RE: [PATCH v9 2/7] PCI: Apply the new generic I/O management on PCI IO hosts

[Gabriele Paoloni]

Hi Bjorn

> -Original Message-
> From: Bjorn Helgaas [mailto:helg...@kernel.org]
> Sent: 26 May 2017 22:20
> To: Gabriele Paoloni
> Cc: catalin.mari...@arm.com; will.dea...@arm.com; robh...@kernel.org;
> frowand.l...@gmail.com; bhelg...@google.com; raf...@kernel.org;
> a...@arndb.de; linux-arm-ker...@lists.infradead.org;
> lorenzo.pieral...@arm.com; mark.rutl...@arm.com; miny...@acm.org;
> b...@kernel.crashing.org; John Garry; linux-kernel@vger.kernel.org;
> xuwei (O); Linuxarm; linux-a...@vger.kernel.org; zhichang.yuan; linux-
> p...@vger.kernel.org; o...@lixom.net; brian.star...@arm.com
> Subject: Re: [PATCH v9 2/7] PCI: Apply the new generic I/O management
> on PCI IO hosts
> 
> On Thu, May 25, 2017 at 12:37:23PM +0100, Gabriele Paoloni wrote:
> > From: "zhichang.yuan" <yuanzhich...@hisilicon.com>
> >
> > After introducing the new generic I/O space management(LOGIC_IO), the
> > original PCI MMIO relevant helpers need to be updated based on the
> new
> > interfaces defined in LOGIC_IO.
> > This patch adapts the corresponding code to match the changes
> introduced
> > by LOGIC_IO.
> >
> > Signed-off-by: zhichang.yuan <yuanzhich...@hisilicon.com>
> > Signed-off-by: Gabriele Paoloni <gabriele.paol...@huawei.com>
> > Signed-off-by: Arnd Bergmann <a...@arndb.de>#earlier draft
> 
> Not sure how you plan to merge this, but here's my ack:
> 
> Acked-by: Bjorn Helgaas <bhelg...@google.com>

Many thanks for your Ack :)

Maybe Arnd can take the whole patchset (or just the lib framework + DT part
leaving the ACPI part to Lorenzo)?

> 
> If you split this as suggested below, add my ack to all three patches.

Sure I will

> 
> > diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> > index b01bd5b..c9fe12b 100644
> > --- a/drivers/pci/pci.c
> > +++ b/drivers/pci/pci.c
> >...
> 
> > -int __weak pci_register_io_range(phys_addr_t addr, resource_size_t
> size)
> > +int pci_register_io_range(struct fwnode_handle *fwnode, phys_addr_t
> addr,
> > +   resource_size_t size)
> 
> It's trivial and nit-picky, but I would do the __weak removal in its
> own patch.  It's obviously fine because there's only one
> implementation, but it's unrelated to the main point of this patch.

Agreed

> 
> I would split the signature change (fwnode addition) to a separate
> patch, too, just to make the actual change more obvious, especially
> since that's the only part that crosses subsystems (ACPI, PCI, OF).

Agreed too

> 
> >  {
> > -   int err = 0;
> > -
> > +   int ret = 0;
> >  #ifdef PCI_IOBASE
> > -   struct io_range *range;
> > -   resource_size_t allocated_size = 0;
> > -
> > -   /* check if the range hasn't been previously recorded */
> > -   spin_lock(_range_lock);
> > -   list_for_each_entry(range, _range_list, list) {
> > -   if (addr >= range->start && addr + size <= range->start +
> size) {
> > -   /* range already registered, bail out */
> > -   goto end_register;
> > -   }
> > -   allocated_size += range->size;
> > -   }
> > -
> > -   /* range not registed yet, check for available space */
> > -   if (allocated_size + size - 1 > IO_SPACE_LIMIT) {
> > -   /* if it's too big check if 64K space can be reserved */
> > -   if (allocated_size + SZ_64K - 1 > IO_SPACE_LIMIT) {
> > -   err = -E2BIG;
> > -   goto end_register;
> > -   }
> > -
> > -   size = SZ_64K;
> > -   pr_warn("Requested IO range too big, new size set to
> 64K\n");
> > -   }
> > +   struct logic_pio_hwaddr *range;
> >
> > -   /* add the range to the list */
> > -   range = kzalloc(sizeof(*range), GFP_ATOMIC);
> > -   if (!range) {
> > -   err = -ENOMEM;
> > -   goto end_register;
> > -   }
> > +   if (!size || addr + size < addr)
> > +   return -EINVAL;
> >
> > -   range->start = addr;
> > +   range = kzalloc(sizeof(*range), GFP_KERNEL);
> > +   if (!range)
> > +   return -ENOMEM;
> 
> Add a blank line here.

