platform/kernel/linux-rpi.git
3 years agoMerge branch 'pci/rockchip-dwc'
Bjorn Helgaas [Thu, 2 Sep 2021 19:56:45 +0000 (14:56 -0500)]
Merge branch 'pci/rockchip-dwc'

- Add Rockchip RK356X host controller driver (Simon Xue)

* pci/rockchip-dwc:
  PCI: rockchip-dwc: Add Rockchip RK356X host controller driver

3 years agoMerge branch 'pci/dwc'
Bjorn Helgaas [Thu, 2 Sep 2021 19:56:45 +0000 (14:56 -0500)]
Merge branch 'pci/dwc'

- Remove surplus break statement (Krzysztof Wilczyński)

* pci/dwc:
  PCI: dwc: Remove surplus break statement after return

3 years agoMerge branch 'pci/artpec6'
Bjorn Helgaas [Thu, 2 Sep 2021 19:56:45 +0000 (14:56 -0500)]
Merge branch 'pci/artpec6'

- Remove surplus break statement and local code block (Krzysztof
  Wilczyński)

* pci/artpec6:
  PCI: artpec6: Remove local code block from switch statement
  PCI: artpec6: Remove surplus break statement after return

3 years agoMerge branch 'pci/misc'
Bjorn Helgaas [Thu, 2 Sep 2021 19:56:44 +0000 (14:56 -0500)]
Merge branch 'pci/misc'

- Add pci_numachip_init() declaration (Krzysztof Wilczyński)

- Allocate pci_dev_str_match_path() string atomically (Dan Carpenter)

- Drop error message when Precision Time Measurement supported but not
  enabled (Jakub Kicinski)

- Correct the pci_iomap.h header guard #endif comment (Jonathan Cameron)

- Add schedule point in proc_bus_pci_read() (Krzysztof Wilczyński)

- Make saved capability state private to core (Bjorn Helgaas)

- Sync __pci_register_driver() stub for CONFIG_PCI=n (Andy Shevchenko)

- Convert sta2x11 from PCI-DMA-API to generic DMA-API (Christophe JAILLET)

* pci/misc:
  x86/PCI: sta2x11: switch from 'pci_' to 'dma_' API
  PCI: Sync __pci_register_driver() stub for CONFIG_PCI=n
  PCI: Make saved capability state private to core
  PCI: Add schedule point in proc_bus_pci_read()
  PCI: Correct the pci_iomap.h header guard #endif comment
  PCI/PTM: Remove error message at boot
  PCI: Fix pci_dev_str_match_path() alloc while atomic bug
  x86/PCI: Add pci_numachip_init() declaration

# Conflicts:
# include/linux/pci.h

3 years agoMerge branch 'pci/vpd'
Bjorn Helgaas [Thu, 2 Sep 2021 19:56:44 +0000 (14:56 -0500)]
Merge branch 'pci/vpd'

- Check Resource Item Names against those defined for type (Bjorn Helgaas)

- Treat initial 0xff as missing EEPROM (Heiner Kallweit)

- Reject resource tags with invalid size (Bjorn Helgaas)

- Don't check Large Resource Item Names for validity (Bjorn Helgaas)

- Allow access to valid parts of VPD if some is invalid (Bjorn Helgaas)

- Remove pci_vpd_size() old_size argument (Heiner Kallweit)

- Make pci_vpd_wait() uninterruptible (Heiner Kallweit)

- Remove struct pci_vpd.flag (Heiner Kallweit)

- Remove struct pci_vpd_ops (Heiner Kallweit)

- Remove struct pci_vpd.valid member (Heiner Kallweit)

- Embed struct pci_vpd in struct pci_dev (Heiner Kallweit)

- Determine VPD size in pci_vpd_init() (Heiner Kallweit)

- Treat invalid VPD like missing VPD capability (Heiner Kallweit)

- Add pci_vpd_alloc() to allocate buffer and read VPD into it (Heiner
  Kallweit)

- Add pci_vpd_find_ro_info_keyword() (Heiner Kallweit)

- Add pci_vpd_check_csum() (Heiner Kallweit)

- Add pci_vpd_find_id_string() (Heiner Kallweit)

- Read VPD with pci_vpd_alloc() (bnx2x, bnxt, sfc, sfc falcon, tg3 drivers)
  (Heiner Kallweit)

- Search VPD with pci_vpd_find_ro_info_keyword() (bnx2, bnx2x, bnxt, cxgb4,
  cxlflash SCSI, sfc, sfc falcon, tg3 drivers) (Heiner Kallweit)

- Search VPD with pci_vpd_find_id_string() (cxgb4 driver) (Heiner Kallweit)

- Validate VPD checksum with pci_vpd_check_csum() (cxgb4, tg3 drivers)
  (Heiner Kallweit)

- Replace open-coded byte swapping with swab32s() in bnx2 (Heiner Kallweit)

- Remove unused vpd_param member ec (Heiner Kallweit)

- Stop exporting pci_vpd_find_tag(), pci_vpd_find_info_keyword() (Heiner
  Kallweit)

- Move several VPD defines and inlines to internal PCI core (Heiner
  Kallweit)

* pci/vpd:
  PCI/VPD: Use unaligned access helpers
  PCI/VPD: Clean up public VPD defines and inline functions
  cxgb4: Use pci_vpd_find_id_string() to find VPD ID string
  PCI/VPD: Add pci_vpd_find_id_string()
  PCI/VPD: Include post-processing in pci_vpd_find_tag()
  PCI/VPD: Stop exporting pci_vpd_find_info_keyword()
  PCI/VPD: Stop exporting pci_vpd_find_tag()
  scsi: cxlflash: Search VPD with pci_vpd_find_ro_info_keyword()
  cxgb4: Search VPD with pci_vpd_find_ro_info_keyword()
  cxgb4: Remove unused vpd_param member ec
  cxgb4: Validate VPD checksum with pci_vpd_check_csum()
  bnxt: Search VPD with pci_vpd_find_ro_info_keyword()
  bnxt: Read VPD with pci_vpd_alloc()
  bnx2x: Search VPD with pci_vpd_find_ro_info_keyword()
  bnx2x: Read VPD with pci_vpd_alloc()
  bnx2: Replace open-coded byte swapping with swab32s()
  bnx2: Search VPD with pci_vpd_find_ro_info_keyword()
  sfc: falcon: Search VPD with pci_vpd_find_ro_info_keyword()
  sfc: falcon: Read VPD with pci_vpd_alloc()
  tg3: Search VPD with pci_vpd_find_ro_info_keyword()
  tg3: Validate VPD checksum with pci_vpd_check_csum()
  tg3: Read VPD with pci_vpd_alloc()
  sfc: Search VPD with pci_vpd_find_ro_info_keyword()
  sfc: Read VPD with pci_vpd_alloc()
  PCI/VPD: Add pci_vpd_check_csum()
  PCI/VPD: Add pci_vpd_find_ro_info_keyword()
  PCI/VPD: Add pci_vpd_alloc()
  PCI/VPD: Treat invalid VPD like missing VPD capability
  PCI/VPD: Determine VPD size in pci_vpd_init()
  PCI/VPD: Embed struct pci_vpd in struct pci_dev
  PCI/VPD: Remove struct pci_vpd.valid member
  PCI/VPD: Remove struct pci_vpd_ops
  PCI/VPD: Reorder pci_read_vpd(), pci_write_vpd()
  PCI/VPD: Remove struct pci_vpd.flag
  PCI/VPD: Make pci_vpd_wait() uninterruptible
  PCI/VPD: Remove pci_vpd_size() old_size argument
  PCI/VPD: Allow access to valid parts of VPD if some is invalid
  PCI/VPD: Don't check Large Resource Item Names for validity
  PCI/VPD: Reject resource tags with invalid size
  PCI/VPD: Treat initial 0xff as missing EEPROM
  PCI/VPD: Check Resource Item Names against those valid for type
  PCI/VPD: Correct diagnostic for VPD read failure

3 years agoMerge branch 'pci/virtualization'
Bjorn Helgaas [Thu, 2 Sep 2021 19:56:43 +0000 (14:56 -0500)]
Merge branch 'pci/virtualization'

- Add ACS quirks for NXP LX2xx0 and LX2xx2 platforms (Wasim Khan)

- Add ACS quirks for Cavium multi-function devices (George Cherian)

- Enforce pci=noats with Transaction Blocking (Alex Williamson)

* pci/virtualization:
  PCI/ACS: Enforce pci=noats with Transaction Blocking
  PCI: Add ACS quirks for Cavium multi-function devices
  PCI: Add ACS quirks for NXP LX2xx0 and LX2xx2 platforms

3 years agoMerge branch 'pci/resource'
Bjorn Helgaas [Thu, 2 Sep 2021 19:56:43 +0000 (14:56 -0500)]
Merge branch 'pci/resource'

- Refactor pci_ioremap_bar() and pci_ioremap_wc_bar() (Krzysztof
  Wilczyński)

- Optimize pci_resource_len() to reduce kernel size (Zhen Lei)

* pci/resource:
  PCI: Optimize pci_resource_len() to reduce kernel size
  PCI: Refactor pci_ioremap_bar() and pci_ioremap_wc_bar()

3 years agoMerge branch 'pci/reset'
Bjorn Helgaas [Thu, 2 Sep 2021 19:56:42 +0000 (14:56 -0500)]
Merge branch 'pci/reset'

- Cache PCIe Device Capabilities register (Amey Narkhede)

- Add pcie_reset_flr() with 'probe' argument (Amey Narkhede)

- Add pdev->reset_methods[] array to track reset method ordering (Amey
  Narkhede)

- Remove reset_fn field from pci_dev (Amey Narkhede)

- Add sysfs interface to query and set device reset mechanism (Amey
  Narkhede)

- Add pci_set_acpi_fwnode() to set ACPI_COMPANION (Shanker Donthineni)

- Use acpi_pci_power_manageable() instead of duplicating logic (Shanker
  Donthineni)

