November 2022 - Linux-stable-mirror

[PATCH 1/1] arm64: dts: rockchip: rk356x: Fix PCIe register and range mappings

by Andrew Powers-Holmes

The register and range mappings for the PCIe controller in Rockchip's RK356x SoCs are incorrect. Replace them with corrected values from the vendor BSP sources, updated to match current DT schema. Cc: stable(a)vger.kernel.org Signed-off-by: Andrew Powers-Holmes <aholmes(a)omnom.net> --- arch/arm64/boot/dts/rockchip/rk3568.dtsi | 14 ++++++++------ arch/arm64/boot/dts/rockchip/rk356x.dtsi | 7 ++++--- 2 files changed, 12 insertions(+), 9 deletions(-) diff --git a/arch/arm64/boot/dts/rockchip/rk3568.dtsi b/arch/arm64/boot/dts/rockchip/rk3568.dtsi index ba67b58f05b7..c1128d0c4406 100644 --- a/arch/arm64/boot/dts/rockchip/rk3568.dtsi +++ b/arch/arm64/boot/dts/rockchip/rk3568.dtsi @@ -94,9 +94,10 @@ pcie3x1: pcie@fe270000 { power-domains = <&power RK3568_PD_PIPE>; reg = <0x3 0xc0400000 0x0 0x00400000>, <0x0 0xfe270000 0x0 0x00010000>, - <0x3 0x7f000000 0x0 0x01000000>; - ranges = <0x01000000 0x0 0x3ef00000 0x3 0x7ef00000 0x0 0x00100000>, - <0x02000000 0x0 0x00000000 0x3 0x40000000 0x0 0x3ef00000>; + <0x0 0xf2000000 0x0 0x00100000>; + ranges = <0x01000000 0x0 0xf2100000 0x0 0xf2100000 0x0 0x00100000>, + <0x02000000 0x0 0xf2200000 0x0 0xf2200000 0x0 0x01e00000>, + <0x03000000 0x0 0x40000000 0x3 0x40000000 0x0 0x40000000>; reg-names = "dbi", "apb", "config"; resets = <&cru SRST_PCIE30X1_POWERUP>; reset-names = "pipe"; @@ -146,9 +147,10 @@ pcie3x2: pcie@fe280000 { power-domains = <&power RK3568_PD_PIPE>; reg = <0x3 0xc0800000 0x0 0x00400000>, <0x0 0xfe280000 0x0 0x00010000>, - <0x3 0xbf000000 0x0 0x01000000>; - ranges = <0x01000000 0x0 0x3ef00000 0x3 0xbef00000 0x0 0x00100000>, - <0x02000000 0x0 0x00000000 0x3 0x80000000 0x0 0x3ef00000>; + <0x0 0xf2000000 0x0 0x01000000>; + ranges = <0x01000000 0x0 0xf0100000 0x0 0xf0100000 0x0 0x00100000>, + <0x02000000 0x0 0xf0200000 0x0 0xf0200000 0x0 0x01e00000>, + <0x03000000 0x0 0x40000000 0x3 0x80000000 0x0 0x40000000>; reg-names = "dbi", "apb", "config"; resets = <&cru SRST_PCIE30X2_POWERUP>; reset-names = "pipe"; diff --git a/arch/arm64/boot/dts/rockchip/rk356x.dtsi b/arch/arm64/boot/dts/rockchip/rk356x.dtsi index 164708f1eb67..eec1d496c617 100644 --- a/arch/arm64/boot/dts/rockchip/rk356x.dtsi +++ b/arch/arm64/boot/dts/rockchip/rk356x.dtsi @@ -951,7 +951,7 @@ pcie2x1: pcie@fe260000 { compatible = "rockchip,rk3568-pcie"; reg = <0x3 0xc0000000 0x0 0x00400000>, <0x0 0xfe260000 0x0 0x00010000>, - <0x3 0x3f000000 0x0 0x01000000>; + <0x0 0xf4000000 0x0 0x00100000>; reg-names = "dbi", "apb", "config"; interrupts = <GIC_SPI 75 IRQ_TYPE_LEVEL_HIGH>, <GIC_SPI 74 IRQ_TYPE_LEVEL_HIGH>, @@ -980,8 +980,9 @@ pcie2x1: pcie@fe260000 { phys = <&combphy2 PHY_TYPE_PCIE>; phy-names = "pcie-phy"; power-domains = <&power RK3568_PD_PIPE>; - ranges = <0x01000000 0x0 0x3ef00000 0x3 0x3ef00000 0x0 0x00100000 - 0x02000000 0x0 0x00000000 0x3 0x00000000 0x0 0x3ef00000>; + ranges = <0x01000000 0x0 0xf4100000 0x0 0xf4100000 0x0 0x00100000>, + <0x02000000 0x0 0xf4200000 0x0 0xf4200000 0x0 0x01e00000>, + <0x03000000 0x0 0x40000000 0x3 0x00000000 0x0 0x40000000>; resets = <&cru SRST_PCIE20_POWERUP>; reset-names = "pipe"; #address-cells = <3>; -- 2.38.0

