From: Khalid Aziz <khalid.aziz@oracle.com>
To: Bjorn Helgaas <helgaas@kernel.org>, Yinghai Lu <yinghai@kernel.org>
Cc: Bjorn Helgaas <bhelgaas@google.com>,
David Miller <davem@davemloft.net>,
Benjamin Herrenschmidt <benh@kernel.crashing.org>,
Wei Yang <weiyang@linux.vnet.ibm.com>, TJ <linux@iam.tj>,
Yijing Wang <wangyijing@huawei.com>,
linux-pci@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH v7 02/60] sparc/PCI: Use correct bus address to resource offset
Date: Tue, 20 Oct 2015 14:07:43 -0600 [thread overview]
Message-ID: <56269F0F.7000309@oracle.com> (raw)
In-Reply-To: <20151020195413.GH8224@localhost>
On 10/20/2015 01:54 PM, Bjorn Helgaas wrote:
> On Thu, Oct 08, 2015 at 02:38:21PM -0700, Yinghai Lu wrote:
>> After we add 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 mem_space.start != offset.
>>
>> Use child_phys_addr to calculate exact offset and recorder offset in
>> pbm. Still assume all mem have same offset, otherwise panic early.
>>
>> 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])
>
> I think this changelog is incorrect. It suggests that this fixes a
> problem with 64-bit apertures, but in this example, *both* mem
> apertures are different from what they would have been before this
> patch.
>
> Before this patch, we used the mem32 aperture base (0x2000000100000)
> as the translation offset for both the mem32 and mem64 apertures, so I
> think we would have had this:
>
> pci_bus 0000:00: root bus resource [mem 0x2000000100000-0x200007effffff] (bus address [0x00000000-0x7eefffff])
> pci_bus 0000:00: root bus resource [mem 0x2000100000000-0x2000dffffffff] (bus address [0xfff00000-0xdffefffff])
>
> After this patch, you compute the real mem32 offset (0x2000000000000)
> and use it for both mem32 and mem64:
>
> 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])
>
>> -v2: to make is simple, do not add mem64_offset, and assume
>> mem64_offset == mem_offset, otherwise would make
>> pci_mmap_resource() path too complicated.
>
> I missed this comment before, and I suppose it has something to do
> with why you decided to panic if the mem64 offset is different from
> the mem32 offset. But I don't understand the reasoning. Other
> architectures handle pci_mmap_resource() just fine, ... oh, I see,
> this is related to the pci_resource_to_user() messiness.
>
> I don't want to argue about pci_resource_to_user() right now, so I
> guess you can just panic if the mem64 offset is different from the
> mem32 offset. But you should at least add a comment about why you're
> doing that. Otherwise it looks like you were just lazy.
I just hit this condition on a sparc platform where mem32 offset !=
mem64 offset. We need to handle this case, and not just call panic. I
had not seen this on 5 platforms and another 3 ldoms I had tested this
on when I tested this code earlier. I have sent panic log to Yinghai
separately.