- Set ACPI fwnode early and at the same time with OF (Shanker Donthineni)

- Add support for ACPI _RST reset method (Shanker Donthineni)

- Change reset function 'probe' argument to bool (Amey Narkhede)

* pci/reset:
  PCI: Change the type of probe argument in reset functions
  PCI: Add support for ACPI _RST reset method
  PCI: Setup ACPI fwnode early and at the same time with OF
  PCI: Use acpi_pci_power_manageable()
  PCI: Add pci_set_acpi_fwnode() to set ACPI_COMPANION
  PCI: Allow userspace to query and set device reset mechanism
  PCI: Remove reset_fn field from pci_dev
  PCI: Add array to track reset method ordering
  PCI: Add pcie_reset_flr() with 'probe' argument
  PCI: Cache PCIe Device Capabilities register

3 years agoMerge branch 'pci/portdrv'
Bjorn Helgaas [Thu, 2 Sep 2021 19:56:42 +0000 (14:56 -0500)]
Merge branch 'pci/portdrv'

- Enable Bandwidth Notification only if port supports it (Stuart Hayes)

* pci/portdrv:
  PCI/portdrv: Enable Bandwidth Notification only if port supports it

3 years agoMerge branch 'pci/irq'
Bjorn Helgaas [Thu, 2 Sep 2021 19:56:42 +0000 (14:56 -0500)]
Merge branch 'pci/irq'

- Convert irq_find_mapping() + generic_handle_irq() to
  generic_handle_domain_irq() (Marc Zyngier)

* pci/irq:
  PCI: Bulk conversion to generic_handle_domain_irq()

3 years agoMerge branch 'pci/iommu'
Bjorn Helgaas [Thu, 2 Sep 2021 19:56:41 +0000 (14:56 -0500)]
Merge branch 'pci/iommu'

- Allow PASID on fake PCIe devices, e.g., HiSilicon KunPeng920 and
  KunPeng930 AMBA devices, without TLP prefixes (Zhangfei Gao)

- Allow SVA / dma-can-stall on fake PCIe devices (Zhangfei Gao)

* pci/iommu:
  PCI: Set dma-can-stall for HiSilicon chips
  PCI: Allow PASID on fake PCIe devices without TLP prefixes

3 years agoMerge branch 'pci/hotplug'
Bjorn Helgaas [Thu, 2 Sep 2021 19:56:41 +0000 (14:56 -0500)]
Merge branch 'pci/hotplug'

- Fix ibmphp double unmap of io_mem (Vishal Aslot)

* pci/hotplug:
  PCI: ibmphp: Fix double unmap of io_mem

3 years agoMerge branch 'pci/enumeration'
Bjorn Helgaas [Thu, 2 Sep 2021 19:56:40 +0000 (14:56 -0500)]
Merge branch 'pci/enumeration'

- Call Max Payload Size-related fixup quirks early, so they're considered
  by pci_configure_mps() (Marek Behún)

- Restrict Max Payload Size Supported to work around ASMedia ASM1062 SATA
  erratum (Marek Behún)

- Return ~0 data on pciconfig_read() CAP_SYS_ADMIN failure (Krzysztof
  Wilczyński)

* pci/enumeration:
  PCI: Return int from pciconfig_read() syscall
  PCI: Return ~0 data on pciconfig_read() CAP_SYS_ADMIN failure
  PCI: Restrict ASMedia ASM1062 SATA Max Payload Size Supported
  PCI: Call Max Payload Size-related fixup quirks early

3 years agoPCI: ibmphp: Fix double unmap of io_mem
Vishal Aslot [Wed, 18 Aug 2021 16:57:51 +0000 (11:57 -0500)]
PCI: ibmphp: Fix double unmap of io_mem

ebda_rsrc_controller() calls iounmap(io_mem) on the error path. Its caller,
ibmphp_access_ebda(), also calls iounmap(io_mem) on good and error paths.

Remove the iounmap(io_mem) invocation from ebda_rsrc_controller().

[bhelgaas: remove item from TODO]
Link: https://lore.kernel.org/r/20210818165751.591185-1-os.vaslot@gmail.com
Signed-off-by: Vishal Aslot <os.vaslot@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agox86/PCI: sta2x11: switch from 'pci_' to 'dma_' API
Christophe JAILLET [Sun, 22 Aug 2021 18:49:20 +0000 (20:49 +0200)]
x86/PCI: sta2x11: switch from 'pci_' to 'dma_' API

The wrappers in include/linux/pci-dma-compat.h should go away.

The patch has been generated with the coccinelle script below.

It has been hand modified to use 'dma_set_mask_and_coherent()' instead of
'pci_set_dma_mask()/pci_set_consistent_dma_mask()' when applicable.
This is less verbose.

It has been compile tested.

@@
@@
-    PCI_DMA_BIDIRECTIONAL
+    DMA_BIDIRECTIONAL

@@
@@
-    PCI_DMA_TODEVICE
+    DMA_TO_DEVICE

@@
@@
-    PCI_DMA_FROMDEVICE
+    DMA_FROM_DEVICE

@@
@@
-    PCI_DMA_NONE
+    DMA_NONE

@@
expression e1, e2, e3;
@@
-    pci_alloc_consistent(e1, e2, e3)
+    dma_alloc_coherent(&e1->dev, e2, e3, GFP_)

@@
expression e1, e2, e3;
@@
-    pci_zalloc_consistent(e1, e2, e3)
+    dma_alloc_coherent(&e1->dev, e2, e3, GFP_)

@@
expression e1, e2, e3, e4;
@@
-    pci_free_consistent(e1, e2, e3, e4)
+    dma_free_coherent(&e1->dev, e2, e3, e4)

@@
expression e1, e2, e3, e4;
@@
-    pci_map_single(e1, e2, e3, e4)
+    dma_map_single(&e1->dev, e2, e3, e4)

@@
expression e1, e2, e3, e4;
@@
-    pci_unmap_single(e1, e2, e3, e4)
+    dma_unmap_single(&e1->dev, e2, e3, e4)

@@
expression e1, e2, e3, e4, e5;
@@
-    pci_map_page(e1, e2, e3, e4, e5)
+    dma_map_page(&e1->dev, e2, e3, e4, e5)

@@
expression e1, e2, e3, e4;
@@
-    pci_unmap_page(e1, e2, e3, e4)
+    dma_unmap_page(&e1->dev, e2, e3, e4)

@@
expression e1, e2, e3, e4;
@@
-    pci_map_sg(e1, e2, e3, e4)
+    dma_map_sg(&e1->dev, e2, e3, e4)

@@
expression e1, e2, e3, e4;
@@
-    pci_unmap_sg(e1, e2, e3, e4)
+    dma_unmap_sg(&e1->dev, e2, e3, e4)

@@
expression e1, e2, e3, e4;
@@
-    pci_dma_sync_single_for_cpu(e1, e2, e3, e4)
+    dma_sync_single_for_cpu(&e1->dev, e2, e3, e4)

@@
expression e1, e2, e3, e4;
@@
-    pci_dma_sync_single_for_device(e1, e2, e3, e4)
+    dma_sync_single_for_device(&e1->dev, e2, e3, e4)

@@
expression e1, e2, e3, e4;
@@
-    pci_dma_sync_sg_for_cpu(e1, e2, e3, e4)
+    dma_sync_sg_for_cpu(&e1->dev, e2, e3, e4)

@@
expression e1, e2, e3, e4;
@@
-    pci_dma_sync_sg_for_device(e1, e2, e3, e4)
+    dma_sync_sg_for_device(&e1->dev, e2, e3, e4)

@@
expression e1, e2;
@@
-    pci_dma_mapping_error(e1, e2)
+    dma_mapping_error(&e1->dev, e2)

@@
expression e1, e2;
@@
-    pci_set_dma_mask(e1, e2)
+    dma_set_mask(&e1->dev, e2)

@@
expression e1, e2;
@@
-    pci_set_consistent_dma_mask(e1, e2)
+    dma_set_coherent_mask(&e1->dev, e2)

Link: https://lore.kernel.org/r/99656452963ba3c63a6cb12e151279d81da365eb.1629658069.git.christophe.jaillet@wanadoo.fr
Link: https://lore.kernel.org/kernel-janitors/20200421081257.GA131897@infradead.org/
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/VPD: Use unaligned access helpers
Heiner Kallweit [Thu, 26 Aug 2021 18:58:07 +0000 (20:58 +0200)]
PCI/VPD: Use unaligned access helpers

Use unaligned access helpers to simplify the code.

Link: https://lore.kernel.org/r/0f1c7e21-5330-72ab-139d-f5ce3c65f04a@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/VPD: Clean up public VPD defines and inline functions
Heiner Kallweit [Thu, 26 Aug 2021 18:57:01 +0000 (20:57 +0200)]
PCI/VPD: Clean up public VPD defines and inline functions

After recent introduction of new VPD API functions and user migration
these defines and inline functions aren't used outside VPD core any
longer.

Link: https://lore.kernel.org/r/d33e06bf-bc5e-ece7-bf35-7245ae224d1b@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agocxgb4: Use pci_vpd_find_id_string() to find VPD ID string
Heiner Kallweit [Thu, 26 Aug 2021 18:56:22 +0000 (20:56 +0200)]
cxgb4: Use pci_vpd_find_id_string() to find VPD ID string

Use pci_vpd_find_id_string() to find the VPD ID string. This simplifies the
code and avoids the need for pci_vpd_lrdt_size().

Link: https://lore.kernel.org/r/19ea2e9b-6e94-288a-6612-88db01b1b417@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/VPD: Add pci_vpd_find_id_string()
Heiner Kallweit [Thu, 26 Aug 2021 18:55:43 +0000 (20:55 +0200)]
PCI/VPD: Add pci_vpd_find_id_string()

Add a pci_vpd_find_id_string() API function to retrieve the ID string from
VPD.

This way callers don't need pci_vpd_lrdt_size() any longer, and it can be
made private to the VPD core.

