kernel/linux.git/drivers/pci, branch v4.19.112

PCI: Increase D3 delay for AMD Ryzen5/7 XHCI controllers

2020-02-24T07:34:41+00:00

[ Upstream commit 3030df209aa8cf831b9963829bd9f94900ee8032 ] On Asus UX434DA (AMD Ryzen7 3700U) and Asus X512DK (AMD Ryzen5 3500U), the XHCI controller fails to resume from runtime suspend or s2idle, and USB becomes unusable from that point. xhci_hcd 0000:03:00.4: Refused to change power state, currently in D3 xhci_hcd 0000:03:00.4: enabling device (0000 -> 0002) xhci_hcd 0000:03:00.4: WARN: xHC restore state timeout xhci_hcd 0000:03:00.4: PCI post-resume error -110! xhci_hcd 0000:03:00.4: HC died; cleaning up During suspend, a transition to D3cold is attempted, however the affected platforms do not seem to cut the power to the PCI device when in this state, so the device stays in D3hot. Upon resume, the D3hot-to-D0 transition is successful only if the D3 delay is increased to 20ms. The transition failure does not appear to be detectable as a CRS condition. Add a PCI quirk to increase the delay on the affected hardware. Link: https://bugzilla.kernel.org/show_bug.cgi?id=205587 Link: http://lkml.kernel.org/r/CAD8Lp47Vh69gQjROYG69=waJgL7hs1PwnLonL9+27S_TcRhixA@mail.gmail.com Link: https://lore.kernel.org/r/20191127053836.31624-2-drake@endlessm.com Signed-off-by: Daniel Drake Signed-off-by: Bjorn Helgaas Reviewed-by: Mika Westerberg Signed-off-by: Sasha Levin

PCI: Add generic quirk for increasing D3hot delay

2020-02-24T07:34:41+00:00

[ Upstream commit 62fe23df067715a21c4aef44068efe7ceaa8f627 ] Separate the D3 delay increase functionality out of quirk_radeon_pm() into its own function so that it can be shared with other quirks, including the AMD Ryzen XHCI quirk that will be introduced in a followup commit. Tweak the function name and message to indicate more clearly that the delay relates to a D3hot-to-D0 transition. Link: https://lore.kernel.org/r/20191127053836.31624-1-drake@endlessm.com Signed-off-by: Daniel Drake Signed-off-by: Bjorn Helgaas Reviewed-by: Mika Westerberg Signed-off-by: Sasha Levin

PCI: iproc: Apply quirk_paxc_bridge() for module as well as built-in

2020-02-24T07:34:41+00:00

[ Upstream commit 574f29036fce385e28617547955dd6911d375025 ] Previously quirk_paxc_bridge() was applied when the iproc driver was built-in, but not when it was compiled as a module. This happened because it was under #ifdef CONFIG_PCIE_IPROC_PLATFORM: PCIE_IPROC_PLATFORM=y causes CONFIG_PCIE_IPROC_PLATFORM to be defined, but PCIE_IPROC_PLATFORM=m causes CONFIG_PCIE_IPROC_PLATFORM_MODULE to be defined. Move quirk_paxc_bridge() to pcie-iproc.c and drop the #ifdef so the quirk is always applied, whether iproc is built-in or a module. [bhelgaas: commit log, move to pcie-iproc.c, not pcie-iproc-platform.c] Link: https://lore.kernel.org/r/20191211174511.89713-1-wei.liu@kernel.org Signed-off-by: Wei Liu Signed-off-by: Bjorn Helgaas Signed-off-by: Sasha Levin

PCI: Don't disable bridge BARs when assigning bus resources

2020-02-14T21:33:23+00:00

commit 9db8dc6d0785225c42a37be7b44d1b07b31b8957 upstream. Some PCI bridges implement BARs in addition to bridge windows. For example, here's a PLX switch: 04:00.0 PCI bridge: PLX Technology, Inc. PEX 8724 24-Lane, 6-Port PCI Express Gen 3 (8 GT/s) Switch, 19 x 19mm FCBGA (rev ca) (prog-if 00 [Normal decode]) Flags: bus master, fast devsel, latency 0, IRQ 30, NUMA node 0 Memory at 90a00000 (32-bit, non-prefetchable) [size=256K] Bus: primary=04, secondary=05, subordinate=0a, sec-latency=0 I/O behind bridge: 00002000-00003fff Memory behind bridge: 90000000-909fffff Prefetchable memory behind bridge: 0000380000800000-0000380000bfffff Previously, when the kernel assigned resource addresses (with the pci=realloc command line parameter, for example) it could clear the struct resource corresponding to the BAR. When this happened, lspci would report this BAR as "ignored": Region 0: Memory at (32-bit, non-prefetchable) [size=256K] This is because the kernel reports a zero start address and zero flags in the corresponding sysfs resource file and in /proc/bus/pci/devices. Investigation with 'lspci -x', however, shows the BIOS-assigned address will still be programmed in the device's BAR registers. It's clearly a bug that the kernel lost track of the BAR value, but in most cases, this still won't result in a visible issue because nothing uses the memory, so nothing is affected. However, when an IOMMU is in use, it will not reserve this space in the IOVA because the kernel no longer thinks the range is valid. (See dmar_init_reserved_ranges() for the Intel implementation of this.) Without the proper reserved range, a DMA mapping may allocate an IOVA that matches a bridge BAR, which results in DMA accesses going to the BAR instead of the intended RAM. The problem was in pci_assign_unassigned_root_bus_resources(). When any resource from a bridge device fails to get assigned, the code set the resource's flags to zero. This makes sense for bridge windows, as they will be re-enabled later, but for regular BARs, it makes the kernel permanently lose track of the fact that they decode address space. Change pci_assign_unassigned_root_bus_resources() and pci_assign_unassigned_bridge_resources() so they only clear "res->flags" for bridge *windows*, not bridge BARs. Fixes: da7822e5ad71 ("PCI: update bridge resources to get more big ranges when allocating space (again)") Link: https://lore.kernel.org/r/20200108213208.4612-1-logang@deltatee.com [bhelgaas: commit log, check for pci_is_bridge()] Reported-by: Kit Chow Signed-off-by: Logan Gunthorpe Signed-off-by: Bjorn Helgaas Signed-off-by: Greg Kroah-Hartman

