From: Bjorn Helgaas <helgaas@kernel.org>
To: Yinghai Lu <yinghai@kernel.org>
Cc: Bjorn Helgaas <bhelgaas@google.com>,
David Miller <davem@davemloft.net>,
Benjamin Herrenschmidt <benh@kernel.crashing.org>,
Linus Torvalds <torvalds@linux-foundation.org>,
Wei Yang <weiyang@linux.vnet.ibm.com>, TJ <linux@iam.tj>,
Yijing Wang <wangyijing@huawei.com>,
Khalid Aziz <khalid.aziz@oracle.com>,
linux-pci@vger.kernel.org, linux-kernel@vger.kernel.org,
Michael Ellerman <mpe@ellerman.id.au>
Subject: Re: [PATCH v11 04/60] sparc/PCI: Use correct offset for bus address to resource
Date: Fri, 22 Apr 2016 15:49:20 -0500 [thread overview]
Message-ID: <20160422204920.GA17215@localhost> (raw)
In-Reply-To: <1460074573-7481-5-git-send-email-yinghai@kernel.org>
[+cc Ben, Michael]
On Thu, Apr 07, 2016 at 05:15:17PM -0700, Yinghai Lu wrote:
> After we added 64bit mmio parsing, we got some "no compatible bridge window"
> warning on anther new model that support 64bit resource.
>
> It turns out that we can not use mem_space.start as 64bit mem space
> offset, aka there is mem_space.start != offset.
>
> Use child_phys_addr to calculate exact offset and record offset in
> pbm.
>
> After patch we get correct offset.
>
> /pci@305: PCI IO [io 0x2007e00000000-0x2007e0fffffff] offset 2007e00000000
> /pci@305: PCI MEM [mem 0x2000000100000-0x200007effffff] offset 2000000000000
> /pci@305: PCI MEM64 [mem 0x2000100000000-0x2000dffffffff] offset 2000000000000
> ...
> pci_sun4v f02ae7f8: PCI host bridge to bus 0000:00
> pci_bus 0000:00: root bus resource [io 0x2007e00000000-0x2007e0fffffff] (bus address [0x0000-0xfffffff])
> pci_bus 0000:00: root bus resource [mem 0x2000000100000-0x200007effffff] (bus address [0x00100000-0x7effffff])
> pci_bus 0000:00: root bus resource [mem 0x2000100000000-0x2000dffffffff] (bus address [0x100000000-0xdffffffff])
> ...
> @@ -733,30 +733,32 @@ int pcibios_enable_device(struct pci_dev *dev, int mask)
> static int __pci_mmap_make_offset_bus(struct pci_dev *pdev, struct vm_area_struct *vma,
> enum pci_mmap_state mmap_state)
> {
> - struct pci_pbm_info *pbm = pdev->dev.archdata.host_controller;
> - unsigned long space_size, user_offset, user_size;
> -
> - if (mmap_state == pci_mmap_io) {
> - space_size = resource_size(&pbm->io_space);
> - } else {
> - space_size = resource_size(&pbm->mem_space);
> - }
> + unsigned long user_offset, user_size;
> + struct resource res, *root_bus_res;
> + struct pci_bus_region region;
> + struct pci_bus *bus;
>
> /* Make sure the request is in range. */
> user_offset = vma->vm_pgoff << PAGE_SHIFT;
> user_size = vma->vm_end - vma->vm_start;
>
> - if (user_offset >= space_size ||
> - (user_offset + user_size) > space_size)
> + region.start = user_offset;
> + region.end = user_offset + user_size - 1;
> + memset(&res, 0, sizeof(res));
> + if (mmap_state == pci_mmap_io)
> + res.flags = IORESOURCE_IO;
> + else
> + res.flags = IORESOURCE_MEM;
> +
> + pcibios_bus_to_resource(pdev->bus, &res, ®ion);
> + bus = pdev->bus;
> + while (bus->parent)
> + bus = bus->parent;
> + root_bus_res = pci_find_bus_resource(bus, &res);
> + if (!root_bus_res)
> return -EINVAL;
>
> - if (mmap_state == pci_mmap_io) {
> - vma->vm_pgoff = (pbm->io_space.start +
> - user_offset) >> PAGE_SHIFT;
> - } else {
> - vma->vm_pgoff = (pbm->mem_space.start +
> - user_offset) >> PAGE_SHIFT;
> - }
> + vma->vm_pgoff = res.start >> PAGE_SHIFT;
>
> return 0;
> }
I'm kind of confused here. There are two ways to mmap PCI BARs:
/proc/bus/pci/00/02.0 (proc_bus_pci_mmap()):
all BARs in one file; MEM/IO determined by ioctl()
mmap offset is a CPU physical address in the PCI resource
/sys/devices/pci0000:00/0000:00:02.0/resource0 (pci_mmap_resource()):
one file per BAR; MEM/IO determined by BAR type
mmap offset is between 0 and BAR size
Both proc_bus_pci_mmap() and pci_mmap_resource() validate the
requested area with pci_mmap_fits() before calling pci_mmap_page_range().
