[5.9,033/152] arm64: pgtable: Fix pte_accessible()

Message ID	20201201084716.224151308@linuxfoundation.org
State	New
Headers	show Return-Path: <stable-owner@kernel.org> From: Greg Kroah-Hartman <gregkh@linuxfoundation.org> To: linux-kernel@vger.kernel.org Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>, stable@vger.kernel.org, Yu Zhao <yuzhao@google.com>, Minchan Kim <minchan@kernel.org>, Catalin Marinas <catalin.marinas@arm.com>, Will Deacon <will@kernel.org> Subject: [PATCH 5.9 033/152] arm64: pgtable: Fix pte_accessible() Date: Tue, 1 Dec 2020 09:52:28 +0100 Message-Id: <20201201084716.224151308@linuxfoundation.org> In-Reply-To: <20201201084711.707195422@linuxfoundation.org> References: <20201201084711.707195422@linuxfoundation.org> User-Agent: quilt/0.66 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Precedence: bulk
Series	None \| expand [5.9,002/152] io_uring: order refnode recycling [5.9,003/152] spi: bcm-qspi: Fix use-after-free on unbind [5.9,004/152] spi: bcm2835: Fix use-after-free on unbind [5.9,005/152] ipv4: use IS_ENABLED instead of ifdef [5.9,006/152] IB/hfi1: Ensure correct mm is used at all times [5.9,007/152] RDMA/i40iw: Address an mmap handler exploit in i40iw [5.9,008/152] btrfs: fix missing delalloc new bit for new delalloc ranges [5.9,009/152] btrfs: tree-checker: add missing return after error in root_item [5.9,010/152] btrfs: tree-checker: add missing returns after data_ref alignment checks [5.9,011/152] btrfs: dont access possibly stale fs_info data for printing duplicate device [5.9,012/152] btrfs: fix lockdep splat when reading qgroup config on mount [5.9,013/152] rtc: pcf2127: fix a bug when not specify interrupts property [5.9,014/152] s390: fix fpu restore in entry.S [5.9,015/152] mm: fix VM_BUG_ON(PageTail) and BUG_ON(PageWriteback) [5.9,016/152] smb3: Call cifs reconnect from demultiplex thread [5.9,017/152] smb3: Avoid Mid pending list corruption [5.9,018/152] smb3: Handle error case during offload read path [5.9,019/152] cifs: fix a memleak with modefromsid [5.9,020/152] powerpc/64s: Fix KVM system reset handling when CONFIG_PPC_PSERIES=y [5.9,021/152] powerpc/64s/exception: KVM Fix for host DSI being taken in HPT guest MMU context [5.9,022/152] KVM: PPC: Book3S HV: XIVE: Fix possible oops when accessing ESB page [5.9,023/152] KVM: arm64: vgic-v3: Drop the reporting of GICR_TYPER.Last for userspace [5.9,024/152] KVM: x86: handle !lapic_in_kernel case in kvm_cpu_*_extint [5.9,025/152] KVM: x86: Fix split-irqchip vs interrupt injection window request [5.9,026/152] iommu/vt-d: Dont read VCCAP register unless it exists [5.9,027/152] firmware: xilinx: Use hash-table for api feature check [5.9,028/152] drm/amdgpu: fix SI UVD firmware validate resume fail [5.9,029/152] io_uring: fix ITER_BVEC check [5.9,030/152] trace: fix potenial dangerous pointer [5.9,031/152] arm64: tegra: Correct the UART for Jetson Xavier NX [5.9,032/152] arm64: tegra: Fix USB_VBUS_EN0 regulator on Jetson TX1 [5.9,033/152] arm64: pgtable: Fix pte_accessible() [5.9,034/152] arm64: pgtable: Ensure dirty bit is preserved across pte_wrprotect() [5.9,035/152] drm/amdgpu: fix a page fault [5.9,036/152] drm/amdgpu: update golden setting for sienna_cichlid [5.9,037/152] drm/amd/amdgpu: fix null pointer in runtime pm [5.9,038/152] drm/amd/display: Avoid HDCP initialization in devices without output [5.9,039/152] HID: uclogic: Add ID for Trust Flex Design Tablet [5.9,040/152] HID: ite: Replace ABS_MISC 120/121 events with touchpad on/off keypresses [5.9,041/152] HID: cypress: Support Varmilo Keyboards media hotkeys [5.9,042/152] HID: add support for Sega Saturn [5.9,043/152] Input: i8042 - allow insmod to succeed on devices without an i8042 controller [5.9,044/152] HID: hid-sensor-hub: Fix issue with devices with