--
Khalid
>
>> Signed-off-by: Yinghai Lu <yinghai@kernel.org>
>> ---
>> arch/sparc/kernel/pci.c | 14 +++++++-------
>> arch/sparc/kernel/pci_common.c | 36 +++++++++++++++++++++++++++---------
>> arch/sparc/kernel/pci_impl.h | 3 +++
>> 3 files changed, 37 insertions(+), 16 deletions(-)
>>
>> diff --git a/arch/sparc/kernel/pci.c b/arch/sparc/kernel/pci.c
>> index badf095..451792e 100644
>> --- a/arch/sparc/kernel/pci.c
>> +++ b/arch/sparc/kernel/pci.c
>> @@ -654,12 +654,12 @@ struct pci_bus *pci_scan_one_pbm(struct pci_pbm_info *pbm,
>> printk("PCI: Scanning PBM %s\n", node->full_name);
>>
>> pci_add_resource_offset(&resources, &pbm->io_space,
>> - pbm->io_space.start);
>> + pbm->io_offset);
>> pci_add_resource_offset(&resources, &pbm->mem_space,
>> - pbm->mem_space.start);
>> + pbm->mem_offset);
>> if (pbm->mem64_space.flags)
>> pci_add_resource_offset(&resources, &pbm->mem64_space,
>> - pbm->mem_space.start);
>> + pbm->mem_offset);
>> pbm->busn.start = pbm->pci_first_busno;
>> pbm->busn.end = pbm->pci_last_busno;
>> pbm->busn.flags = IORESOURCE_BUS;
>> @@ -751,10 +751,10 @@ static int __pci_mmap_make_offset_bus(struct pci_dev *pdev, struct vm_area_struc
>> return -EINVAL;
>>
>> if (mmap_state == pci_mmap_io) {
>> - vma->vm_pgoff = (pbm->io_space.start +
>> + vma->vm_pgoff = (pbm->io_offset +
>> user_offset) >> PAGE_SHIFT;
>> } else {
>> - vma->vm_pgoff = (pbm->mem_space.start +
>> + vma->vm_pgoff = (pbm->mem_offset +
>> user_offset) >> PAGE_SHIFT;
>> }
>>
>> @@ -981,9 +981,9 @@ void pci_resource_to_user(const struct pci_dev *pdev, int bar,
>> unsigned long offset;
>>
>> if (rp->flags & IORESOURCE_IO)
>> - offset = pbm->io_space.start;
>> + offset = pbm->io_offset;
>> else
>> - offset = pbm->mem_space.start;
>> + offset = pbm->mem_offset;
>>
>> *start = rp->start - offset;
>> *end = rp->end - offset;
>> diff --git a/arch/sparc/kernel/pci_common.c b/arch/sparc/kernel/pci_common.c
>> index 33524c1..28e976a 100644
>> --- a/arch/sparc/kernel/pci_common.c
>> +++ b/arch/sparc/kernel/pci_common.c
>> @@ -393,6 +393,7 @@ static void pci_register_iommu_region(struct pci_pbm_info *pbm)
>> void pci_determine_mem_io_space(struct pci_pbm_info *pbm)
>> {
>> const struct linux_prom_pci_ranges *pbm_ranges;
>> + resource_size_t mem64_offset = 0;
>> int i, saw_mem, saw_io;
>> int num_pbm_ranges;
>>
>> @@ -410,13 +411,16 @@ void pci_determine_mem_io_space(struct pci_pbm_info *pbm)
>>
>> for (i = 0; i < num_pbm_ranges; i++) {
>> const struct linux_prom_pci_ranges *pr = &pbm_ranges[i];
>> - unsigned long a, size;
>> + unsigned long a, size, region_a;
>> u32 parent_phys_hi, parent_phys_lo;
>> + u32 child_phys_mid, child_phys_lo;
>> u32 size_hi, size_lo;
>> int type;
>>
>> parent_phys_hi = pr->parent_phys_hi;
>> parent_phys_lo = pr->parent_phys_lo;
>> + child_phys_mid = pr->child_phys_mid;
>> + child_phys_lo = pr->child_phys_lo;
>> if (tlb_type == hypervisor)
>> parent_phys_hi &= 0x0fffffff;
>>
>> @@ -426,6 +430,8 @@ void pci_determine_mem_io_space(struct pci_pbm_info *pbm)
>> type = (pr->child_phys_hi >> 24) & 0x3;
>> a = (((unsigned long)parent_phys_hi << 32UL) |
>> ((unsigned long)parent_phys_lo << 0UL));
>> + region_a = (((unsigned long)child_phys_mid << 32UL) |
>> + ((unsigned long)child_phys_lo << 0UL));
>> size = (((unsigned long)size_hi << 32UL) |
>> ((unsigned long)size_lo << 0UL));
>>
>> @@ -440,6 +446,7 @@ void pci_determine_mem_io_space(struct pci_pbm_info *pbm)
>> pbm->io_space.start = a;
>> pbm->io_space.end = a + size - 1UL;
>> pbm->io_space.flags = IORESOURCE_IO;
>> + pbm->io_offset = a - region_a;