In the proc_bus_pci_mmap() path, the offset in vma->vm_pgoff must be
within the pdev->resource[], so the user must be supplying a CPU
physical address (not an address obtained from pci_resource_to_user()).
That vma->vm_pgoff is passed unchanged to pci_mmap_page_range().
In the pci_mmap_resource() path, vma->vm_pgoff must be between 0 and
the BAR size. Then we add in the pci_resource_to_user() information
before passing it to pci_mmap_page_range(). The comment in
pci_mmap_resource() says pci_mmap_page_range() expects a "user
visible" address, but I don't really believe that based on how
proc_bus_pci_mmap() works.
Do both proc_bus_pci_mmap() and pci_mmap_resource() work on sparc?
It looks like they call pci_mmap_page_range() with different
assumptions, so I don't see how they can both work.
In any case, pci_mmap_page_range() on sparc converts that
vma->vm_pgoff back to a CPU physical address, so there's an awful lot
of work going on in the pci_mmap_resource() path to convert the mmap
offset to a "user" address and then convert it back again.
There's also quite a bit of validation done in the arch code (in
__pci_mmap_make_offset() and __pci_mmap_make_offset_bus()) that looks
partly redundant with pci_mmap_fits() and not necessarily
arch-specific.
As far as I can see, pci_mmap_resource() doesn't need to have any
connection at all with pci_resource_to_user(). All it needs is the
pdev->resource[] (which has the CPU physical address of the BAR) and
vma->vm_pgoff (the offset into the BAR).
I don't think pci_mmap_resource() should call pci_resource_to_user(),
and I think pci_mmap_page_range() should expect a normal VMA that
contains a valid CPU PFN in vm_pgoff (or a valid CPU I/O port number,
in the case of an I/O port mmap).
The original pci_resource_to_user() was added for powerpc by
2311b1f2bbd3 ("[PATCH] PCI: fix-pci-mmap-on-ppc-and-ppc64.patch"),
but I couldn't find the linux-pci discussion it mentions.