Yep I will

Cheers
Gab

> 
> > +   range->fwnode = fwnode;
> > range->size = size;
> > +   range->hw_start = addr;
> > +   range->flags = PIO_CPU_MMIO;
> >
> > -   list_add_tail(>list, _range_list);
> > -
> > -end_register:
> > -   spin_unlock(_range_lock);
> > +   ret = logic_pio_register_range(range);
> > +   if (ret)
> > +   kfree(range);
> >  #endif
> >
> > -   return err;
> > +   return ret;
> >  }

RE: [PATCH v9 2/7] PCI: Apply the new generic I/O management on PCI IO hosts

[Gabriele Paoloni]

Hi Bjorn

> -Original Message-
> From: Bjorn Helgaas [mailto:helg...@kernel.org]
> Sent: 26 May 2017 22:20
> To: Gabriele Paoloni
> Cc: catalin.mari...@arm.com; will.dea...@arm.com; robh...@kernel.org;
> frowand.l...@gmail.com; bhelg...@google.com; raf...@kernel.org;
> a...@arndb.de; linux-arm-ker...@lists.infradead.org;
> lorenzo.pieral...@arm.com; mark.rutl...@arm.com; miny...@acm.org;
> b...@kernel.crashing.org; John Garry; linux-kernel@vger.kernel.org;
> xuwei (O); Linuxarm; linux-a...@vger.kernel.org; zhichang.yuan; linux-
> p...@vger.kernel.org; o...@lixom.net; brian.star...@arm.com
> Subject: Re: [PATCH v9 2/7] PCI: Apply the new generic I/O management
> on PCI IO hosts
> 
> On Thu, May 25, 2017 at 12:37:23PM +0100, Gabriele Paoloni wrote:
> > From: "zhichang.yuan" 
> >
> > After introducing the new generic I/O space management(LOGIC_IO), the
> > original PCI MMIO relevant helpers need to be updated based on the
> new
> > interfaces defined in LOGIC_IO.
> > This patch adapts the corresponding code to match the changes
> introduced
> > by LOGIC_IO.
> >
> > Signed-off-by: zhichang.yuan 
> > Signed-off-by: Gabriele Paoloni 
> > Signed-off-by: Arnd Bergmann #earlier draft
> 
> Not sure how you plan to merge this, but here's my ack:
> 
> Acked-by: Bjorn Helgaas 

Many thanks for your Ack :)

Maybe Arnd can take the whole patchset (or just the lib framework + DT part
leaving the ACPI part to Lorenzo)?

> 
> If you split this as suggested below, add my ack to all three patches.

Sure I will

> 
> > diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> > index b01bd5b..c9fe12b 100644
> > --- a/drivers/pci/pci.c
> > +++ b/drivers/pci/pci.c
> >...
> 
> > -int __weak pci_register_io_range(phys_addr_t addr, resource_size_t
> size)
> > +int pci_register_io_range(struct fwnode_handle *fwnode, phys_addr_t
> addr,
> > +   resource_size_t size)
> 
> It's trivial and nit-picky, but I would do the __weak removal in its
> own patch.  It's obviously fine because there's only one
> implementation, but it's unrelated to the main point of this patch.

Agreed

> 
> I would split the signature change (fwnode addition) to a separate
> patch, too, just to make the actual change more obvious, especially
> since that's the only part that crosses subsystems (ACPI, PCI, OF).