2 years, 1 month

2
1
0 0

[PATCH 3/6] phy: qcom-qmp-combo: fix sc8180x reset

by Johan Hovold

The SC8180X has two resets but the DP configuration erroneously described only one. In case the DP part of the PHY is initialised before the USB part (e.g. depending on probe order), then only the first reset would be asserted. Fixes: 1633802cd4ac ("phy: qcom: qmp: Add SC8180x USB/DP combo") Cc: stable(a)vger.kernel.org # 5.15 Signed-off-by: Johan Hovold <johan+linaro(a)kernel.org> --- drivers/phy/qualcomm/phy-qcom-qmp-combo.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/phy/qualcomm/phy-qcom-qmp-combo.c b/drivers/phy/qualcomm/phy-qcom-qmp-combo.c index cc53e2f99121..40c25a0ead23 100644 --- a/drivers/phy/qualcomm/phy-qcom-qmp-combo.c +++ b/drivers/phy/qualcomm/phy-qcom-qmp-combo.c @@ -1177,8 +1177,8 @@ static const struct qmp_phy_cfg sc8180x_dpphy_cfg = { .clk_list = qmp_v3_phy_clk_l, .num_clks = ARRAY_SIZE(qmp_v3_phy_clk_l), - .reset_list = sc7180_usb3phy_reset_l, - .num_resets = ARRAY_SIZE(sc7180_usb3phy_reset_l), + .reset_list = msm8996_usb3phy_reset_l, + .num_resets = ARRAY_SIZE(msm8996_usb3phy_reset_l), .vreg_list = qmp_phy_vreg_l, .num_vregs = ARRAY_SIZE(qmp_phy_vreg_l), .regs = qmp_v3_usb3phy_regs_layout, -- 2.37.4