Link: https://lore.kernel.org/r/c5225bf6-8d29-970d-e271-0d7b52252630@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/VPD: Include post-processing in pci_vpd_find_tag()
Heiner Kallweit [Thu, 26 Aug 2021 18:55:07 +0000 (20:55 +0200)]
PCI/VPD: Include post-processing in pci_vpd_find_tag()

Move pci_vpd_find_tag() post-processing from pci_vpd_find_ro_info_keyword()
to pci_vpd_find_tag(). This simplifies function pci_vpd_find_id_string()
that will be added in a subsequent patch.

Link: https://lore.kernel.org/r/fb15393f-d3b2-e140-2643-570d3abd7382@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/VPD: Stop exporting pci_vpd_find_info_keyword()
Heiner Kallweit [Thu, 26 Aug 2021 18:54:23 +0000 (20:54 +0200)]
PCI/VPD: Stop exporting pci_vpd_find_info_keyword()

Now that the last users have been migrated to pci_vpd_find_ro_keyword()
we can stop exporting this function. It's still used in VPD core code.

Link: https://lore.kernel.org/r/96ca2a56-383e-9b61-9cba-4f1e5611dc15@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/VPD: Stop exporting pci_vpd_find_tag()
Heiner Kallweit [Thu, 26 Aug 2021 18:53:42 +0000 (20:53 +0200)]
PCI/VPD: Stop exporting pci_vpd_find_tag()

Now that the last users have been migrated to pci_vpd_find_ro_keyword()
we can stop exporting this function. It's still used in VPD core code.

Link: https://lore.kernel.org/r/71131eca-0502-7878-365f-30b6614161cf@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI: Set dma-can-stall for HiSilicon chips
Zhangfei Gao [Tue, 13 Jul 2021 02:54:36 +0000 (10:54 +0800)]
PCI: Set dma-can-stall for HiSilicon chips

HiSilicon KunPeng920 and KunPeng930 have devices that appear as PCI but are
actually on the AMBA bus. These fake PCI devices can support SVA via the
SMMU stall feature.

DT systems can indicate this in the device tree, but ACPI systems don't
have that mechanism, so add a "dma-can-stall" property manually for them.

[bhelgaas: add text from Robin as comment near quirk]
Link: https://lore.kernel.org/r/1626144876-11352-4-git-send-email-zhangfei.gao@linaro.org
Signed-off-by: Zhangfei Gao <zhangfei.gao@linaro.org>
Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Signed-off-by: Zhou Wang <wangzhou1@hisilicon.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Robin Murphy <robin.murphy@arm.com>
3 years agoPCI: rockchip-dwc: Add Rockchip RK356X host controller driver
Simon Xue [Fri, 25 Jun 2021 06:55:11 +0000 (14:55 +0800)]
PCI: rockchip-dwc: Add Rockchip RK356X host controller driver

Add a driver for the DesignWare-based PCIe controller found on
RK356X. The existing pcie-rockchip-host driver is only used for
the Rockchip-designed IP found on RK3399.

Link: https://lore.kernel.org/r/20210625065511.1096935-1-xxm@rock-chips.com
Tested-by: Peter Geis <pgwipeout@gmail.com>
Signed-off-by: Simon Xue <xxm@rock-chips.com>
Signed-off-by: Shawn Lin <shawn.lin@rock-chips.com>
Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Rob Herring <robh@kernel.org>
3 years agoPCI: dwc: Remove surplus break statement after return
Krzysztof Wilczyński [Thu, 1 Jul 2021 21:02:52 +0000 (21:02 +0000)]
PCI: dwc: Remove surplus break statement after return

As part of code refactoring completed in a0fd361db8e5 ("PCI: dwc: Move
"dbi", "dbi2", and "addr_space" resource setup into common code"),
dw_plat_add_pcie_ep() was removed and the call to the dw_pcie_ep_init() was
moved into dw_plat_pcie_probe().

This left a break statement behind that is not needed any more as as
dw_plat_pcie_probe() returns immediately after calling dw_pcie_ep_init().

Remove this surplus break statement that became dead code.

Suggested-by: Bjorn Helgaas <bhelgaas@google.com>
Link: https://lore.kernel.org/r/20210701210252.1638709-1-kw@linux.com
Signed-off-by: Krzysztof Wilczyński <kw@linux.com>
Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI: artpec6: Remove local code block from switch statement
Krzysztof Wilczyński [Thu, 1 Jul 2021 20:44:01 +0000 (20:44 +0000)]
PCI: artpec6: Remove local code block from switch statement

The switch statement in the artpec6_pcie_probe() has a local code block
where "val" is defined and immediately used by the artpec6_pcie_readl().

This extra code block adds brackets at the same indentation level as the
switch statement itself which can hinder readability of the code.

Move the "val" declaration to the top of the function and remove
the extra code block from the switch statement.

Suggested-by: Bjorn Helgaas <bhelgaas@google.com>
Link: https://lore.kernel.org/r/20210701204401.1636562-2-kw@linux.com
Signed-off-by: Krzysztof Wilczyński <kw@linux.com>
Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Jesper Nilsson <jesper.nilsson@axis.com>
3 years agoPCI: artpec6: Remove surplus break statement after return
Krzysztof Wilczyński [Thu, 1 Jul 2021 20:44:00 +0000 (20:44 +0000)]
PCI: artpec6: Remove surplus break statement after return

As part of code refactoring completed in a0fd361db8e5 ("PCI: dwc: Move
"dbi", "dbi2", and "addr_space" resource setup into common code"),
artpec6_add_pcie_ep() was removed and the call to the dw_pcie_ep_init()
was moved into artpec6_pcie_probe().

This left a break statement behind that is not needed any more as
artpec6_pcie_probe() returns immediately after calling dw_pcie_ep_init().

Remove this surplus break statement that became dead code.

Link: https://lore.kernel.org/r/20210701204401.1636562-1-kw@linux.com
Signed-off-by: Krzysztof Wilczyński <kw@linux.com>
Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Jesper Nilsson <jesper.nilsson@axis.com>
3 years agoPCI/portdrv: Enable Bandwidth Notification only if port supports it
Stuart Hayes [Wed, 12 May 2021 21:33:14 +0000 (03:03 +0530)]
PCI/portdrv: Enable Bandwidth Notification only if port supports it

Previously we assumed that all Root Ports and Switch Downstream Ports
supported Link Bandwidth Notification.  Per spec, this is only required
for Ports supporting Links wider than x1 and/or multiple Link speeds
(PCIe r5.0, sec 7.5.3.6).

Because we assumed all Ports supported it, we tried to set up a Bandwidth
Notification IRQ, which failed for devices that don't support IRQs at all,
which meant pcieport didn't attach to the Port at all.

Check the Link Bandwidth Notification Capability bit and enable the service
only when the Port supports it.

[bhelgaas: commit log]
Fixes: e8303bb7a75c ("PCI/LINK: Report degraded links via link bandwidth notification")
Link: https://lore.kernel.org/r/20210512213314.7778-1-stuart.w.hayes@gmail.com
Signed-off-by: Stuart Hayes <stuart.w.hayes@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Lukas Wunner <lukas@wunner.de>
Cc: stable@vger.kernel.org
3 years agoPCI: Allow PASID on fake PCIe devices without TLP prefixes
Zhangfei Gao [Tue, 13 Jul 2021 02:54:34 +0000 (10:54 +0800)]
PCI: Allow PASID on fake PCIe devices without TLP prefixes

Some systems, e.g., HiSilicon KunPeng920 and KunPeng930, have devices that
appear as PCI but are actually on the AMBA bus.  Some of these fake PCI
devices support a PASID-like feature and they do have a working PASID
capability even though they do not use the PCIe Transport Layer Protocol
and do not support TLP prefixes.

Add a pasid_no_tlp bit for this "PASID works without TLP prefixes" case and
update pci_enable_pasid() so it can enable PASID on these devices.

Set this bit for HiSilicon KunPeng920 and KunPeng930.

[bhelgaas: squashed, commit log]
Suggested-by: Bjorn Helgaas <bhelgaas@google.com>
Link: https://lore.kernel.org/r/1626144876-11352-2-git-send-email-zhangfei.gao@linaro.org
Link: https://lore.kernel.org/r/1626144876-11352-3-git-send-email-zhangfei.gao@linaro.org
Signed-off-by: Zhangfei Gao <zhangfei.gao@linaro.org>
Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Signed-off-by: Zhou Wang <wangzhou1@hisilicon.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoscsi: cxlflash: Search VPD with pci_vpd_find_ro_info_keyword()
Heiner Kallweit [Sun, 22 Aug 2021 14:01:08 +0000 (16:01 +0200)]
scsi: cxlflash: Search VPD with pci_vpd_find_ro_info_keyword()

Use pci_vpd_find_ro_info_keyword() to search for keywords in VPD to
simplify the code.

Link: https://lore.kernel.org/r/b5f71c97-61fb-86cb-6bec-84b042392ce7@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agocxgb4: Search VPD with pci_vpd_find_ro_info_keyword()
Heiner Kallweit [Sun, 22 Aug 2021 13:59:21 +0000 (15:59 +0200)]
cxgb4: Search VPD with pci_vpd_find_ro_info_keyword()

Use pci_vpd_find_ro_info_keyword() to search for keywords in VPD to
simplify the code.

Link: https://lore.kernel.org/r/db576a3e-e877-b37b-98ed-cfc03d225ab3@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agocxgb4: Remove unused vpd_param member ec
Heiner Kallweit [Sun, 22 Aug 2021 13:58:26 +0000 (15:58 +0200)]
cxgb4: Remove unused vpd_param member ec

Member ec isn't used, so remove it.

Link: https://lore.kernel.org/r/30648e95-bfb9-9af3-0c8f-dd3e34df8b6b@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agocxgb4: Validate VPD checksum with pci_vpd_check_csum()
Heiner Kallweit [Sun, 22 Aug 2021 13:57:27 +0000 (15:57 +0200)]
cxgb4: Validate VPD checksum with pci_vpd_check_csum()

Validate the VPD checksum with pci_vpd_check_csum() to simplify the code.