>> saw_io = 1;
>> break;
>>
>> @@ -448,6 +455,7 @@ void pci_determine_mem_io_space(struct pci_pbm_info *pbm)
>> pbm->mem_space.start = a;
>> pbm->mem_space.end = a + size - 1UL;
>> pbm->mem_space.flags = IORESOURCE_MEM;
>> + pbm->mem_offset = a - region_a;
>> saw_mem = 1;
>> break;
>>
>> @@ -456,6 +464,7 @@ void pci_determine_mem_io_space(struct pci_pbm_info *pbm)
>> pbm->mem64_space.start = a;
>> pbm->mem64_space.end = a + size - 1UL;
>> pbm->mem64_space.flags = IORESOURCE_MEM;
>> + mem64_offset = a - region_a;
>> saw_mem = 1;
>> break;
>>
>> @@ -471,14 +480,23 @@ void pci_determine_mem_io_space(struct pci_pbm_info *pbm)
>> prom_halt();
>> }
>>
>> - printk("%s: PCI IO[%llx] MEM[%llx]",
>> - pbm->name,
>> - pbm->io_space.start,
>> - pbm->mem_space.start);
>> - if (pbm->mem64_space.flags)
>> - printk(" MEM64[%llx]",
>> - pbm->mem64_space.start);
>> - printk("\n");
>> + if (pbm->io_space.flags)
>> + printk("%s: PCI IO %pR offset %llx\n",
>> + pbm->name, &pbm->io_space, pbm->io_offset);
>> + if (pbm->mem_space.flags)
>> + printk("%s: PCI MEM %pR offset %llx\n",
>> + pbm->name, &pbm->mem_space, pbm->mem_offset);
>> + if (pbm->mem64_space.flags) {
>> + if (pbm->mem_space.flags) {
>> + if (mem64_offset != pbm->mem_offset)
>> + panic("mem offset %llx != mem64 offset %llx\n",
>> + pbm->mem_offset, mem64_offset);
>> + } else
>> + pbm->mem_offset = mem64_offset;
>> +
>> + printk("%s: PCI MEM64 %pR offset %llx\n",
>> + pbm->name, &pbm->mem64_space, pbm->mem_offset);
>> + }
>>
>> pbm->io_space.name = pbm->mem_space.name = pbm->name;
>> pbm->mem64_space.name = pbm->name;
>> diff --git a/arch/sparc/kernel/pci_impl.h b/arch/sparc/kernel/pci_impl.h
>> index 37222ca..08b74af 100644
>> --- a/arch/sparc/kernel/pci_impl.h
>> +++ b/arch/sparc/kernel/pci_impl.h
>> @@ -99,6 +99,9 @@ struct pci_pbm_info {
>> struct resource mem_space;
>> struct resource mem64_space;
>> struct resource busn;
>> + /* offset */
>> + resource_size_t io_offset;
>> + resource_size_t mem_offset;
>>
>> /* Base of PCI Config space, can be per-PBM or shared. */
>> unsigned long config_space;
>> --
>> 1.8.4.5
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-pci" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at http://vger.kernel.org/majordomo-info.html
next prev parent reply other threads:[~2015-10-20 20:08 UTC|newest]
Thread overview: 101+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-10-08 21:38 [PATCH v7 00/60] PCI: Resource allocation cleanup for v4.4 Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 01/60] sparc/PCI: Add mem64 resource parsing for root bus Yinghai Lu
2015-10-09 16:15 ` Khalid Aziz
2015-10-20 16:42 ` Bjorn Helgaas
2015-10-08 21:38 ` [PATCH v7 02/60] sparc/PCI: Use correct bus address to resource offset Yinghai Lu
2015-10-09 16:16 ` Khalid Aziz
2015-10-20 16:45 ` Bjorn Helgaas
2015-10-20 19:54 ` Bjorn Helgaas
2015-10-20 20:07 ` Khalid Aziz [this message]
2015-10-20 20:34 ` Yinghai Lu
2015-10-20 20:49 ` Khalid Aziz
2015-10-20 23:51 ` Khalid Aziz
2015-10-21 0:03 ` Yinghai Lu
2015-10-21 0:25 ` Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 03/60] sparc/PCI: Unify pci_register_region() Yinghai Lu
2015-10-09 16:16 ` Khalid Aziz
2015-10-20 21:57 ` Bjorn Helgaas
2015-10-20 22:16 ` Yinghai Lu
2015-10-20 22:30 ` Yinghai Lu
2015-10-21 2:35 ` Bjorn Helgaas
2015-10-21 18:16 ` Yinghai Lu
2015-10-21 18:50 ` Bjorn Helgaas
2015-10-21 22:32 ` Yinghai Lu
2015-10-21 22:59 ` Yinghai Lu
2015-10-22 1:15 ` Yinghai Lu
2015-10-22 19:56 ` Yinghai Lu
2015-10-22 21:43 ` Yinghai Lu
2015-10-22 1:27 ` David Miller