Agreed too

> 
> >  {
> > -   int err = 0;
> > -
> > +   int ret = 0;
> >  #ifdef PCI_IOBASE
> > -   struct io_range *range;
> > -   resource_size_t allocated_size = 0;
> > -
> > -   /* check if the range hasn't been previously recorded */
> > -   spin_lock(_range_lock);
> > -   list_for_each_entry(range, _range_list, list) {
> > -   if (addr >= range->start && addr + size <= range->start +
> size) {
> > -   /* range already registered, bail out */
> > -   goto end_register;
> > -   }
> > -   allocated_size += range->size;
> > -   }
> > -
> > -   /* range not registed yet, check for available space */
> > -   if (allocated_size + size - 1 > IO_SPACE_LIMIT) {
> > -   /* if it's too big check if 64K space can be reserved */
> > -   if (allocated_size + SZ_64K - 1 > IO_SPACE_LIMIT) {
> > -   err = -E2BIG;
> > -   goto end_register;
> > -   }
> > -
> > -   size = SZ_64K;
> > -   pr_warn("Requested IO range too big, new size set to
> 64K\n");
> > -   }
> > +   struct logic_pio_hwaddr *range;
> >
> > -   /* add the range to the list */
> > -   range = kzalloc(sizeof(*range), GFP_ATOMIC);
> > -   if (!range) {
> > -   err = -ENOMEM;
> > -   goto end_register;
> > -   }
> > +   if (!size || addr + size < addr)
> > +   return -EINVAL;
> >
> > -   range->start = addr;
> > +   range = kzalloc(sizeof(*range), GFP_KERNEL);
> > +   if (!range)
> > +   return -ENOMEM;
> 
> Add a blank line here.

Yep I will

Cheers
Gab

> 
> > +   range->fwnode = fwnode;
> > range->size = size;
> > +   range->hw_start = addr;
> > +   range->flags = PIO_CPU_MMIO;
> >
> > -   list_add_tail(>list, _range_list);
> > -
> > -end_register:
> > -   spin_unlock(_range_lock);
> > +   ret = logic_pio_register_range(range);
> > +   if (ret)
> > +   kfree(range);
> >  #endif
> >
> > -   return err;
> > +   return ret;
> >  }

Re: [PATCH v9 2/7] PCI: Apply the new generic I/O management on PCI IO hosts

[Bjorn Helgaas]

On Thu, May 25, 2017 at 12:37:23PM +0100, Gabriele Paoloni wrote:
> From: "zhichang.yuan" 
> 
> After introducing the new generic I/O space management(LOGIC_IO), the
> original PCI MMIO relevant helpers need to be updated based on the new
> interfaces defined in LOGIC_IO.
> This patch adapts the corresponding code to match the changes introduced
> by LOGIC_IO.
> 
> Signed-off-by: zhichang.yuan 
> Signed-off-by: Gabriele Paoloni 
> Signed-off-by: Arnd Bergmann #earlier draft

Not sure how you plan to merge this, but here's my ack:

Acked-by: Bjorn Helgaas 

If you split this as suggested below, add my ack to all three patches.

> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> index b01bd5b..c9fe12b 100644
> --- a/drivers/pci/pci.c
> +++ b/drivers/pci/pci.c
>...

> -int __weak pci_register_io_range(phys_addr_t addr, resource_size_t size)
> +int pci_register_io_range(struct fwnode_handle *fwnode, phys_addr_t addr,
> + resource_size_t size)

It's trivial and nit-picky, but I would do the __weak removal in its
own patch.  It's obviously fine because there's only one
implementation, but it's unrelated to the main point of this patch.

I would split the signature change (fwnode addition) to a separate
patch, too, just to make the actual change more obvious, especially
since that's the only part that crosses subsystems (ACPI, PCI, OF).

>  {
> - int err = 0;
> -
> + int ret = 0;
>  #ifdef PCI_IOBASE
> - struct io_range *range;
> - resource_size_t allocated_size = 0;
> -
> - /* check if the range hasn't been previously recorded */
> - spin_lock(_range_lock);
> - list_for_each_entry(range, _range_list, list) {
> - if (addr >= range->start && addr + size <= range->start + size) 
> {
> - /* range already registered, bail out */
> - goto end_register;
> - }
> - allocated_size += range->size;
> - }
> -
> - /* range not registed yet, check for available space */
> - if (allocated_size + size - 1 > IO_SPACE_LIMIT) {
> - /* if it's too big check if 64K space can be reserved */
> - if (allocated_size + SZ_64K - 1 > IO_SPACE_LIMIT) {
> - err = -E2BIG;
> - goto end_register;
> - }
> -
> - size = SZ_64K;
> - pr_warn("Requested IO range too big, new size set to 64K\n");
> - }
> + struct logic_pio_hwaddr *range;
>  
> - /* add the range to the list */
> - range = kzalloc(sizeof(*range), GFP_ATOMIC);
> - if (!range) {
> - err = -ENOMEM;
> - goto end_register;
> - }
> + if (!size || addr + size < addr)
> + return -EINVAL;
>  
> - range->start = addr;
> + range = kzalloc(sizeof(*range), GFP_KERNEL);
> + if (!range)
> + return -ENOMEM;

Add a blank line here.