2 years, 1 month

2
1
0 0

[PATCH 2/6] phy: qcom-qmp-combo: fix sdm845 reset

by Johan Hovold

The SDM845 has two resets but the DP configuration erroneously described only one. In case the DP part of the PHY is initialised before the USB part (e.g. depending on probe order), then only the first reset would be asserted. Add a dedicated configuration for SDM845 rather than reuse the incompatible SC7180 configuration. Fixes: d88497fb6bbd ("phy: qualcomm: phy-qcom-qmp: add support for combo USB3+DP phy on SDM845") Cc: stable(a)vger.kernel.org # 6.1 Signed-off-by: Johan Hovold <johan+linaro(a)kernel.org> --- drivers/phy/qualcomm/phy-qcom-qmp-combo.c | 39 ++++++++++++++++++++++- 1 file changed, 38 insertions(+), 1 deletion(-) diff --git a/drivers/phy/qualcomm/phy-qcom-qmp-combo.c b/drivers/phy/qualcomm/phy-qcom-qmp-combo.c index bb38b18258ca..cc53e2f99121 100644 --- a/drivers/phy/qualcomm/phy-qcom-qmp-combo.c +++ b/drivers/phy/qualcomm/phy-qcom-qmp-combo.c @@ -1084,9 +1084,46 @@ static const struct qmp_phy_cfg sdm845_usb3phy_cfg = { .has_pwrdn_delay = true, }; +static const struct qmp_phy_cfg sdm845_dpphy_cfg = { + .type = PHY_TYPE_DP, + .lanes = 2, + + .serdes_tbl = qmp_v3_dp_serdes_tbl, + .serdes_tbl_num = ARRAY_SIZE(qmp_v3_dp_serdes_tbl), + .tx_tbl = qmp_v3_dp_tx_tbl, + .tx_tbl_num = ARRAY_SIZE(qmp_v3_dp_tx_tbl), + + .serdes_tbl_rbr = qmp_v3_dp_serdes_tbl_rbr, + .serdes_tbl_rbr_num = ARRAY_SIZE(qmp_v3_dp_serdes_tbl_rbr), + .serdes_tbl_hbr = qmp_v3_dp_serdes_tbl_hbr, + .serdes_tbl_hbr_num = ARRAY_SIZE(qmp_v3_dp_serdes_tbl_hbr), + .serdes_tbl_hbr2 = qmp_v3_dp_serdes_tbl_hbr2, + .serdes_tbl_hbr2_num = ARRAY_SIZE(qmp_v3_dp_serdes_tbl_hbr2), + .serdes_tbl_hbr3 = qmp_v3_dp_serdes_tbl_hbr3, + .serdes_tbl_hbr3_num = ARRAY_SIZE(qmp_v3_dp_serdes_tbl_hbr3), + + .swing_hbr_rbr = &qmp_dp_v3_voltage_swing_hbr_rbr, + .pre_emphasis_hbr_rbr = &qmp_dp_v3_pre_emphasis_hbr_rbr, + .swing_hbr3_hbr2 = &qmp_dp_v3_voltage_swing_hbr3_hbr2, + .pre_emphasis_hbr3_hbr2 = &qmp_dp_v3_pre_emphasis_hbr3_hbr2, + + .clk_list = qmp_v3_phy_clk_l, + .num_clks = ARRAY_SIZE(qmp_v3_phy_clk_l), + .reset_list = msm8996_usb3phy_reset_l, + .num_resets = ARRAY_SIZE(msm8996_usb3phy_reset_l), + .vreg_list = qmp_phy_vreg_l, + .num_vregs = ARRAY_SIZE(qmp_phy_vreg_l), + .regs = qmp_v3_usb3phy_regs_layout, + + .dp_aux_init = qcom_qmp_v3_phy_dp_aux_init, + .configure_dp_tx = qcom_qmp_v3_phy_configure_dp_tx, + .configure_dp_phy = qcom_qmp_v3_phy_configure_dp_phy, + .calibrate_dp_phy = qcom_qmp_v3_dp_phy_calibrate, +}; + static const struct qmp_phy_combo_cfg sdm845_usb3dpphy_cfg = { .usb_cfg = &sdm845_usb3phy_cfg, - .dp_cfg = &sc7180_dpphy_cfg, + .dp_cfg = &sdm845_dpphy_cfg, }; static const struct qmp_phy_cfg sm8150_usb3phy_cfg = { -- 2.37.4

