Re: [PATCH v12 01/15] PCI: Let pci_mmap_page_range() take extra resource pointer

From: Yinghai Lu <yinghai@kernel.org>
To: linux-arm-kernel@lists.infradead.org
Subject: Re: [PATCH v12 01/15] PCI: Let pci_mmap_page_range() take extra resource pointer
Date: Thu, 09 Jun 2016 00:00:47 +0000	[thread overview]
Message-ID: <CAE9FiQWw0pUB=1iDrX_1qyMFAUGQidSaV7CPc0aNb2CzPB-fZw@mail.gmail.com> (raw)
In-Reply-To: <CAE9FiQXPmG6UYYGHG52_i8vaBJ5yPm6Z4Zfx_BhQxVhyWC5fnw@mail.gmail.com>

[-- Attachment #1: Type: text/plain, Size: 317 bytes --]

On Wed, Jun 8, 2016 at 3:35 PM, Yinghai Lu <yinghai@kernel.org> wrote:

> At the same time, can you kill __pci_mmap_set_pgprot() for powerpc.

Can you please put your two patches and this attached one into to pci/next?

Then I could send updated PCI: Let pci_mmap_page_range() take resource address.

Thanks

Yinghai

[-- Attachment #2: remove_powerpc_pci_prot.patch --]
[-- Type: text/x-patch, Size: 2652 bytes --]

From: Bjorn Helgaas <bhelgaas@google.com>
Subject: [PATCH] powerpc/PCI: Remove __pci_mmap_set_pgprot()

  PCI: Ignore write-combining when mapping I/O port space
already handle the io port mmap path.

For mmio mmap path, caller should state that correctly if write_combine
is really needed.

via proc path it should look like:
  mmap(fd, ...)                           # default is I/O, non-combining
  ioctl(fd, PCIIOC_WRITE_COMBINE, 1);     # request write-combining
  ioctl(fd, PCIIOC_MMAP_IS_MEM);          # request memory space
  mmap(fd, ...)

sysfs path, it should use resource]?]_wc.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>

---
 arch/powerpc/kernel/pci-common.c |   37 ++++---------------------------------
 1 file changed, 4 insertions(+), 33 deletions(-)

Index: linux-2.6/arch/powerpc/kernel/pci-common.c
===================================================================

--- linux-2.6.orig/arch/powerpc/kernel/pci-common.c
+++ linux-2.6/arch/powerpc/kernel/pci-common.c
@@ -356,36 +356,6 @@ static struct resource *__pci_mmap_make_
 }
 
 /*
- * Set vm_page_prot of VMA, as appropriate for this architecture, for a pci
- * device mapping.
- */
-static pgprot_t __pci_mmap_set_pgprot(struct pci_dev *dev, struct resource *rp,
-				      pgprot_t protection,
-				      enum pci_mmap_state mmap_state,
-				      int write_combine)
-{
-
-	/* Write combine is always 0 on non-memory space mappings. On
-	 * memory space, if the user didn't pass 1, we check for a
-	 * "prefetchable" resource. This is a bit hackish, but we use
-	 * this to workaround the inability of /sysfs to provide a write
-	 * combine bit
-	 */
-	if (mmap_state != pci_mmap_mem)
-		write_combine = 0;
-	else if (write_combine == 0) {
-		if (rp->flags & IORESOURCE_PREFETCH)
-			write_combine = 1;
-	}
-
-	/* XXX would be nice to have a way to ask for write-through */
-	if (write_combine)
-		return pgprot_noncached_wc(protection);
-	else
-		return pgprot_noncached(protection);
-}
-
-/*
  * This one is used by /dev/mem and fbdev who have no clue about the
  * PCI device, it tries to find the PCI device first and calls the
  * above routine
@@ -458,9 +428,10 @@ int pci_mmap_page_range(struct pci_dev *
 		return -EINVAL;
 
 	vma->vm_pgoff = offset >> PAGE_SHIFT;
-	vma->vm_page_prot = __pci_mmap_set_pgprot(dev, rp,
-						  vma->vm_page_prot,
-						  mmap_state, write_combine);
+	if (write_combine)
+		vma->vm_page_prot = pgprot_noncached_wc(vma->vm_page_prot);
+	else
+		vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
 
 	ret = remap_pfn_range(vma, vma->vm_start, vma->vm_pgoff,
 			       vma->vm_end - vma->vm_start, vma->vm_page_prot);