no report ID [5.9,045/152] staging: ralink-gdma: fix kconfig dependency bug for DMA_RALINK [5.9,046/152] HID: add HID_QUIRK_INCREMENT_USAGE_ON_DUPLICATE for Gamevice devices [5.9,047/152] dmaengine: xilinx_dma: use readl_poll_timeout_atomic variant [5.9,048/152] x86/xen: dont unbind uninitialized lock_kicker_irq [5.9,049/152] kunit: fix display of failed expectations for strings [5.9,050/152] HID: logitech-hidpp: Add HIDPP_CONSUMER_VENDOR_KEYS quirk for the Dinovo Edge [5.9,051/152] HID: Add Logitech Dinovo Edge battery quirk [5.9,052/152] proc: dont allow async path resolution of /proc/self components [5.9,053/152] nvme: free sq/cq dbbuf pointers when dbbuf set fails [5.9,054/152] io_uring: handle -EOPNOTSUPP on path resolution [5.9,055/152] net: stmmac: dwmac_lib: enlarge dma reset timeout [5.9,056/152] vdpasim: fix "mac_pton" undefined error [5.9,057/152] vhost: add helper to check if a vq has been setup [5.9,058/152] vhost scsi: alloc cmds per vq instead of session [5.9,059/152] vhost scsi: fix cmd completion race [5.9,060/152] cpuidle: tegra: Annotate tegra_pm_set_cpu_in_lp2() with RCU_NONIDLE [5.9,061/152] dmaengine: pl330: _prep_dma_memcpy: Fix wrong burst size [5.9,062/152] scsi: libiscsi: Fix NOP race condition [5.9,063/152] scsi: target: iscsi: Fix cmd abort fabric stop race [5.9,064/152] lockdep: Put graph lock/unlock under lock_recursion protection [5.9,065/152] perf/x86: fix sysfs type mismatches [5.9,066/152] xtensa: uaccess: Add missing __user to strncpy_from_user() prototype [5.9,067/152] x86/dumpstack: Do not try to access user space code of other tasks [5.9,068/152] net: dsa: mv88e6xxx: Wait for EEPROM done after HW reset [5.9,069/152] bus: ti-sysc: Fix reset status check for modules with quirks [5.9,070/152] bus: ti-sysc: Fix bogus resetdone warning on enable for cpsw [5.9,071/152] ARM: OMAP2+: Manage MPU state properly for omap_enter_idle_coupled() [5.9,072/152] phy: tegra: xusb: Fix dangling pointer on probe failure [5.9,073/152] iwlwifi: mvm: use the HOT_SPOT_CMD to cancel an AUX ROC [5.9,074/152] iwlwifi: mvm: properly cancel a session protection for P2P [5.9,075/152] iwlwifi: mvm: write queue_sync_state only for sync [5.9,076/152] KVM: s390: pv: Mark mm as protected after the set secure parameters and improve cle... [5.9,077/152] KVM: s390: remove diag318 reset code [5.9,078/152] btrfs: qgroup: dont commit transaction when we already hold the handle [5.9,079/152] batman-adv: set .owner to THIS_MODULE [5.9,080/152] usb: cdns3: gadget: fix some endian issues [5.9,081/152] usb: cdns3: gadget: calculate TD_SIZE based on TD [5.9,082/152] phy: qualcomm: usb: Fix SuperSpeed PHY OF dependency [5.9,083/152] phy: qualcomm: Fix 28 nm Hi-Speed USB PHY OF dependency [5.9,084/152] arch: pgtable: define MAX_POSSIBLE_PHYSMEM_BITS where needed [5.9,085/152] bus: ti-sysc: suppress err msg for timers used as clockevent/source [5.9,086/152] ARM: dts: dra76x: m_can: fix order of clocks [5.9,087/152] scsi: ufs: Fix race between shutdown and runtime resume flow [5.9,088/152] bnxt_en: fix error return code in bnxt_init_one() [5.9,089/152] bnxt_en: fix error return code in bnxt_init_board() [5.9,090/152] video: hyperv_fb: Fix the cache type when mapping the VRAM [5.9,091/152] bnxt_en: Release PCI regions when DMA mask setup fails during probe. [5.9,092/152] block/keyslot-manager: prevent crash when num_slots=1 [5.9,093/152] cxgb4: fix the panic caused by non smac rewrite [5.9,094/152] dpaa2-eth: select XGMAC_MDIO for MDIO bus support [5.9,095/152] s390/qeth: make af_iucv TX notification call more robust [5.9,096/152] s390/qeth: fix af_iucv notification race [5.9,097/152] s390/qeth: fix tear down of async TX buffers [5.9,098/152] drm/mediatek: dsi: Modify horizontal front/back porch byte formula [5.9,099/152] bonding: