linux-pci.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Bjorn Helgaas <bhelgaas@google.com>
To: matt@masarand.com
Cc: linux-pci@vger.kernel.org
Subject: Re: [PATCH 04/18] Delayed powerpc setup of PCI IRQs to bus scan time
Date: Tue, 14 Oct 2014 12:20:48 -0600	[thread overview]
Message-ID: <20141014182048.GD10125@google.com> (raw)
In-Reply-To: <1412222866-21068-5-git-send-email-matt@masarand.com>

On Thu, Oct 02, 2014 at 05:07:32AM +0100, matt@masarand.com wrote:
> From: Matthew Minter <matt@masarand.com>
> 
> The powerpc PCI init code is currently initialising PCI device IRQ during
> the boot time pcibios phase. This results in devices which are connected
> after boot time not being asigned IRQs which can cause various problems.
> 
> This patch as part of its set fixes this be moving the IRQ initialisation
> into the PCI device enable code path so that it is run for boot time and
> hot-plugged devices alike.
> 
> Signed-off-by: Matthew Minter <matt@masarand.com>
> 
> ---
>  arch/powerpc/kernel/pci-common.c | 26 +++++++++++++-------------
>  1 file changed, 13 insertions(+), 13 deletions(-)
> 
> diff --git a/arch/powerpc/kernel/pci-common.c b/arch/powerpc/kernel/pci-common.c
> index b2814e2..a8be5a5 100644
> --- a/arch/powerpc/kernel/pci-common.c
> +++ b/arch/powerpc/kernel/pci-common.c
> @@ -217,7 +217,7 @@ struct pci_controller* pci_find_hose_for_OF_device(struct device_node* node)
>   * If the interrupt is used, then gets the interrupt line from the
>   * openfirmware and sets it in the pci_dev and pci_config line.
>   */
> -static int pci_read_irq_line(struct pci_dev *pci_dev)
> +static int pci_read_irq_line(struct pci_dev *pci_dev, u8 pin)
>  {
>  	struct of_phandle_args oirq;
>  	unsigned int virq;
> @@ -229,7 +229,7 @@ static int pci_read_irq_line(struct pci_dev *pci_dev)
>  #endif
>  	/* Try to get a mapping from the device-tree */
>  	if (of_irq_parse_pci(pci_dev, &oirq)) {
> -		u8 line, pin;
> +		u8 line;
>  
>  		/* If that fails, lets fallback to what is in the config
>  		 * space and map that through the default controller. We
> @@ -238,10 +238,6 @@ static int pci_read_irq_line(struct pci_dev *pci_dev)
>  		 * either provide a proper interrupt tree or don't use this
>  		 * function.
>  		 */
> -		if (pci_read_config_byte(pci_dev, PCI_INTERRUPT_PIN, &pin))
> -			return -1;
> -		if (pin == 0)
> -			return -1;
>  		if (pci_read_config_byte(pci_dev, PCI_INTERRUPT_LINE, &line) ||
>  		    line == 0xff || line == 0) {
>  			return -1;
> @@ -266,9 +262,16 @@ static int pci_read_irq_line(struct pci_dev *pci_dev)
>  
>  	pr_debug(" Mapped to linux irq %d\n", virq);
>  
> -	pci_dev->irq = virq;
> +	return virq;
> +}
>  
> -	return 0;
> +int pci_map_irq(struct pci_dev *dev, u8 slot, u8 pin)

Should be static.

> +{
> +	/* Read default IRQs and fixup if necessary */
> +	int irq = pci_read_irq_line(dev, pin);
> +	if (ppc_md.pci_irq_fixup)
> +		ppc_md.pci_irq_fixup(dev);
> +	return irq;

The .pci_irq_fixup() can change dev->irq, so wouldn't you want to return
dev->irq, not irq, here?  Otherwise, pdev_assign_irq() (the caller of
pci_map_irq()) will overwrite the dev->irq value set by the fixup.