Link: https://lore.kernel.org/r/70404ece-0036-c0ce-f824-f5637e54115e@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agobnxt: Search VPD with pci_vpd_find_ro_info_keyword()
Heiner Kallweit [Sun, 22 Aug 2021 13:56:24 +0000 (15:56 +0200)]
bnxt: Search VPD with pci_vpd_find_ro_info_keyword()

Use pci_vpd_find_ro_info_keyword() to search for keywords in VPD to
simplify the code.

Link: https://lore.kernel.org/r/f062921c-ad33-3b3e-8ada-b53427a9cd4a@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agobnxt: Read VPD with pci_vpd_alloc()
Heiner Kallweit [Sun, 22 Aug 2021 13:55:23 +0000 (15:55 +0200)]
bnxt: Read VPD with pci_vpd_alloc()

Use pci_vpd_alloc() to dynamically allocate a properly sized buffer and
read the full VPD data into it.

This simplifies the code, and we no longer have to make assumptions about
VPD size.

Link: https://lore.kernel.org/r/62522a24-f39a-2b35-1577-1fbb41695bed@gmail.com
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agobnx2x: Search VPD with pci_vpd_find_ro_info_keyword()
Heiner Kallweit [Sun, 22 Aug 2021 13:54:23 +0000 (15:54 +0200)]
bnx2x: Search VPD with pci_vpd_find_ro_info_keyword()

Use pci_vpd_find_ro_info_keyword() to search for keywords in VPD to
simplify the code.

Use strncasecmp() to match Vendor ID instead of comparing with lower- and
upper-case hex string.

[bhelgaas: convert to strncasecmp()]
Link: https://lore.kernel.org/r/a9f730cf-e31e-902b-7b39-0ff2e99636e0@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agobnx2x: Read VPD with pci_vpd_alloc()
Heiner Kallweit [Sun, 22 Aug 2021 13:53:23 +0000 (15:53 +0200)]
bnx2x: Read VPD with pci_vpd_alloc()

Use pci_vpd_alloc() to dynamically allocate a properly sized buffer and
read the full VPD data into it.

This simplifies the code, and we no longer have to make assumptions about
VPD size.

Link: https://lore.kernel.org/r/821a334d-ff9d-386e-5f42-9b620ab3dbfa@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agobnx2: Replace open-coded byte swapping with swab32s()
Heiner Kallweit [Sun, 22 Aug 2021 13:52:01 +0000 (15:52 +0200)]
bnx2: Replace open-coded byte swapping with swab32s()

Read NVRAM directly into buffer and use swab32s() to byte swap it in-place
instead of reading it into the end of the buffer and swapping it manually
while copying it.

[bhelgaas: commit log]
Link: https://lore.kernel.org/r/e4ac6229-1df5-8760-3a87-1ad0ace87137@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agobnx2: Search VPD with pci_vpd_find_ro_info_keyword()
Heiner Kallweit [Sun, 22 Aug 2021 13:50:50 +0000 (15:50 +0200)]
bnx2: Search VPD with pci_vpd_find_ro_info_keyword()

Use pci_vpd_find_ro_info_keyword() to search for keywords in VPD to
simplify the code.

Link: https://lore.kernel.org/r/7ca2b8b5-4c94-f644-1d80-b2ffb8df2d05@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agosfc: falcon: Search VPD with pci_vpd_find_ro_info_keyword()
Heiner Kallweit [Sun, 22 Aug 2021 13:49:36 +0000 (15:49 +0200)]
sfc: falcon: Search VPD with pci_vpd_find_ro_info_keyword()

Use pci_vpd_find_ro_info_keyword() to search for keywords in VPD to
simplify the code.

Replace netif_err() with pci_err() because the netdevice isn't registered
yet, which results in very ugly messages.

Use kmemdup_nul() instead of open-coding it.

This is the same as 37838aa437c7 ("sfc: Search VPD with
pci_vpd_find_ro_info_keyword()"), just for the falcon chip version.

Link: https://lore.kernel.org/r/898282a1-13bd-17bc-2e9a-d3dcd336b46c@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agosfc: falcon: Read VPD with pci_vpd_alloc()
Heiner Kallweit [Sun, 22 Aug 2021 13:48:33 +0000 (15:48 +0200)]
sfc: falcon: Read VPD with pci_vpd_alloc()

Use pci_vpd_alloc() to dynamically allocate a properly sized buffer and
read the full VPD data into it.

This avoids having to allocate a buffer on the stack, and we don't have to
make any assumptions on VPD size and location of information in VPD.

This is the same as 5119e20facfa ("sfc: Read VPD with pci_vpd_alloc()"),
just for the falcon chip version.

Link: https://lore.kernel.org/r/2a8d069e-9516-50d8-6520-2614222c8f5f@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI: Sync __pci_register_driver() stub for CONFIG_PCI=n
Andy Shevchenko [Fri, 13 Aug 2021 15:36:19 +0000 (18:36 +0300)]
PCI: Sync __pci_register_driver() stub for CONFIG_PCI=n

The CONFIG_PCI=y case got a new parameter long time ago.  Sync the stub as
well.

[bhelgaas: add parameter names]
Fixes: 725522b5453d ("PCI: add the sysfs driver name to all modules")
Link: https://lore.kernel.org/r/20210813153619.89574-1-andriy.shevchenko@linux.intel.com
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI: Optimize pci_resource_len() to reduce kernel size
Zhen Lei [Tue, 13 Jul 2021 07:22:36 +0000 (15:22 +0800)]
PCI: Optimize pci_resource_len() to reduce kernel size

pci_resource_end() can be 0 only when pci_resource_start() is 0.
Otherwise, it is definitely an error. In this case, pci_resource_len()
should be regarded as 0. Therefore, determining whether
pci_resource_start() and pci_resource_end() are both 0 can be reduced to
determining only whether pci_resource_end() is 0.

Although only one condition judgment is reduced, the macro function
pci_resource_len() is widely referenced in the kernel. I used defconfig to
compile the latest kernel on X86, and its binary code size was reduced by
about 3KB.

  Before:
   [ 2] .rela.text        RELA             0000000000000000  093bfcb0
        0000000001a67168  0000000000000018   I      68     1     8

  After:
   [ 2] .rela.text        RELA             0000000000000000  093bfcb0
        0000000001a66598  0000000000000018   I      68     1     8

Link: https://lore.kernel.org/r/20210713072236.3043-1-thunder.leizhen@huawei.com
Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI: Make saved capability state private to core
Bjorn Helgaas [Mon, 2 Aug 2021 22:17:28 +0000 (17:17 -0500)]
PCI: Make saved capability state private to core

Interfaces and structs for saving and restoring PCI Capability state were
declared in include/linux/pci.h, but aren't needed outside drivers/pci/.

Move these to drivers/pci/pci.h:

  struct pci_cap_saved_data
  struct pci_cap_saved_state
  void pci_allocate_cap_save_buffers()
  void pci_free_cap_save_buffers()
  int pci_add_cap_save_buffer()
  int pci_add_ext_cap_save_buffer()
  struct pci_cap_saved_state *pci_find_saved_cap()
  struct pci_cap_saved_state *pci_find_saved_ext_cap()

Link: https://lore.kernel.org/r/20210802221728.1469304-1-helgaas@kernel.org
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
3 years agoPCI: Add schedule point in proc_bus_pci_read()
Krzysztof Wilczyński [Sun, 15 Aug 2021 15:08:24 +0000 (15:08 +0000)]
PCI: Add schedule point in proc_bus_pci_read()

PCI configuration space reads from /proc/bus/pci can often take several
milliseconds to complete.

Add a schedule point in proc_bus_pci_read() to reduce the maximum latency.

A similar change was made for sysfs by 2ce02a864ac1 ("PCI: Add schedule
point in pci_read_config()").

Link: https://lore.kernel.org/r/20200824052025.48362-1-benbjiang@tencent.com
Link: https://lore.kernel.org/r/20210815150824.96773-1-kw@linux.com
Signed-off-by: Krzysztof Wilczyński <kw@linux.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI: Correct the pci_iomap.h header guard #endif comment
Jonathan Cameron [Tue, 3 Aug 2021 12:30:14 +0000 (20:30 +0800)]
PCI: Correct the pci_iomap.h header guard #endif comment

Update the include/asm-generic/pci_iomap.h header guard #endif comment to
match the corresponding #ifndef.

Link: https://lore.kernel.org/r/20210803123014.2963814-1-Jonathan.Cameron@huawei.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/ACS: Enforce pci=noats with Transaction Blocking
Alex Williamson [Fri, 18 Jun 2021 20:55:14 +0000 (14:55 -0600)]
PCI/ACS: Enforce pci=noats with Transaction Blocking

PCIe Address Translation Services (ATS) provides a mechanism for a device
to provide an on-device caching translation agent (device IOTLB).  We
already have a means to disable support for this feature via the pci=noats
option.  For untrusted and externally facing devices, we not only disable
ATS support for the device, but we use Access Control Services (ACS)
Transaction Blocking to actively prevent devices from sending TLPs with
non-default AT field values.

Extend pci=noats to also make use of PCI_ACS_TB so that not only is ATS
disabled at the device, but blocked at the downstream ports.  This provides
a means to further lock-down ATS for cases such as device assignment, where
it may not be the hardware configuration of the device that makes it
untrusted, but the driver running on the device.

Link: https://lore.kernel.org/r/162404966325.2362347.12176138291577486015.stgit@omen
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Cc: Rajat Jain <rajatja@google.com>
3 years agoPCI: Add ACS quirks for Cavium multi-function devices
George Cherian [Tue, 10 Aug 2021 12:24:25 +0000 (17:54 +0530)]
PCI: Add ACS quirks for Cavium multi-function devices

Some Cavium endpoints are implemented as multi-function devices without ACS
capability, but they actually don't support peer-to-peer transactions.