wait for sysfs kobject destruction before freeing struct slave [5.9,100/152] ibmvnic: fix call_netdevice_notifiers in do_reset [5.9,101/152] ibmvnic: notify peers when failover and migration happen [5.9,102/152] powerpc/64s: Fix allnoconfig build since uaccess flush [5.9,103/152] iommu: Check return of __iommu_attach_device() [5.9,104/152] IB/mthca: fix return value of error branch in mthca_init_cq() [5.9,105/152] i40e: Fix removing driver while bare-metal VFs pass traffic [5.9,106/152] firmware: xilinx: Fix SD DLL node reset issue [5.9,107/152] spi: imx: fix the unbalanced spi runtime pm management [5.9,108/152] io_uring: fix shift-out-of-bounds when round up cq size [5.9,109/152] aquantia: Remove the build_skb path [5.9,110/152] nfc: s3fwrn5: use signed integer for parsing GPIO numbers [5.9,111/152] net: ena: handle bad request id in ena_netdev [5.9,112/152] net: ena: set initial DMA width to avoid intel iommu issue [5.9,113/152] net: ena: fix packets addresses for rx_offset feature [5.9,114/152] ibmvnic: fix NULL pointer dereference in reset_sub_crq_queues [5.9,115/152] ibmvnic: fix NULL pointer dereference in ibmvic_reset_crq [5.9,116/152] ibmvnic: enhance resetting status check during module exit [5.9,117/152] optee: add writeback to valid memory type [5.9,118/152] x86/tboot: Dont disable swiotlb when iommu is forced on [5.9,119/152] arm64: tegra: Wrong AON HSP reg property size [5.9,120/152] efi/efivars: Set generic ops before loading SSDT [5.9,121/152] efivarfs: revert "fix memory leak in efivarfs_create()" [5.9,122/152] efi: EFI_EARLYCON should depend on EFI [5.9,123/152] riscv: Explicitly specify the build id style in vDSO Makefile again [5.9,124/152] RISC-V: Add missing jump label initialization [5.9,125/152] RISC-V: fix barrier() use in <vdso/processor.h> [5.9,126/152] net: stmmac: fix incorrect merge of patch upstream [5.9,127/152] enetc: Let the hardware auto-advance the taprio base-time of 0 [5.9,128/152] ptp: clockmatrix: bug fix for idtcm_strverscmp [5.9,129/152] drm/nouveau: fix relocations applying logic and a double-free [5.9,130/152] can: gs_usb: fix endianess problem with candleLight firmware [5.9,131/152] platform/x86: thinkpad_acpi: Send tablet mode switch at wakeup time [5.9,132/152] platform/x86: toshiba_acpi: Fix the wrong variable assignment [5.9,133/152] RDMA/hns: Fix wrong field of SRQ number the device supports [5.9,134/152] RDMA/hns: Fix retry_cnt and rnr_cnt when querying QP [5.9,135/152] RDMA/hns: Bugfix for memory window mtpt configuration [5.9,136/152] can: m_can: m_can_open(): remove IRQF_TRIGGER_FALLING from request_threaded_irq()s ... [5.9,137/152] can: m_can: fix nominal bitiming tseg2 min for version >= 3.1 [5.9,138/152] perf record: Synthesize cgroup events only if needed [5.9,139/152] perf stat: Use proper cpu for shadow stats [5.9,140/152] perf probe: Fix to die_entrypc() returns error correctly [5.9,141/152] spi: bcm2835aux: Restore err assignment in bcm2835aux_spi_probe [5.9,142/152] USB: core: Change %pK for __user pointers to %px [5.9,143/152] usb: gadget: f_midi: Fix memleak in f_midi_alloc [5.9,144/152] USB: core: Fix regression in Hercules audio card [5.9,145/152] USB: quirks: Add USB_QUIRK_DISCONNECT_SUSPEND quirk for Lenovo A630Z TIO built-in u... [5.9,146/152] usb: gadget: Fix memleak in gadgetfs_fill_super [5.9,147/152] irqchip/exiu: Fix the index of fwspec for IRQ type [5.9,148/152] x86/mce: Do not overwrite no_way_out if mce_end() fails [5.9,149/152] x86/speculation: Fix prctl() when spectre_v2_user={seccomp,prctl},ibpb [5.9,150/152] x86/resctrl: Remove superfluous kernfs_get() calls to prevent refcount leak [5.9,151/152] x86/resctrl: Add necessary kernfs_put() calls to prevent refcount leak [5.9,152/152] drm/amdgpu: add rlc iram and dram firmware support