2015-10-22 3:47 ` Benjamin Herrenschmidt
2015-10-23 5:38 ` Bjorn Helgaas
2015-10-08 21:38 ` [PATCH v7 04/60] sparc/PCI: Reserve legacy mmio after PCI mmio Yinghai Lu
2015-10-09 16:19 ` Khalid Aziz
2015-10-08 21:38 ` [PATCH v7 05/60] sparc/PCI: Add IORESOURCE_MEM_64 for 64-bit resource in OF parsing Yinghai Lu
2015-10-09 16:19 ` Khalid Aziz
2015-10-08 21:38 ` [PATCH v7 06/60] sparc/PCI: Keep resource idx order with bridge register number Yinghai Lu
2015-10-09 16:19 ` Khalid Aziz
2015-10-08 21:38 ` [PATCH v7 07/60] PCI: kill wrong quirk about M7101 Yinghai Lu
2015-10-09 16:20 ` Khalid Aziz
2015-10-12 19:46 ` Meelis Roos
2015-10-12 22:41 ` Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 08/60] powerpc/PCI: Keep resource idx order with bridge register number Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 09/60] powerpc/PCI: Add IORESOURCE_MEM_64 for 64-bit resource in OF parsing Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 10/60] OF/PCI: Add IORESOURCE_MEM_64 for 64-bit resource Yinghai Lu
2015-10-09 16:20 ` Khalid Aziz
2015-10-08 21:38 ` [PATCH v7 11/60] PCI: Check pref compatible bit for mem64 resource of PCIe device Yinghai Lu
2015-10-09 16:20 ` Khalid Aziz
2015-10-08 21:38 ` [PATCH v7 12/60] PCI: Only treat non-pref mmio64 as pref if all bridges have MEM_64 Yinghai Lu
2015-10-09 16:21 ` Khalid Aziz
2015-10-08 21:38 ` [PATCH v7 13/60] PCI: Add has_mem64 for struct host_bridge Yinghai Lu
2015-10-09 16:21 ` Khalid Aziz
2015-10-08 21:38 ` [PATCH v7 14/60] PCI: Only treat non-pref mmio64 as pref if host bridge has mmio64 Yinghai Lu
2015-10-09 16:21 ` Khalid Aziz
2015-10-08 21:38 ` [PATCH v7 15/60] PCI: Restore pref MMIO allocation logic for host bridge without mmio64 Yinghai Lu
2015-10-09 16:22 ` Khalid Aziz
2015-10-08 21:38 ` [PATCH v7 16/60] PCI: Don't release fixed resource for realloc Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 17/60] PCI: Claim fixed resource during remove/rescan path Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 18/60] PCI: Set resource to FIXED for LSI devices Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 19/60] PCI: Separate realloc list checking after allocation Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 20/60] PCI: Treat optional as required in first try for bridge rescan Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 21/60] PCI: Get new realloc size for bridge for last try Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 22/60] PCI: Don't release sibling bridge resources during hotplug Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 23/60] PCI: Cleanup res_to_dev_res() printout Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 24/60] PCI: Reuse res_to_dev_res() in reassign_resources_sorted() Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 25/60] PCI: Use correct align for optional only resources during sorting Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 26/60] PCI: Optimize bus min_align/size calculation during sizing Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 27/60] PCI: Optimize bus align/size calculation for optional " Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 28/60] PCI: Don't add too much optional size for hotplug bridge MMIO Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 29/60] PCI: Reorder resources list for required/optional resources Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 30/60] PCI: Remove duplicated code for resource sorting Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 31/60] PCI: Rename pdev_sort_resources() to