Add ACS quirks to declare DMA isolation for the following devices:

  - BGX device found on Octeon-TX (8xxx)
  - CGX device found on Octeon-TX2 (9xxx)
  - RPM device found on Octeon-TX3 (10xxx)

Link: https://lore.kernel.org/r/20210810122425.1115156-1-george.cherian@marvell.com
Signed-off-by: George Cherian <george.cherian@marvell.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/PTM: Remove error message at boot
Jakub Kicinski [Wed, 11 Aug 2021 18:59:55 +0000 (11:59 -0700)]
PCI/PTM: Remove error message at boot

Since 39850ed51062 ("PCI/PTM: Save/restore Precision Time Measurement
Capability for suspend/resume"), devices that have PTM capability but
don't enable it see this message on calls to pci_save_state():

  no suspend buffer for PTM

Drop the message, it's perfectly fine not to use a capability.

Fixes: 39850ed51062 ("PCI/PTM: Save/restore Precision Time Measurement Capability for suspend/resume")
Link: https://lore.kernel.org/r/20210811185955.3112534-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: David E. Box <david.e.box@linux.intel.com>
3 years agotg3: Search VPD with pci_vpd_find_ro_info_keyword()
Heiner Kallweit [Wed, 18 Aug 2021 19:06:40 +0000 (21:06 +0200)]
tg3: Search VPD with pci_vpd_find_ro_info_keyword()

Use pci_vpd_find_ro_info_keyword() to search for keywords in VPD to
simplify the code.

Link: https://lore.kernel.org/r/0ae9d4c0-590d-682a-a0af-2272e5f71630@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agotg3: Validate VPD checksum with pci_vpd_check_csum()
Heiner Kallweit [Wed, 18 Aug 2021 19:05:26 +0000 (21:05 +0200)]
tg3: Validate VPD checksum with pci_vpd_check_csum()

Validate the VPD checksum with pci_vpd_check_csum() to simplify the code.

Link: https://lore.kernel.org/r/7297fce9-47db-3b86-366e-10b9ef43beaf@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agotg3: Read VPD with pci_vpd_alloc()
Heiner Kallweit [Wed, 18 Aug 2021 19:04:37 +0000 (21:04 +0200)]
tg3: Read VPD with pci_vpd_alloc()

Use pci_vpd_alloc() to dynamically allocate a properly sized buffer and
read the full VPD data into it.

This simplifies the code, and we no longer have to make assumptions about
VPD size.

Link: https://lore.kernel.org/r/bd3cd19c-b74f-9704-5786-476bf35ab5de@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agosfc: Search VPD with pci_vpd_find_ro_info_keyword()
Heiner Kallweit [Wed, 18 Aug 2021 19:03:44 +0000 (21:03 +0200)]
sfc: Search VPD with pci_vpd_find_ro_info_keyword()

Use pci_vpd_find_ro_info_keyword() to search for keywords in VPD to
simplify the code.

Replace netif_err() with pci_err() because the netdevice isn't registered
yet, which results in very ugly messages.

Use kmemdup_nul() instead of open-coding it.

Link: https://lore.kernel.org/r/bf5d4ba9-61a9-2bfe-19ec-75472732d74d@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agosfc: Read VPD with pci_vpd_alloc()
Heiner Kallweit [Wed, 18 Aug 2021 19:02:33 +0000 (21:02 +0200)]
sfc: Read VPD with pci_vpd_alloc()

Use pci_vpd_alloc() to dynamically allocate a properly sized buffer and
read the full VPD data into it.

This avoids having to allocate a buffer on the stack, and we don't have to
make any assumptions on VPD size and location of information in VPD.

Link: https://lore.kernel.org/r/e58f1e40-c043-0266-9a0f-e5a7f3f6883c@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/VPD: Add pci_vpd_check_csum()
Heiner Kallweit [Fri, 20 Aug 2021 20:32:42 +0000 (15:32 -0500)]
PCI/VPD: Add pci_vpd_check_csum()

VPD checksum information and checksum calculation are specified by PCIe
r5.0, sec 6.28.2.2.  Therefore checksum handling can and should be moved
into the PCI VPD core.

Add pci_vpd_check_csum() to validate the VPD checksum.

[bhelgaas: split to separate patch]
Link: https://lore.kernel.org/r/1643bd7a-088e-1028-c9b0-9d112cf48d63@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/VPD: Add pci_vpd_find_ro_info_keyword()
Heiner Kallweit [Wed, 18 Aug 2021 19:00:57 +0000 (21:00 +0200)]
PCI/VPD: Add pci_vpd_find_ro_info_keyword()

All users of pci_vpd_find_info_keyword() are interested in the VPD RO
section only. In addition all calls are followed by the same activities to
calculate start of tag data area and size of the data area.

Add pci_vpd_find_ro_info_keyword() that combines these functionalities.

pci_vpd_find_info_keyword() can be phased out once all users are converted.

[bhelgaas: split pci_vpd_check_csum() to separate patch]
Link: https://lore.kernel.org/r/1643bd7a-088e-1028-c9b0-9d112cf48d63@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/VPD: Add pci_vpd_alloc()
Heiner Kallweit [Wed, 18 Aug 2021 18:59:31 +0000 (20:59 +0200)]
PCI/VPD: Add pci_vpd_alloc()

Several users of the VPD API use a fixed-size buffer and read the VPD into
it for further usage. This requires special handling for the case that the
buffer isn't big enough to hold the full VPD data.  Also the buffer is
often allocated on the stack, which isn't too nice.

Add pci_vpd_alloc() to dynamically allocate buffer of the correct size and
read VPD into it.

Link: https://lore.kernel.org/r/955ff598-0021-8446-f856-0c2c077635d7@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI: Change the type of probe argument in reset functions
Amey Narkhede [Tue, 17 Aug 2021 18:05:00 +0000 (23:35 +0530)]
PCI: Change the type of probe argument in reset functions

Change the type of probe argument in functions which implement reset
methods from int to bool to make the context and intent clear.

Suggested-by: Alex Williamson <alex.williamson@redhat.com>
Link: https://lore.kernel.org/r/20210817180500.1253-10-ameynarkhede03@gmail.com
Signed-off-by: Amey Narkhede <ameynarkhede03@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI: Add support for ACPI _RST reset method
Shanker Donthineni [Tue, 17 Aug 2021 18:04:59 +0000 (23:34 +0530)]
PCI: Add support for ACPI _RST reset method

_RST is a standard ACPI method that performs a function level reset of a
device (ACPI v6.3, sec 7.3.25).

Add pci_dev_acpi_reset() to probe for _RST method and execute if present.
The default priority of this reset is set to below device-specific and
above hardware resets.

Suggested-by: Alex Williamson <alex.williamson@redhat.com>
Link: https://lore.kernel.org/r/20210817180500.1253-9-ameynarkhede03@gmail.com
Signed-off-by: Shanker Donthineni <sdonthineni@nvidia.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Sinan Kaya <okaya@kernel.org>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
3 years agoPCI: Setup ACPI fwnode early and at the same time with OF
Shanker Donthineni [Tue, 17 Aug 2021 18:04:58 +0000 (23:34 +0530)]
PCI: Setup ACPI fwnode early and at the same time with OF

Previously, the ACPI_COMPANION() of a pci_dev was usually set by
acpi_bind_one() in this path:

  pci_device_add
    pci_configure_device
    pci_init_capabilities
    device_add
      device_platform_notify
acpi_platform_notify
  acpi_device_notify  # KOBJ_ADD
    acpi_bind_one
      ACPI_COMPANION_SET

However, things like pci_configure_device() and pci_init_capabilities()
that run before device_add() need the ACPI_COMPANION, e.g.,
acpi_pci_bridge_d3() uses a _DSD method to learn about D3 support.  These
places had special-case code to manually look up the ACPI_COMPANION.

Set the ACPI_COMPANION earlier, in pci_setup_device(), so it will be
available while configuring the device.  This covers both paths to creating
pci_dev objects:

  pci_scan_single_device           # for normal non-SR-IOV devices
    pci_scan_device
      pci_setup_device
pci_set_acpi_fwnode
    pci_device_add

  pci_iov_add_virtfn               # for SR-IOV virtual functions
    pci_setup_device
      pci_set_acpi_fwnode

Also move the OF fwnode setup to the same spot.

[bhelgaas: commit log]
Link: https://lore.kernel.org/r/20210817180500.1253-8-ameynarkhede03@gmail.com
Signed-off-by: Shanker Donthineni <sdonthineni@nvidia.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
3 years agoPCI: Use acpi_pci_power_manageable()
Shanker Donthineni [Tue, 17 Aug 2021 21:09:47 +0000 (16:09 -0500)]
PCI: Use acpi_pci_power_manageable()

Use acpi_pci_power_manageable() instead of duplicating the logic in
acpi_pci_bridge_d3().  No functional change intended.

[bhelgaas: split out from
https://lore.kernel.org/r/20210817180500.1253-8-ameynarkhede03@gmail.com]
Signed-off-by: Shanker Donthineni <sdonthineni@nvidia.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI: Add pci_set_acpi_fwnode() to set ACPI_COMPANION
Shanker Donthineni [Tue, 17 Aug 2021 18:04:57 +0000 (23:34 +0530)]
PCI: Add pci_set_acpi_fwnode() to set ACPI_COMPANION

Move the existing logic from acpi_pci_bridge_d3() to a separate function
pci_set_acpi_fwnode() to set the ACPI fwnode.  No functional change
intended.

Link: https://lore.kernel.org/r/20210817180500.1253-7-ameynarkhede03@gmail.com
Signed-off-by: Shanker Donthineni <sdonthineni@nvidia.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
3 years agoPCI: Allow userspace to query and set device reset mechanism
Amey Narkhede [Tue, 17 Aug 2021 18:04:56 +0000 (23:34 +0530)]
PCI: Allow userspace to query and set device reset mechanism

Add "reset_method" sysfs attribute to enable user to query and set
preferred device reset methods and their ordering.