> @@ -977,16 +979,12 @@ void pci_resource_to_user(const struct pci_dev *pdev, int bar,
> const struct resource *rp, resource_size_t *start,
> resource_size_t *end)
> {
> - struct pci_pbm_info *pbm = pdev->dev.archdata.host_controller;
> - unsigned long offset;
> + struct pci_bus_region region;
>
> - if (rp->flags & IORESOURCE_IO)
> - offset = pbm->io_space.start;
> - else
> - offset = pbm->mem_space.start;
> + pcibios_resource_to_bus(pdev->bus, ®ion, (struct resource *)rp);
>
> - *start = rp->start - offset;
> - *end = rp->end - offset;
> + *start = region.start;
> + *end = region.end;
> }
next prev parent reply other threads:[~2016-04-22 20:49 UTC|newest]
Thread overview: 86+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-04-08 0:15 [PATCH v11 00/60] PCI: Resource allocation cleanup for v4.7 Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 01/60] PCI: Fix iomem_is_exclusive() checking in pci_mmap_resource() Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 02/60] alpha/PCI: Only check iomem_is_exclusive() for IORESOURCE_MEM, not IORESOURCE_IO Yinghai Lu
2016-04-25 21:01 ` Bjorn Helgaas
2016-04-08 0:15 ` [PATCH v11 03/60] PCI: Add pci_find_bus_resource() Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 04/60] sparc/PCI: Use correct offset for bus address to resource Yinghai Lu
2016-04-22 20:49 ` Bjorn Helgaas [this message]
2016-04-28 4:55 ` Yinghai Lu
2016-04-28 13:56 ` Bjorn Helgaas
2016-04-29 7:19 ` Yinghai Lu
2016-05-03 22:52 ` Yinghai Lu
2016-05-04 0:37 ` Benjamin Herrenschmidt
2016-05-04 1:25 ` Bjorn Helgaas
2016-05-04 5:08 ` Yinghai Lu
2016-05-04 5:52 ` Yinghai Lu
2016-05-04 15:17 ` Bjorn Helgaas
2016-05-04 18:46 ` Yinghai Lu
2016-05-05 0:25 ` Yinghai Lu
2016-05-05 15:53 ` Yinghai Lu
2016-05-05 22:02 ` Benjamin Herrenschmidt
2016-05-06 0:56 ` Yinghai Lu
2016-05-06 4:18 ` Yinghai Lu
2016-05-06 18:26 ` Bjorn Helgaas
2016-05-10 6:18 ` Yinghai Lu
2016-05-04 4:17 ` David Miller
2016-04-08 0:15 ` [PATCH v11 05/60] sparc/PCI: Reserve legacy mmio after PCI mmio Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 06/60] sparc/PCI: Add IORESOURCE_MEM_64 for 64-bit resource in OF parsing Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 07/60] sparc/PCI: Keep resource idx order with bridge register number Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 08/60] PCI: Kill wrong quirk about M7101 Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 09/60] powerpc/PCI: Keep resource idx order with bridge register number Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 10/60] powerpc/PCI: Add IORESOURCE_MEM_64 for 64-bit resource in OF parsing Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 11/60] OF/PCI: Add IORESOURCE_MEM_64 for 64-bit resource Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 12/60] PCI: Check pref compatible bit for mem64 resource of PCIe device Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 13/60] PCI: Only treat non-pref mmio64 as pref if all bridges have MEM_64 Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 14/60] PCI: Add has_mem64 for struct host_bridge Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 15/60] PCI: Only treat non-pref mmio64 as pref if host bridge has mmio64 Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 16/60] PCI: Restore pref MMIO allocation logic for host bridge without mmio64 Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 17/60] PCI: Don't release fixed resource for realloc Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 18/60] PCI: Claim fixed resource during remove/rescan path Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 19/60] PCI: Set resource to FIXED for LSI devices Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 20/60] PCI: Separate realloc list checking after allocation Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 21/60] PCI: Treat optional as required in first try for bridge rescan Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 22/60] PCI: Get new realloc size for bridge for last try Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 23/60] PCI: Don't release sibling bridge resources during hotplug Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 24/60] PCI: Cleanup res_to_dev_res() printout Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 25/60] PCI: Reuse res_to_dev_res() in reassign_resources_sorted() Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 26/60] PCI: Use correct align for optional only resources during