>  }
>  
>  /*
> @@ -766,6 +769,8 @@ int pci_proc_domain(struct pci_bus *bus)
>  
>  int pcibios_root_bridge_prepare(struct pci_host_bridge *bridge)
>  {
> +	bridge->swizzle_irq = NULL;
> +	bridge->map_irq = pci_map_irq;
>  	if (ppc_md.pcibios_root_bridge_prepare)
>  		return ppc_md.pcibios_root_bridge_prepare(bridge);
>  
> @@ -968,11 +973,6 @@ static void pcibios_setup_device(struct pci_dev *dev)
>  	/* Additional platform DMA/iommu setup */
>  	if (ppc_md.pci_dma_dev_setup)
>  		ppc_md.pci_dma_dev_setup(dev);
> -
> -	/* Read default IRQs and fixup if necessary */
> -	pci_read_irq_line(dev);
> -	if (ppc_md.pci_irq_fixup)
> -		ppc_md.pci_irq_fixup(dev);
>  }
>  
>  int pcibios_add_device(struct pci_dev *dev)
> -- 
> 2.1.0
> 

  reply	other threads:[~2014-10-14 18:20 UTC|newest]

Thread overview: 36+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-10-02  4:07 [PATCH V3] Delay allocation of PCI device IRQs from boot time until bus scan time to fix PCI hotplugging matt
2014-10-02  4:07 ` [PATCH 01/18] Added way to register deferred PCI IRQ assignment handlers matt
2014-10-02 10:33   ` Liviu Dudau
2014-10-14 23:25     ` Bjorn Helgaas
2014-10-14 17:15   ` Bjorn Helgaas
2014-10-02  4:07 ` [PATCH 02/18] Delayed x86 setup of PCI IRQs to bus scan time matt
2014-10-02 10:51   ` Liviu Dudau
2014-10-14 18:11   ` Bjorn Helgaas
2014-10-02  4:07 ` [PATCH 03/18] Delayed arm " matt
2014-10-14 18:14   ` Bjorn Helgaas
2014-10-02  4:07 ` [PATCH 04/18] Delayed powerpc " matt
2014-10-14 18:20   ` Bjorn Helgaas [this message]
2014-10-02  4:07 ` [PATCH 05/18] Delayed sh " matt
2014-10-14 18:25   ` Bjorn Helgaas
2014-10-02  4:07 ` [PATCH 06/18] Delayed alpha " matt
2014-10-14 18:27   ` Bjorn Helgaas
2014-10-02  4:07 ` [PATCH 07/18] Delayed cris " matt
2014-10-14 18:34   ` Bjorn Helgaas
2014-10-02  4:07 ` [PATCH 08/18] Delayed frv " matt
2014-10-14 18:37   ` Bjorn Helgaas
2014-10-02  4:07 ` [PATCH 09/18] Delayed m68k " matt
2014-10-14 18:38   ` Bjorn Helgaas
2014-10-02  4:07 ` [PATCH 10/18] Delayed microblaze " matt
2014-10-02  4:07 ` [PATCH 11/18] Delayed mips " matt
2014-10-02  4:07 ` [PATCH 12/18] Delayed mn10300 " matt
2014-10-14 18:46   ` Bjorn Helgaas
2014-10-02  4:07 ` [PATCH 13/18] Delayed sparc " matt
2014-10-14 18:51   ` Bjorn Helgaas
2014-10-02  4:07 ` [PATCH 14/18] Delayed tile " matt
2014-10-02  4:07 ` [PATCH 15/18] Delayed unicore32 " matt
2014-10-02  4:07 ` [PATCH 16/18] Disabled bus scan time PCI IRQ assignment on ia64 matt
2014-10-14 18:53   ` Bjorn Helgaas
2014-10-02  4:07 ` [PATCH 17/18] Disabled bus scan time PCI IRQ assignment on parisc matt
2014-10-02  4:07 ` [PATCH 18/18] Disabled bus scan time PCI IRQ assignment on s390 matt
2015-01-31 14:56 ` [PATCH V3] Delay allocation of PCI device IRQs from boot time until bus scan time to fix PCI hotplugging Bjorn Helgaas
  -- strict thread matches above, loose matches on Subject: below --
2014-10-02  3:50 matt
2014-10-02  3:50 ` [PATCH 04/18] Delayed powerpc setup of PCI IRQs to bus scan time matt

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20141014182048.GD10125@google.com \
    --to=bhelgaas@google.com \
    --cc=linux-pci@vger.kernel.org \
    --cc=matt@masarand.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).