[bhelgaas: on invalid sysfs input, return error and preserve previous
config, as in earlier patch versions]
Co-developed-by: Alex Williamson <alex.williamson@redhat.com>
Link: https://lore.kernel.org/r/20210817180500.1253-6-ameynarkhede03@gmail.com
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Amey Narkhede <ameynarkhede03@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Raphael Norwitz <raphael.norwitz@nutanix.com>
3 years agoPCI: Remove reset_fn field from pci_dev
Amey Narkhede [Tue, 17 Aug 2021 18:04:55 +0000 (23:34 +0530)]
PCI: Remove reset_fn field from pci_dev

"reset_fn" indicates whether the device supports any reset mechanism.
Remove the use of reset_fn in favor of the reset_methods array that tracks
supported reset mechanisms of a device and their ordering.

The octeon driver incorrectly used reset_fn to detect whether the device
supports FLR or not. Use pcie_reset_flr() to probe whether it supports FLR.

Co-developed-by: Alex Williamson <alex.williamson@redhat.com>
Link: https://lore.kernel.org/r/20210817180500.1253-5-ameynarkhede03@gmail.com
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Amey Narkhede <ameynarkhede03@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
Reviewed-by: Raphael Norwitz <raphael.norwitz@nutanix.com>
3 years agoPCI: Add array to track reset method ordering
Amey Narkhede [Tue, 17 Aug 2021 18:04:54 +0000 (23:34 +0530)]
PCI: Add array to track reset method ordering

Add reset_methods[] in struct pci_dev to keep track of reset mechanisms
supported by the device and their ordering.

Refactor probing and reset functions to take advantage of calling
convention of reset functions.

Co-developed-by: Alex Williamson <alex.williamson@redhat.com>
Link: https://lore.kernel.org/r/20210817180500.1253-4-ameynarkhede03@gmail.com
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Amey Narkhede <ameynarkhede03@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Raphael Norwitz <raphael.norwitz@nutanix.com>
3 years agoPCI: Add pcie_reset_flr() with 'probe' argument
Amey Narkhede [Tue, 17 Aug 2021 18:04:53 +0000 (23:34 +0530)]
PCI: Add pcie_reset_flr() with 'probe' argument

Most reset methods are of the form "pci_*_reset(dev, probe)".  pcie_flr()
was an exception because it relied on a separate pcie_has_flr() function
instead of taking a "probe" argument.

Add "pcie_reset_flr(dev, probe)" to follow the convention.  Remove
pcie_has_flr().

Some pcie_flr() callers that did not use pcie_has_flr() remain.

[bhelgaas: commit log, rework pcie_reset_flr() to use dev->devcap directly]
Link: https://lore.kernel.org/r/20210817180500.1253-3-ameynarkhede03@gmail.com
Signed-off-by: Amey Narkhede <ameynarkhede03@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Raphael Norwitz <raphael.norwitz@nutanix.com>
3 years agoPCI: Cache PCIe Device Capabilities register
Amey Narkhede [Tue, 17 Aug 2021 18:04:52 +0000 (23:34 +0530)]
PCI: Cache PCIe Device Capabilities register

Add a new member called devcap in struct pci_dev for caching the PCIe
Device Capabilities register to avoid reading PCI_EXP_DEVCAP multiple
times.

Refactor pcie_has_flr() to use cached device capabilities.

Link: https://lore.kernel.org/r/20210817180500.1253-2-ameynarkhede03@gmail.com
Signed-off-by: Amey Narkhede <ameynarkhede03@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Raphael Norwitz <raphael.norwitz@nutanix.com>
3 years agoPCI: Fix pci_dev_str_match_path() alloc while atomic bug
Dan Carpenter [Thu, 12 Aug 2021 07:00:04 +0000 (10:00 +0300)]
PCI: Fix pci_dev_str_match_path() alloc while atomic bug

pci_dev_str_match_path() is often called with a spinlock held so the
allocation has to be atomic.  The call tree is:

  pci_specified_resource_alignment() <-- takes spin_lock();
    pci_dev_str_match()
      pci_dev_str_match_path()

Fixes: 45db33709ccc ("PCI: Allow specifying devices using a base bus and path of devfns")
Link: https://lore.kernel.org/r/20210812070004.GC31863@kili
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Logan Gunthorpe <logang@deltatee.com>
3 years agox86/PCI: Add pci_numachip_init() declaration
Krzysztof Wilczyński [Thu, 12 Aug 2021 17:17:17 +0000 (17:17 +0000)]
x86/PCI: Add pci_numachip_init() declaration

numachip.c defines pci_numachip_init(), but neglected to include its
declaration, causing the following sparse and compile time warnings:

  arch/x86/pci/numachip.c:108:12: warning: no previous prototype for function 'pci_numachip_init' [-Wmissing-prototypes]
  arch/x86/pci/numachip.c:108:12: warning: symbol 'pci_numachip_init' was not declared. Should it be static?

Include asm/numachip/numachip.h, which includes the missing declaration.

Link: https://lore.kernel.org/r/20210812171717.1471243-1-kw@linux.com
Signed-off-by: Krzysztof Wilczyński <kw@linux.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/VPD: Treat invalid VPD like missing VPD capability
Heiner Kallweit [Sun, 8 Aug 2021 17:23:57 +0000 (19:23 +0200)]
PCI/VPD: Treat invalid VPD like missing VPD capability

Exporting sysfs files that can't be accessed doesn't make much sense.
Therefore, if either a quirk or the dynamic size calculation result in VPD
being marked as invalid, treat this as though the device has no VPD
capability.  One consequence is that the "vpd" sysfs file is not visible.

Link: https://lore.kernel.org/r/6a02b204-4ed2-4553-c3b2-eacf9554fa8d@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/VPD: Determine VPD size in pci_vpd_init()
Heiner Kallweit [Sun, 8 Aug 2021 17:22:52 +0000 (19:22 +0200)]
PCI/VPD: Determine VPD size in pci_vpd_init()

Determine VPD size in pci_vpd_init().

Quirks set dev->vpd.len to a non-zero value, so they cause us to skip the
dynamic size calculation.  Prerequisite is that we move the quirks from
FINAL to HEADER so they are run before pci_vpd_init().

Link: https://lore.kernel.org/r/cc4a6538-557a-294d-4f94-e6d1d3c91589@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/VPD: Embed struct pci_vpd in struct pci_dev
Heiner Kallweit [Sun, 8 Aug 2021 17:21:56 +0000 (19:21 +0200)]
PCI/VPD: Embed struct pci_vpd in struct pci_dev

Now that struct pci_vpd is really small, simplify the code by embedding
struct pci_vpd directly in struct pci_dev instead of dynamically allocating
it.

Link: https://lore.kernel.org/r/d898489e-22ba-71f1-2f31-f1a78dc15849@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/VPD: Remove struct pci_vpd.valid member
Heiner Kallweit [Sun, 8 Aug 2021 17:21:02 +0000 (19:21 +0200)]
PCI/VPD: Remove struct pci_vpd.valid member

Instead of having a separate flag, use vp->len != 0 as indicator that VPD
validity has been checked.  Now vpd->len == PCI_VPD_SZ_INVALID indicates
that VPD is invalid.

Link: https://lore.kernel.org/r/9f777bc7-5316-e1b8-e5d4-f9f609bdb5dd@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/VPD: Remove struct pci_vpd_ops
Heiner Kallweit [Sun, 8 Aug 2021 17:20:05 +0000 (19:20 +0200)]
PCI/VPD: Remove struct pci_vpd_ops

Some multi-function devices share VPD hardware across functions and don't
work correctly for concurrent VPD accesses to different functions.

Struct pci_vpd_ops was added by 932c435caba8 ("PCI: Add dev_flags bit to
access VPD through function 0") so that on these devices, VPD accesses to
any function would always go to function 0.

It's easier to just check for the PCI_DEV_FLAGS_VPD_REF_F0 quirk bit in the
two places we need it than to deal with the struct pci_vpd_ops.

Simplify the code by removing struct pci_vpd_ops and removing the indirect
calls.

[bhelgaas: check for !func0_dev earlier, commit log]
Link: https://lore.kernel.org/r/b2532a41-df8b-860f-461f-d5c066c819d0@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/VPD: Reorder pci_read_vpd(), pci_write_vpd()
Heiner Kallweit [Sun, 8 Aug 2021 17:19:10 +0000 (19:19 +0200)]
PCI/VPD: Reorder pci_read_vpd(), pci_write_vpd()

Reorder pci_read_vpd() and pci_write_vpd() to prepare for future patches.
No functional change intended.

Link: https://lore.kernel.org/r/89f0f5a2-293b-a017-fc67-a36473a792bf@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/VPD: Remove struct pci_vpd.flag
Heiner Kallweit [Thu, 13 May 2021 21:02:01 +0000 (23:02 +0200)]
PCI/VPD: Remove struct pci_vpd.flag

The struct pci_vpd.flag member was used only to communicate between
pci_vpd_wait() and its callers.  Remove the flag member and pass the value
directly to pci_vpd_wait() to simplify the code.

[bhelgaas: commit log]
Link: https://lore.kernel.org/r/e4ef6845-6b23-1646-28a0-d5c5a28347b6@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/VPD: Make pci_vpd_wait() uninterruptible
Heiner Kallweit [Thu, 13 May 2021 20:56:41 +0000 (22:56 +0200)]
PCI/VPD: Make pci_vpd_wait() uninterruptible

Reading/writing 4 bytes should be fast enough even on a slow bus, therefore
pci_vpd_wait() doesn't have to be interruptible.  Making it uninterruptible
allows to simplify the code.

In addition make VPD writes uninterruptible in general.  It's about vital
data, and allowing writes to be interruptible may leave the VPD in an
inconsistent state.

