR_1; i <= BAR_5; i++) {
> drivers/pci/controller/dwc/pci-dra7xx.c:        for (bar = BAR_0; bar <= BAR_5; bar++)
> drivers/pci/controller/dwc/pci-layerscape-ep.c: for (bar = BAR_0; bar <= BAR_5; bar++)
> drivers/pci/controller/dwc/pcie-artpec6.c:      for (bar = BAR_0; bar <= BAR_5; bar++)
> drivers/pci/controller/dwc/pcie-designware-plat.c:      for (bar = BAR_0; bar <= BAR_5; bar++)
> drivers/pci/endpoint/functions/pci-epf-test.c:  for (bar = BAR_0; bar <= BAR_5; bar++) {
> include/linux/pci-epc.h:        u64     bar_fixed_size[BAR_5 + 1];
> drivers/scsi/pm8001/pm8001_hwi.c:       for (bar = 0; bar < 6; bar++) {
> drivers/scsi/pm8001/pm8001_init.c:      for (bar = 0; bar < 6; bar++) {
> drivers/ata/sata_nv.c:  for (bar = 0; bar < 6; bar++)
> drivers/video/fbdev/core/fbmem.c:       for (idx = 0, bar = 0; bar < PCI_ROM_RESOURCE; bar++) {
> drivers/staging/gasket/gasket_core.c:   for (i = 0; i < GASKET_NUM_BARS; i++) {
> drivers/tty/serial/8250/8250_pci.c:     for (i = 0; i < PCI_NUM_BAR_RESOURCES; i++) { <-----------
> 
> It looks like BARs are often iterated with PCI_NUM_BAR_RESOURCES, there
> are a load of these too found with:
> 
> git grep PCI_ROM_RESOURCE | grep "< "
> 
> I'm happy to share patches if preferred.
> 

I'm not sure about lib/devres.c
265:#define PCIM_IOMAP_MAX      PCI_ROM_RESOURCE
268:    void __iomem *table[PCIM_IOMAP_MAX];
277:    for (i = 0; i < PCIM_IOMAP_MAX; i++)
324:    BUG_ON(bar >= PCIM_IOMAP_MAX);
352:    for (i = 0; i < PCIM_IOMAP_MAX; i++)
455:    for (i = 0; i < PCIM_IOMAP_MAX; i++) {

Is it worth changing?
#define PCIM_IOMAP_MAX  PCI_STD_NUM_BARS

Thanks,
Denis