sorting Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 27/60] PCI: Optimize bus min_align/size calculation during sizing Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 28/60] PCI: Optimize bus align/size calculation for optional " Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 29/60] PCI: Don't add too much optional size for hotplug bridge MMIO Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 30/60] PCI: Reorder resources list for required/optional resources Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 31/60] PCI: Remove duplicated code for resource sorting Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 32/60] PCI: Rename pdev_sort_resources() to pdev_assign_resources_prepare() Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 33/60] PCI: Treat ROM resource as optional during realloc Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 34/60] PCI: Add debug printout during releasing partial assigned resources Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 35/60] PCI: Simplify res reference using in __assign_resources_sorted() Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 36/60] PCI: Add __add_to_list() Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 37/60] PCI: Cache window alignment value during bus sizing Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 38/60] PCI: Check if resource is allocated before trying to assign one Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 39/60] PCI: Separate out save_resources()/restore_resources() Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 40/60] PCI: Move comment to pci_need_to_release() Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 41/60] PCI: Separate required+optional assigning to another function Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 42/60] PCI: Skip required+optional if there is no optional Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 43/60] PCI: Move saved required resource list out of required+optional assigning Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 44/60] PCI: Add alt_size ressource allocation support Yinghai Lu
2016-04-08 0:56 ` Linus Torvalds
2016-04-08 5:50 ` Yinghai Lu
2016-04-08 6:24 ` Benjamin Herrenschmidt
2016-04-08 0:15 ` [PATCH v11 45/60] PCI: Add support for more than two alt_size entries under same bridge Yinghai Lu
2016-04-08 0:15 ` [PATCH v11 46/60] PCI: Fix size calculation with old_size on rescan path Yinghai Lu
2016-04-08 0:16 ` [PATCH v11 47/60] PCI: Don't add too much optional size for hotplug bridge io Yinghai Lu
2016-04-08 0:16 ` [PATCH v11 48/60] PCI: Move ISA io port align out of calculate_iosize() Yinghai Lu
2016-04-08 0:16 ` [PATCH v11 49/60] PCI: Don't add too much io port for hotplug bridge with old size Yinghai Lu
2016-04-08 0:16 ` [PATCH v11 50/60] PCI: Unify calculate_size() for io port and MMIO Yinghai Lu
2016-04-08 0:16 ` [PATCH v11 51/60] PCI: Allow bridge optional only io port resource required size to be 0 Yinghai Lu
2016-04-08 0:16 ` [PATCH v11 52/60] PCI: Unify skip_ioresource_align() Yinghai Lu
2016-04-08 0:16 ` [PATCH v11 53/60] PCI: Kill macro checking for bus io port sizing Yinghai Lu
2016-04-08 0:16 ` [PATCH v11 54/60] resources: Make allocate_resource() return best fit resource Yinghai Lu
2016-04-08 0:16 ` [PATCH v11 55/60] PCI, x86: Allocate from high in available window for MMIO Yinghai Lu
2016-04-08 0:16 ` [PATCH v11 56/60] PCI: Add debug print out for min_align and alt_size Yinghai Lu
2016-04-08 0:16 ` [PATCH v11 57/60] PCI, x86: Add pci=assign_pref_bars to reallocate pref BARs Yinghai Lu
2016-04-08 0:16 ` [PATCH v11 58/60] PCI: Introduce resource_disabled() Yinghai Lu
2016-04-08 0:16 ` [PATCH v11 59/60] PCI: Don't set flags to 0 when assign resource fail Yinghai Lu
2016-04-08 0:16 ` [PATCH v11 60/60] PCI: Only try to assign io port only for root bus that support it Yinghai Lu
2016-04-08 0:51 ` [PATCH v11 00/60] PCI: Resource allocation cleanup for v4.7 Linus Torvalds
2016-04-09 5:29 ` Yinghai Lu
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=20160422204920.GA17215@localhost \
--to=helgaas@kernel.org \
--cc=benh@kernel.crashing.org \
--cc=bhelgaas@google.com \
--cc=davem@davemloft.net \
--cc=khalid.aziz@oracle.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-pci@vger.kernel.org \
--cc=linux@iam.tj \
--cc=mpe@ellerman.id.au \
--cc=torvalds@linux-foundation.org \
--cc=wangyijing@huawei.com \
--cc=weiyang@linux.vnet.ibm.com \
--cc=yinghai@kernel.org \
/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).