Link: https://lore.kernel.org/r/258bf994-bc2a-2907-9181-2c7a562986d5@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/VPD: Remove pci_vpd_size() old_size argument
Heiner Kallweit [Thu, 13 May 2021 20:56:09 +0000 (22:56 +0200)]
PCI/VPD: Remove pci_vpd_size() old_size argument

vpd->len is initialized to PCI_VPD_MAX_SIZE, and if a quirk is used to set
a specific VPD size, then pci_vpd_set_size() sets vpd->valid, resulting in
pci_vpd_size() not being called. Therefore we can remove the old_size
argument. Note that we don't have to check off < PCI_VPD_MAX_SIZE because
that's implicitly done by pci_read_vpd().

Link: https://lore.kernel.org/r/ede36c16-5335-6867-43a1-293641348430@gmail.com
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI/VPD: Allow access to valid parts of VPD if some is invalid
Bjorn Helgaas [Thu, 15 Jul 2021 21:59:59 +0000 (16:59 -0500)]
PCI/VPD: Allow access to valid parts of VPD if some is invalid

Previously, if we found any error in the VPD, we returned size 0, which
prevents access to all of VPD.  But there may be valid resources in VPD
before the error, and there's no reason to prevent access to those.

"off" covers only VPD resources known to have valid header tags.  In case
of error, return "off" (which may be zero if we haven't found any valid
header tags at all).

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
3 years agoPCI/VPD: Don't check Large Resource Item Names for validity
Bjorn Helgaas [Thu, 15 Jul 2021 21:59:58 +0000 (16:59 -0500)]
PCI/VPD: Don't check Large Resource Item Names for validity

VPD consists of a series of Small and Large Resources.  Computing the size
of VPD requires only the length of each, which is specified in the generic
tag of each resource.  We only expect to see ID_STRING, RO_DATA, and
RW_DATA in VPD, but it's not a problem if it contains other resource types
because all we care about is the size.

Drop the validity checking of Large Resource items.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
3 years agoPCI/VPD: Reject resource tags with invalid size
Bjorn Helgaas [Thu, 15 Jul 2021 21:59:57 +0000 (16:59 -0500)]
PCI/VPD: Reject resource tags with invalid size

VPD is limited in size by the 15-bit VPD Address field in the VPD
Capability.  Each resource tag includes a length that determines the
overall size of the resource.  Reject any resources that would extend past
the maximum VPD size.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
3 years agoPCI: Add ACS quirks for NXP LX2xx0 and LX2xx2 platforms
Wasim Khan [Thu, 29 Jul 2021 12:17:47 +0000 (14:17 +0200)]
PCI: Add ACS quirks for NXP LX2xx0 and LX2xx2 platforms

Root Ports in NXP LX2xx0 and LX2xx2, where each Root Port is a Root Complex
with unique segment numbers, do provide isolation features to disable peer
transactions and validate bus numbers in requests, but do not provide an
actual PCIe ACS capability.

Add ACS quirks for NXP LX2xx0 A/C/E/N and LX2xx2 A/C/E/N platforms.

  LX2xx0A : without security features + CAN-FD
    LX2160A (0x8d81) - 16 cores
    LX2120A (0x8da1) - 12 cores
    LX2080A (0x8d83) -  8 cores

  LX2xx0C : security features + CAN-FD
    LX2160C (0x8d80) - 16 cores
    LX2120C (0x8da0) - 12 cores
    LX2080C (0x8d82) -  8 cores

  LX2xx0E : security features + CAN
    LX2160E (0x8d90) - 16 cores
    LX2120E (0x8db0) - 12 cores
    LX2080E (0x8d92) -  8 cores

  LX2xx0N : without security features + CAN
    LX2160N (0x8d91) - 16 cores
    LX2120N (0x8db1) - 12 cores
    LX2080N (0x8d93) -  8 cores

  LX2xx2A : without security features + CAN-FD
    LX2162A (0x8d89) - 16 cores
    LX2122A (0x8da9) - 12 cores
    LX2082A (0x8d8b) -  8 cores

  LX2xx2C : security features + CAN-FD
    LX2162C (0x8d88) - 16 cores
    LX2122C (0x8da8) - 12 cores
    LX2082C (0x8d8a) -  8 cores

  LX2xx2E : security features + CAN
    LX2162E (0x8d98) - 16 cores
    LX2122E (0x8db8) - 12 cores
    LX2082E (0x8d9a) -  8 cores

  LX2xx2N : without security features + CAN
    LX2162N (0x8d99) - 16 cores
    LX2122N (0x8db9) - 12 cores
    LX2082N (0x8d9b) -  8 cores

[bhelgaas: put PCI_VENDOR_ID_NXP definition next to PCI_VENDOR_ID_FREESCALE
as a clue that they share the same Device ID namespace]
Link: https://lore.kernel.org/r/20210729121747.1823086-1-wasim.khan@oss.nxp.com
Link: https://lore.kernel.org/r/20210803180021.3252886-1-wasim.khan@oss.nxp.com
Signed-off-by: Wasim Khan <wasim.khan@nxp.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI: Return int from pciconfig_read() syscall
Krzysztof Wilczyński [Thu, 29 Jul 2021 23:37:55 +0000 (23:37 +0000)]
PCI: Return int from pciconfig_read() syscall

Change pciconfig_read() syscall "err" return value from long to int.  This
makes it consistent with pciconfig_write().

[bhelgaas: commit log]
Link: https://lore.kernel.org/r/20210729233755.1509616-2-kw@linux.com
Signed-off-by: Krzysztof Wilczyński <kw@linux.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoPCI: Return ~0 data on pciconfig_read() CAP_SYS_ADMIN failure
Krzysztof Wilczyński [Thu, 29 Jul 2021 23:37:54 +0000 (23:37 +0000)]
PCI: Return ~0 data on pciconfig_read() CAP_SYS_ADMIN failure

The pciconfig_read() syscall reads PCI configuration space using
hardware-dependent config accessors.

If the read fails on PCI, most accessors don't return an error; they
pretend the read was successful and got ~0 data from the device, so the
syscall returns success with ~0 data in the buffer.

When the accessor does return an error, pciconfig_read() normally fills the
user's buffer with ~0 and returns an error in errno.  But after
e4585da22ad0 ("pci syscall.c: Switch to refcounting API"), we don't fill
the buffer with ~0 for the EPERM "user lacks CAP_SYS_ADMIN" error.

Userspace may rely on the ~0 data to detect errors, but after e4585da22ad0,
that would not detect CAP_SYS_ADMIN errors.

Restore the original behaviour of filling the buffer with ~0 when the
CAP_SYS_ADMIN check fails.

[bhelgaas: commit log, fold in Nathan's fix
https://lore.kernel.org/r/20210803200836.500658-1-nathan@kernel.org]
Fixes: e4585da22ad0 ("pci syscall.c: Switch to refcounting API")
Link: https://lore.kernel.org/r/20210729233755.1509616-1-kw@linux.com
Signed-off-by: Krzysztof Wilczyński <kw@linux.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Cc: stable@vger.kernel.org
3 years agoPCI: Bulk conversion to generic_handle_domain_irq()
Marc Zyngier [Mon, 2 Aug 2021 16:26:19 +0000 (17:26 +0100)]
PCI: Bulk conversion to generic_handle_domain_irq()

Wherever possible, replace constructs that match either
generic_handle_irq(irq_find_mapping()) or
generic_handle_irq(irq_linear_revmap()) to a single call to
generic_handle_domain_irq().

Link: https://lore.kernel.org/r/20210802162630.2219813-4-maz@kernel.org
Signed-off-by: Marc Zyngier <maz@kernel.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
3 years agoPCI/VPD: Treat initial 0xff as missing EEPROM
Heiner Kallweit [Thu, 29 Jul 2021 17:22:25 +0000 (12:22 -0500)]
PCI/VPD: Treat initial 0xff as missing EEPROM

Previously we assumed that the first tag being 0x00 meant an EEPROM was
missing.  The first tag being 0xff means the same thing; check for that
also.

[bhelgaas: rework error mesage]
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
3 years agoPCI/VPD: Check Resource Item Names against those valid for type
Bjorn Helgaas [Thu, 15 Jul 2021 21:59:56 +0000 (16:59 -0500)]
PCI/VPD: Check Resource Item Names against those valid for type

Previously, we checked for PCI_VPD_STIN_END, PCI_VPD_LTIN_ID_STRING, etc.,
outside the Large and Small Resource cases, so we checked Large Resource
Item Names against a Small Resource name and vice versa.

Move these tests into the Large and Small Resource cases, so we only check
PCI_VPD_STIN_END for Small Resources and PCI_VPD_LTIN_* for Large
Resources.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
3 years agoPCI/VPD: Correct diagnostic for VPD read failure
Bjorn Helgaas [Thu, 15 Jul 2021 21:59:55 +0000 (16:59 -0500)]
PCI/VPD: Correct diagnostic for VPD read failure

Previously, when a VPD read failed, we warned about an "invalid large
VPD tag".  Warn about the VPD read failure instead.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
3 years agoPCI: Restrict ASMedia ASM1062 SATA Max Payload Size Supported
Marek Behún [Thu, 24 Jun 2021 17:14:18 +0000 (19:14 +0200)]
PCI: Restrict ASMedia ASM1062 SATA Max Payload Size Supported

The ASMedia ASM1062 SATA controller advertises Max_Payload_Size_Supported
of 512, but in fact it cannot handle incoming TLPs with payload size of
512.

We discovered this issue on PCIe controllers capable of MPS = 512 (Aardvark
and DesignWare), where the issue presents itself as an External Abort.
Bjorn Helgaas says:

  Probably ASM1062 reports a Malformed TLP error when it receives a data
  payload of 512 bytes, and Aardvark, DesignWare, etc convert this to an
  arm64 External Abort. [1]

To avoid this problem, limit the ASM1062 Max Payload Size Supported to 256
bytes, so we set the Max Payload Size of devices that may send TLPs to the
ASM1062 to 256 or less.