Message ID

20201201084716.224151308@linuxfoundation.org

State

New

Headers

From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To: linux-kernel@vger.kernel.org
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	stable@vger.kernel.org, Yu Zhao <yuzhao@google.com>,
	Minchan Kim <minchan@kernel.org>,
	Catalin Marinas <catalin.marinas@arm.com>, Will Deacon <will@kernel.org>
Subject: [PATCH 5.9 033/152] arm64: pgtable: Fix pte_accessible()
Date: Tue,  1 Dec 2020 09:52:28 +0100
Message-Id: <20201201084716.224151308@linuxfoundation.org>
In-Reply-To: <20201201084711.707195422@linuxfoundation.org>
References: <20201201084711.707195422@linuxfoundation.org>
User-Agent: quilt/0.66
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Precedence: bulk

Series

None | expand

Commit Message

Greg Kroah-Hartman Dec. 1, 2020, 8:52 a.m. UTC

From: Will Deacon <will@kernel.org>

commit 07509e10dcc77627f8b6a57381e878fe269958d3 upstream.

pte_accessible() is used by ptep_clear_flush() to figure out whether TLB
invalidation is necessary when unmapping pages for reclaim. Although our
implementation is correct according to the architecture, returning true
only for valid, young ptes in the absence of racing page-table
modifications, this is in fact flawed due to lazy invalidation of old
ptes in ptep_clear_flush_young() where we elide the expensive DSB
instruction for completing the TLB invalidation.

Rather than penalise the aging path, adjust pte_accessible() to return
true for any valid pte, even if the access flag is cleared.

Cc: <stable@vger.kernel.org>
Fixes: 76c714be0e5e ("arm64: pgtable: implement pte_accessible()")
Reported-by: Yu Zhao <yuzhao@google.com>
Acked-by: Yu Zhao <yuzhao@google.com>
Reviewed-by: Minchan Kim <minchan@kernel.org>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Link: https://lore.kernel.org/r/20201120143557.6715-2-will@kernel.org
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

---
 arch/arm64/include/asm/pgtable.h |    7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

--- a/arch/arm64/include/asm/pgtable.h
+++ b/arch/arm64/include/asm/pgtable.h
@@ -108,8 +108,6 @@  extern unsigned long empty_zero_page[PAG
 #define pte_valid(pte)		(!!(pte_val(pte) & PTE_VALID))
 #define pte_valid_not_user(pte) \
 	((pte_val(pte) & (PTE_VALID | PTE_USER)) == PTE_VALID)
-#define pte_valid_young(pte) \
-	((pte_val(pte) & (PTE_VALID | PTE_AF)) == (PTE_VALID | PTE_AF))
 #define pte_valid_user(pte) \
 	((pte_val(pte) & (PTE_VALID | PTE_USER)) == (PTE_VALID | PTE_USER))
 
@@ -117,9 +115,12 @@  extern unsigned long empty_zero_page[PAG
  * Could the pte be present in the TLB? We must check mm_tlb_flush_pending
  * so that we don't erroneously return false for pages that have been
  * remapped as PROT_NONE but are yet to be flushed from the TLB.
+ * Note that we can't make any assumptions based on the state of the access
+ * flag, since ptep_clear_flush_young() elides a DSB when invalidating the
+ * TLB.
  */
 #define pte_accessible(mm, pte)	\
-	(mm_tlb_flush_pending(mm) ? pte_present(pte) : pte_valid_young(pte))
+	(mm_tlb_flush_pending(mm) ? pte_present(pte) : pte_valid(pte))
 
 /*
  * p??_access_permitted() is true for valid user mappings (subject to the

[5.9,033/152] arm64: pgtable: Fix pte_accessible()

Commit Message

Patch