2 years, 1 month

2
1
0 0

[PATCH 1/6] phy: qcom-qmp-combo: fix out-of-bounds clock access

by Johan Hovold

The SM8250 only uses three clocks but the DP configuration erroneously described four clocks. In case the DP part of the PHY is initialised before the USB part, this would lead to uninitialised memory beyond the bulk-clocks array to be treated as a clock pointer as the clocks are requested based on the USB configuration. Fixes: aff188feb5e1 ("phy: qcom-qmp: add support for sm8250-usb3-dp phy") Cc: stable(a)vger.kernel.org # 5.13 Signed-off-by: Johan Hovold <johan+linaro(a)kernel.org> --- drivers/phy/qualcomm/phy-qcom-qmp-combo.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/phy/qualcomm/phy-qcom-qmp-combo.c b/drivers/phy/qualcomm/phy-qcom-qmp-combo.c index 5e11b6a1d189..bb38b18258ca 100644 --- a/drivers/phy/qualcomm/phy-qcom-qmp-combo.c +++ b/drivers/phy/qualcomm/phy-qcom-qmp-combo.c @@ -1270,8 +1270,8 @@ static const struct qmp_phy_cfg sm8250_dpphy_cfg = { .swing_hbr3_hbr2 = &qmp_dp_v3_voltage_swing_hbr3_hbr2, .pre_emphasis_hbr3_hbr2 = &qmp_dp_v3_pre_emphasis_hbr3_hbr2, - .clk_list = qmp_v4_phy_clk_l, - .num_clks = ARRAY_SIZE(qmp_v4_phy_clk_l), + .clk_list = qmp_v4_sm8250_usbphy_clk_l, + .num_clks = ARRAY_SIZE(qmp_v4_sm8250_usbphy_clk_l), .reset_list = msm8996_usb3phy_reset_l, .num_resets = ARRAY_SIZE(msm8996_usb3phy_reset_l), .vreg_list = qmp_phy_vreg_l, -- 2.37.4

2 years, 1 month

2
1
0 0

+ hugetlb-dont-delete-vma_lock-in-hugetlb-madv_dontneed-processing.patch added to mm-hotfixes-unstable branch