[1] https://lore.kernel.org/linux-pci/20210601170907.GA1949035@bjorn-Precision-5520/
BugLink: https://bugzilla.kernel.org/show_bug.cgi?id=212695
Link: https://lore.kernel.org/r/20210624171418.27194-2-kabel@kernel.org
Reported-by: Rötti <espressobinboardarmbiantempmailaddress@posteo.de>
Signed-off-by: Marek Behún <kabel@kernel.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Krzysztof Wilczyński <kw@linux.com>
Reviewed-by: Pali Rohár <pali@kernel.org>
Cc: stable@vger.kernel.org
3 years agoPCI: Call Max Payload Size-related fixup quirks early
Marek Behún [Thu, 24 Jun 2021 17:14:17 +0000 (19:14 +0200)]
PCI: Call Max Payload Size-related fixup quirks early

pci_device_add() calls HEADER fixups after pci_configure_device(), which
configures Max Payload Size.

Convert MPS-related fixups to EARLY fixups so pci_configure_mps() takes
them into account.

Fixes: 27d868b5e6cfa ("PCI: Set MPS to match upstream bridge")
Link: https://lore.kernel.org/r/20210624171418.27194-1-kabel@kernel.org
Signed-off-by: Marek Behún <kabel@kernel.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Cc: stable@vger.kernel.org
3 years agoPCI: Refactor pci_ioremap_bar() and pci_ioremap_wc_bar()
Krzysztof Wilczyński [Tue, 13 Jul 2021 10:24:36 +0000 (10:24 +0000)]
PCI: Refactor pci_ioremap_bar() and pci_ioremap_wc_bar()

pci_ioremap_bar() and pci_ioremap_wc_bar() shared similar implementations
but differed in unimportant ways.  Align them by adding a shared helper,
__pci_ioremap_resource().

Upgrade warning message to error level, since it indicates a driver defect.
Remove WARN_ON() from WC path in favor of the error message.

[bhelgaas: commit log, use ioremap() since pci_iomap_range() doesn't add
anything]
Link: https://lore.kernel.org/r/20210713102436.304693-1-kw@linux.com
Signed-off-by: Krzysztof Wilczyński <kw@linux.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
3 years agoLinux 5.14-rc1
Linus Torvalds [Sun, 11 Jul 2021 22:07:40 +0000 (15:07 -0700)]
Linux 5.14-rc1

3 years agomm/rmap: try_to_migrate() skip zone_device !device_private
Hugh Dickins [Wed, 7 Jul 2021 20:13:33 +0000 (13:13 -0700)]
mm/rmap: try_to_migrate() skip zone_device !device_private

I know nothing about zone_device pages and !device_private pages; but if
try_to_migrate_one() will do nothing for them, then it's better that
try_to_migrate() filter them first, than trawl through all their vmas.

Signed-off-by: Hugh Dickins <hughd@google.com>
Reviewed-by: Shakeel Butt <shakeelb@google.com>
Reviewed-by: Alistair Popple <apopple@nvidia.com>
Link: https://lore.kernel.org/lkml/1241d356-8ec9-f47b-a5ec-9b2bf66d242@google.com/
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Jason Gunthorpe <jgg@nvidia.com>
Cc: Ralph Campbell <rcampbell@nvidia.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Yang Shi <shy828301@gmail.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agomm/rmap: fix new bug: premature return from page_mlock_one()
Hugh Dickins [Wed, 7 Jul 2021 20:11:24 +0000 (13:11 -0700)]
mm/rmap: fix new bug: premature return from page_mlock_one()

In the unlikely race case that page_mlock_one() finds VM_LOCKED has been
cleared by the time it got page table lock, page_vma_mapped_walk_done()
must be called before returning, either explicitly, or by a final call
to page_vma_mapped_walk() - otherwise the page table remains locked.

Fixes: cd62734ca60d ("mm/rmap: split try_to_munlock from try_to_unmap")
Signed-off-by: Hugh Dickins <hughd@google.com>
Reviewed-by: Alistair Popple <apopple@nvidia.com>
Reviewed-by: Shakeel Butt <shakeelb@google.com>
Reported-by: kernel test robot <oliver.sang@intel.com>
Link: https://lore.kernel.org/lkml/20210711151446.GB4070@xsang-OptiPlex-9020/
Link: https://lore.kernel.org/lkml/f71f8523-cba7-3342-40a7-114abc5d1f51@google.com/
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Jason Gunthorpe <jgg@nvidia.com>
Cc: Ralph Campbell <rcampbell@nvidia.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Yang Shi <shy828301@gmail.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agomm/rmap: fix old bug: munlocking THP missed other mlocks
Hugh Dickins [Wed, 7 Jul 2021 20:08:53 +0000 (13:08 -0700)]
mm/rmap: fix old bug: munlocking THP missed other mlocks

The kernel recovers in due course from missing Mlocked pages: but there
was no point in calling page_mlock() (formerly known as
try_to_munlock()) on a THP, because nothing got done even when it was
found to be mapped in another VM_LOCKED vma.

It's true that we need to be careful: Mlocked accounting of pte-mapped
THPs is too difficult (so consistently avoided); but Mlocked accounting
of only-pmd-mapped THPs is supposed to work, even when multiple mappings
are mlocked and munlocked or munmapped.  Refine the tests.

There is already a VM_BUG_ON_PAGE(PageDoubleMap) in page_mlock(), so
page_mlock_one() does not even have to worry about that complication.

(I said the kernel recovers: but would page reclaim be likely to split
THP before rediscovering that it's VM_LOCKED? I've not followed that up)

Fixes: 9a73f61bdb8a ("thp, mlock: do not mlock PTE-mapped file huge pages")
Signed-off-by: Hugh Dickins <hughd@google.com>
Reviewed-by: Shakeel Butt <shakeelb@google.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Link: https://lore.kernel.org/lkml/cfa154c-d595-406-eb7d-eb9df730f944@google.com/
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Jason Gunthorpe <jgg@nvidia.com>
Cc: Ralph Campbell <rcampbell@nvidia.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Yang Shi <shy828301@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agomm/rmap: fix comments left over from recent changes
Hugh Dickins [Wed, 7 Jul 2021 20:06:17 +0000 (13:06 -0700)]
mm/rmap: fix comments left over from recent changes

Parallel developments in mm/rmap.c have left behind some out-of-date
comments: try_to_migrate_one() also accepts TTU_SYNC (already commented
in try_to_migrate() itself), and try_to_migrate() returns nothing at
all.

TTU_SPLIT_FREEZE has just been deleted, so reword the comment about it
in mm/huge_memory.c; and TTU_IGNORE_ACCESS was removed in 5.11, so
delete the "recently referenced" comment from try_to_unmap_one() (once
upon a time the comment was near the removed codeblock, but they drifted
apart).

Signed-off-by: Hugh Dickins <hughd@google.com>
Reviewed-by: Shakeel Butt <shakeelb@google.com>
Reviewed-by: Alistair Popple <apopple@nvidia.com>
Link: https://lore.kernel.org/lkml/563ce5b2-7a44-5b4d-1dfd-59a0e65932a9@google.com/
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Jason Gunthorpe <jgg@nvidia.com>
Cc: Ralph Campbell <rcampbell@nvidia.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Yang Shi <shy828301@gmail.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agoMerge tag 'irq-urgent-2021-07-11' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 11 Jul 2021 18:17:57 +0000 (11:17 -0700)]
Merge tag 'irq-urgent-2021-07-11' of git://git./linux/kernel/git/tip/tip

Pull irq fixes from Ingo Molnar:
 "Two fixes:

   - Fix a MIPS IRQ handling RCU bug

   - Remove a DocBook annotation for a parameter that doesn't exist
     anymore"

* tag 'irq-urgent-2021-07-11' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  irqchip/mips: Fix RCU violation when using irqdomain lookup on interrupt entry
  genirq/irqdesc: Drop excess kernel-doc entry @lookup

3 years agoMerge tag 'sched-urgent-2021-07-11' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 11 Jul 2021 18:13:57 +0000 (11:13 -0700)]
Merge tag 'sched-urgent-2021-07-11' of git://git./linux/kernel/git/tip/tip

Pull scheduler fixes from Ingo Molnar:
 "Three fixes:

   - Fix load tracking bug/inconsistency

   - Fix a sporadic CFS bandwidth constraints enforcement bug

   - Fix a uclamp utilization tracking bug for newly woken tasks"

* tag 'sched-urgent-2021-07-11' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  sched/uclamp: Ignore max aggregation if rq is idle
  sched/fair: Fix CFS bandwidth hrtimer expiry type
  sched/fair: Sync load_sum with load_avg after dequeue

3 years agoMerge tag 'perf-urgent-2021-07-11' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 11 Jul 2021 18:10:48 +0000 (11:10 -0700)]
Merge tag 'perf-urgent-2021-07-11' of git://git./linux/kernel/git/tip/tip

Pull perf fixes from Ingo Molnar:
 "A fix and a hardware-enablement addition:

   - Robustify uncore_snbep's skx_iio_set_mapping()'s error cleanup

   - Add cstate event support for Intel ICELAKE_X and ICELAKE_D"

* tag 'perf-urgent-2021-07-11' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf/x86/intel/uncore: Clean up error handling path of iio mapping
  perf/x86/cstate: Add ICELAKE_X and ICELAKE_D support

3 years agoMerge tag 'locking-urgent-2021-07-11' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 11 Jul 2021 18:06:09 +0000 (11:06 -0700)]
Merge tag 'locking-urgent-2021-07-11' of git://git./linux/kernel/git/tip/tip

Pull locking fixes from Ingo Molnar:

 - Fix a Sparc crash

 - Fix a number of objtool warnings

 - Fix /proc/lockdep output on certain configs

 - Restore a kprobes fail-safe

* tag 'locking-urgent-2021-07-11' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  locking/atomic: sparc: Fix arch_cmpxchg64_local()
  kprobe/static_call: Restore missing static_call_text_reserved()
  static_call: Fix static_call_text_reserved() vs __init
  jump_label: Fix jump_label_text_reserved() vs __init
  locking/lockdep: Fix meaningless /proc/lockdep output of lock classes on !CONFIG_PROVE_LOCKING