PCI/switchtec: Fix vep_vector_number ioread width

2020-02-14T21:33:23+00:00

commit 9375646b4cf03aee81bc6c305aa18cc80b682796 upstream. vep_vector_number is actually a 16 bit register which should be read with ioread16() instead of ioread32(). Fixes: 080b47def5e5 ("MicroSemi Switchtec management interface driver") Link: https://lore.kernel.org/r/20200106190337.2428-3-logang@deltatee.com Reported-by: Doug Meyer Signed-off-by: Logan Gunthorpe Signed-off-by: Bjorn Helgaas Signed-off-by: Greg Kroah-Hartman

PCI/IOV: Fix memory leak in pci_iov_add_virtfn()

2020-02-14T21:33:23+00:00

commit 8c386cc817878588195dde38e919aa6ba9409d58 upstream. In the implementation of pci_iov_add_virtfn() the allocated virtfn is leaked if pci_setup_device() fails. The error handling is not calling pci_stop_and_remove_bus_device(). Change the goto label to failed2. Fixes: 156c55325d30 ("PCI: Check for pci_setup_device() failure in pci_iov_add_virtfn()") Link: https://lore.kernel.org/r/20191125195255.23740-1-navid.emamdoost@gmail.com Signed-off-by: Navid Emamdoost Signed-off-by: Bjorn Helgaas Signed-off-by: Greg Kroah-Hartman

PCI: keystone: Fix link training retries initiation

2020-02-11T12:34:00+00:00

[ Upstream commit 6df19872d881641e6394f93ef2938cffcbdae5bb ] ks_pcie_stop_link() function does not clear LTSSM_EN_VAL bit so link training was not triggered more than once after startup. In configurations where link can be unstable during early boot, for example, under low temperature, it will never be established. Fixes: 0c4ffcfe1fbc ("PCI: keystone: Add TI Keystone PCIe driver") Signed-off-by: Yurii Monakov Signed-off-by: Lorenzo Pieralisi Acked-by: Andrew Murray Cc: stable@vger.kernel.org Signed-off-by: Sasha Levin

PCI: tegra: Fix return value check of pm_runtime_get_sync()

2020-02-11T12:33:59+00:00

commit 885199148442f56b880995d703d2ed03b6481a3c upstream. pm_runtime_get_sync() returns the device's usage counter. This might be >0 if the device is already powered up or CONFIG_PM is disabled. Abort probe function on real error only. Fixes: da76ba50963b ("PCI: tegra: Add power management support") Link: https://lore.kernel.org/r/20191216111825.28136-1-david.engraf@sysgo.com Signed-off-by: David Engraf Signed-off-by: Bjorn Helgaas Signed-off-by: Lorenzo Pieralisi Acked-by: Andrew Murray Cc: stable@vger.kernel.org # v4.17+ Signed-off-by: Greg Kroah-Hartman

PCI: Add DMA alias quirk for Intel VCA NTB

2020-02-01T09:37:09+00:00

[ Upstream commit 56b4cd4b7da9ee95778eb5c8abea49f641ebfd91 ] Intel Visual Compute Accelerator (VCA) is a family of PCIe add-in devices exposing computational units via Non Transparent Bridges (NTB, PEX 87xx). Similarly to MIC x200, we need to add DMA aliases to allow buffer access when IOMMU is enabled. Add aliases to allow computational unit access to host memory. These aliases mark the whole VCA device as one IOMMU group. All possible slot numbers (0x20) are used, since we are unable to tell what slot is used on other side. This quirk is intended for both host and computational unit sides. The VCA devices have up to five functions: four for DMA channels and one additional. Link: https://lore.kernel.org/r/5683A335CC8BE1438C3C30C49DCC38DF637CED8E@IRSMSX102.ger.corp.intel.com Signed-off-by: Slawomir Pawlowski Signed-off-by: Przemek Kitszel Signed-off-by: Bjorn Helgaas Signed-off-by: Sasha Levin

PCI: Mark AMD Navi14 GPU rev 0xc5 ATS as broken

2020-01-29T15:43:18+00:00

commit 5e89cd303e3a4505752952259b9f1ba036632544 upstream. To account for parts of the chip that are "harvested" (disabled) due to silicon flaws, caches on some AMD GPUs must be initialized before ATS is enabled. ATS is normally enabled by the IOMMU driver before the GPU driver loads, so this cache initialization would have to be done in a quirk, but that's too complex to be practical. For Navi14 (device ID 0x7340), this initialization is done by the VBIOS, but apparently some boards went to production with an older VBIOS that doesn't do it. Disable ATS for those boards. Link: https://lore.kernel.org/r/20200114205523.1054271-3-alexander.deucher@amd.com Bug: https://gitlab.freedesktop.org/drm/amd/issues/1015 See-also: d28ca864c493 ("PCI: Mark AMD Stoney Radeon R7 GPU ATS as broken") See-also: 9b44b0b09dec ("PCI: Mark AMD Stoney GPU ATS as broken") [bhelgaas: squash into one patch, simplify slightly, commit log] Signed-off-by: Alex Deucher Signed-off-by: Bjorn Helgaas Cc: stable@vger.kernel.org Signed-off-by: Greg Kroah-Hartman