> + range->fwnode = fwnode;
>   range->size = size;
> + range->hw_start = addr;
> + range->flags = PIO_CPU_MMIO;
>  
> - list_add_tail(>list, _range_list);
> -
> -end_register:
> - spin_unlock(_range_lock);
> + ret = logic_pio_register_range(range);
> + if (ret)
> + kfree(range);
>  #endif
>  
> - return err;
> + return ret;
>  }

Re: [PATCH v9 2/7] PCI: Apply the new generic I/O management on PCI IO hosts

[Bjorn Helgaas]

On Thu, May 25, 2017 at 12:37:23PM +0100, Gabriele Paoloni wrote:
> From: "zhichang.yuan" 
> 
> After introducing the new generic I/O space management(LOGIC_IO), the
> original PCI MMIO relevant helpers need to be updated based on the new
> interfaces defined in LOGIC_IO.
> This patch adapts the corresponding code to match the changes introduced
> by LOGIC_IO.
> 
> Signed-off-by: zhichang.yuan 
> Signed-off-by: Gabriele Paoloni 
> Signed-off-by: Arnd Bergmann #earlier draft

Not sure how you plan to merge this, but here's my ack:

Acked-by: Bjorn Helgaas 

If you split this as suggested below, add my ack to all three patches.

> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> index b01bd5b..c9fe12b 100644
> --- a/drivers/pci/pci.c
> +++ b/drivers/pci/pci.c
>...

> -int __weak pci_register_io_range(phys_addr_t addr, resource_size_t size)
> +int pci_register_io_range(struct fwnode_handle *fwnode, phys_addr_t addr,
> + resource_size_t size)

It's trivial and nit-picky, but I would do the __weak removal in its
own patch.  It's obviously fine because there's only one
implementation, but it's unrelated to the main point of this patch.

I would split the signature change (fwnode addition) to a separate
patch, too, just to make the actual change more obvious, especially
since that's the only part that crosses subsystems (ACPI, PCI, OF).

>  {
> - int err = 0;
> -
> + int ret = 0;
>  #ifdef PCI_IOBASE
> - struct io_range *range;
> - resource_size_t allocated_size = 0;
> -
> - /* check if the range hasn't been previously recorded */
> - spin_lock(_range_lock);
> - list_for_each_entry(range, _range_list, list) {
> - if (addr >= range->start && addr + size <= range->start + size) 
> {
> - /* range already registered, bail out */
> - goto end_register;
> - }
> - allocated_size += range->size;
> - }
> -
> - /* range not registed yet, check for available space */
> - if (allocated_size + size - 1 > IO_SPACE_LIMIT) {
> - /* if it's too big check if 64K space can be reserved */
> - if (allocated_size + SZ_64K - 1 > IO_SPACE_LIMIT) {
> - err = -E2BIG;
> - goto end_register;
> - }
> -
> - size = SZ_64K;
> - pr_warn("Requested IO range too big, new size set to 64K\n");
> - }
> + struct logic_pio_hwaddr *range;
>  
> - /* add the range to the list */
> - range = kzalloc(sizeof(*range), GFP_ATOMIC);
> - if (!range) {
> - err = -ENOMEM;
> - goto end_register;
> - }
> + if (!size || addr + size < addr)
> + return -EINVAL;
>  
> - range->start = addr;
> + range = kzalloc(sizeof(*range), GFP_KERNEL);
> + if (!range)
> + return -ENOMEM;

Add a blank line here.

> + range->fwnode = fwnode;
>   range->size = size;
> + range->hw_start = addr;
> + range->flags = PIO_CPU_MMIO;
>  
> - list_add_tail(>list, _range_list);
> -
> -end_register:
> - spin_unlock(_range_lock);
> + ret = logic_pio_register_range(range);
> + if (ret)
> + kfree(range);
>  #endif
>  
> - return err;
> + return ret;
>  }