pdev_assign_resources_prepare() Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 32/60] PCI: Treat ROM resource as optional during realloc Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 33/60] PCI: Add debug printout during releasing partial assigned resources Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 34/60] PCI: Simplify res reference using in __assign_resources_sorted() Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 35/60] PCI: Add __add_to_list() Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 36/60] PCI: Cache window alignment value during bus sizing Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 37/60] PCI: Check if resource is allocated before trying to assign one Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 38/60] PCI: Separate out save_resources()/restore_resources() Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 39/60] PCI: Move comment to pci_need_to_release() Yinghai Lu
2015-10-08 21:38 ` [PATCH v7 40/60] PCI: Separate required+optional assigning to another function Yinghai Lu
2015-10-08 21:39 ` [PATCH v7 41/60] PCI: Skip required+optional if there is no optional Yinghai Lu
2015-10-08 21:39 ` [PATCH v7 42/60] PCI: Move saved required resource list out of required+optional assigning Yinghai Lu
2015-10-08 21:39 ` [PATCH v7 43/60] PCI: Add alt_size ressource allocation support Yinghai Lu
2015-10-08 21:39 ` [PATCH v7 44/60] PCI: Add support for more than two alt_size entries under same bridge Yinghai Lu
2015-10-08 21:39 ` [PATCH v7 45/60] PCI: Fix size calculation with old_size on rescan path Yinghai Lu
2015-10-08 21:39 ` [PATCH v7 46/60] PCI: Don't add too much optional size for hotplug bridge io Yinghai Lu
2015-10-08 21:39 ` [PATCH v7 47/60] PCI: Move ISA io port align out of calculate_iosize() Yinghai Lu
2015-10-08 21:39 ` [PATCH v7 48/60] PCI: Don't add too much io port for hotplug bridge with old size Yinghai Lu
2015-10-08 21:39 ` [PATCH v7 49/60] PCI: Unify calculate_size() for io port and MMIO Yinghai Lu
2015-10-08 21:39 ` [PATCH v7 50/60] PCI: Allow bridge optional only io port resource required size to be 0 Yinghai Lu
2015-10-08 21:39 ` [PATCH v7 51/60] PCI: Unify skip_ioresource_align() Yinghai Lu
2015-10-13 9:45 ` [v7,51/60] " Michael Ellerman
2015-10-08 21:39 ` [PATCH v7 52/60] PCI: Kill macro checking for bus io port sizing Yinghai Lu
2015-10-08 21:39 ` [PATCH v7 53/60] resources: Split out __allocate_resource() Yinghai Lu
2015-10-08 21:39 ` [PATCH v7 54/60] resources: Make allocate_resource() return best fit resource Yinghai Lu
2015-10-08 21:39 ` [PATCH v7 55/60] PCI, x86: Allocate from high in available window for MMIO Yinghai Lu
2015-10-08 21:39 ` [PATCH v7 56/60] PCI: Add debug print out for min_align and alt_size Yinghai Lu
2015-10-08 21:39 ` [PATCH v7 57/60] PCI, x86: Add pci=assign_pref_bars to reallocate pref BARs Yinghai Lu
2015-10-08 21:39 ` [PATCH v7 58/60] PCI: Introduce resource_disabled() Yinghai Lu
2015-10-13 9:48 ` [v7,58/60] " Michael Ellerman
2015-10-08 21:39 ` [PATCH v7 59/60] PCI: Don't set flags to 0 when assign resource fail Yinghai Lu
2015-10-08 21:39 ` [PATCH v7 60/60] PCI: Only try to assign io port only for root bus that support it 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=56269F0F.7000309@oracle.com \
--to=khalid.aziz@oracle.com \
--cc=benh@kernel.crashing.org \
--cc=bhelgaas@google.com \
--cc=davem@davemloft.net \
--cc=helgaas@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-pci@vger.kernel.org \
--cc=linux@iam.tj \
--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).