by Andrew Morton

The patch titled Subject: hugetlb: don't delete vma_lock in hugetlb MADV_DONTNEED processing has been added to the -mm mm-hotfixes-unstable branch. Its filename is hugetlb-dont-delete-vma_lock-in-hugetlb-madv_dontneed-processing.patch This patch will shortly appear at https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patche… This patch will later appear in the mm-hotfixes-unstable branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm Before you just go and hit "reply", please: a) Consider who else should be cc'ed b) Prefer to cc a suitable mailing list as well c) Ideally: find the original patch on the mailing list and do a reply-to-all to that, adding suitable additional cc's *** Remember to use Documentation/process/submit-checklist.rst when testing your code *** The -mm tree is included into linux-next via the mm-everything branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm and is updated there every 2-3 working days ------------------------------------------------------ From: Mike Kravetz <mike.kravetz(a)oracle.com> Subject: hugetlb: don't delete vma_lock in hugetlb MADV_DONTNEED processing Date: Fri, 11 Nov 2022 15:26:28 -0800 madvise(MADV_DONTNEED) ends up calling zap_page_range() to clear page tables associated with the address range. For hugetlb vmas, zap_page_range will call __unmap_hugepage_range_final. However, __unmap_hugepage_range_final assumes the passed vma is about to be removed and deletes the vma_lock to prevent pmd sharing as the vma is on the way out. In the case of madvise(MADV_DONTNEED) the vma remains, but the missing vma_lock prevents pmd sharing and could potentially lead to issues with truncation/fault races. This issue was originally reported here [1] as a BUG triggered in page_try_dup_anon_rmap. Prior to the introduction of the hugetlb vma_lock, __unmap_hugepage_range_final cleared the VM_MAYSHARE flag to prevent pmd sharing. Subsequent faults on this vma were confused as VM_MAYSHARE indicates a sharable vma, but was not set so page_mapping was not set in new pages added to the page table. This resulted in pages that appeared anonymous in a VM_SHARED vma and triggered the BUG. Address issue by adding a new zap flag ZAP_FLAG_UNMAP to indicate an unmap call from unmap_vmas(). This is used to indicate the 'final' unmapping of a hugetlb vma. When called via MADV_DONTNEED, this flag is not set and the vm_lock is not deleted. [1] https://lore.kernel.org/lkml/CAO4mrfdLMXsao9RF4fUE8-Wfde8xmjsKrTNMNC9wjUb6J… Link: https://lkml.kernel.org/r/20221111232628.290160-4-mike.kravetz@oracle.com Fixes: 90e7e7f5ef3f ("mm: enable MADV_DONTNEED for hugetlb mappings") Signed-off-by: Mike Kravetz <mike.kravetz(a)oracle.com> Reported-by: Wei Chen <harperchen1110(a)gmail.com> Cc: <stable(a)vger.kernel.org> Cc: Axel Rasmussen <axelrasmussen(a)google.com> Cc: David Hildenbrand <david(a)redhat.com> Cc: Matthew Wilcox <willy(a)infradead.org> Cc: Mina Almasry <almasrymina(a)google.com> Cc: Nadav Amit <nadav.amit(a)gmail.com> Cc: Naoya Horiguchi <naoya.horiguchi(a)linux.dev> Cc: Peter Xu <peterx(a)redhat.com> Cc: Rik van Riel <riel(a)surriel.com> Cc: Vlastimil Babka <vbabka(a)suse.cz> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> --- include/linux/mm.h | 2 ++ mm/hugetlb.c | 27 ++++++++++++++++----------- mm/memory.c | 2 +- 3 files changed, 19 insertions(+), 12 deletions(-) --- a/include/linux/mm.h~hugetlb-dont-delete-vma_lock-in-hugetlb-madv_dontneed-processing +++ a/include/linux/mm.h @@ -1868,6 +1868,8 @@ struct zap_details { * default, the flag is not set. */ #define ZAP_FLAG_DROP_MARKER ((__force zap_flags_t) BIT(0)) +/* Set in unmap_vmas() to indicate a final unmap call. Only used by hugetlb */ +#define ZAP_FLAG_UNMAP ((__force zap_flags_t) BIT(1)) #ifdef CONFIG_MMU extern bool can_do_mlock(void); --- a/mm/hugetlb.c~hugetlb-dont-delete-vma_lock-in-hugetlb-madv_dontneed-processing +++ a/mm/hugetlb.c @@ -5196,17 +5196,22 @@ void __unmap_hugepage_range_final(struct /* mmu notification performed in caller */ __unmap_hugepage_range(tlb, vma, start, end, ref_page, zap_flags); - /* - * Unlock and free the vma lock before releasing i_mmap_rwsem. When - * the vma_lock is freed, this makes the vma ineligible for pmd - * sharing. And, i_mmap_rwsem is required to set up pmd sharing. - * This is important as page tables for this unmapped range will - * be asynchrously deleted. If the page tables are shared, there - * will be issues when accessed by someone else. - */ - __hugetlb_vma_unlock_write_free(vma); - - i_mmap_unlock_write(vma->vm_file->f_mapping); + if (zap_flags & ZAP_FLAG_UNMAP) { /* final unmap */ + /* + * Unlock and free the vma lock before releasing i_mmap_rwsem. + * When the vma_lock is freed, this makes the vma ineligible + * for pmd sharing. And, i_mmap_rwsem is required to set up + * pmd sharing. This is important as page tables for this + * unmapped range will be asynchrously deleted. If the page + * tables are shared, there will be issues when accessed by + * someone else. + */ + __hugetlb_vma_unlock_write_free(vma); + i_mmap_unlock_write(vma->vm_file->f_mapping); + } else { + i_mmap_unlock_write(vma->vm_file->f_mapping); + hugetlb_vma_unlock_write(vma); + } } void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start, --- a/mm/memory.c~hugetlb-dont-delete-vma_lock-in-hugetlb-madv_dontneed-processing +++ a/mm/memory.c @@ -1711,7 +1711,7 @@ void unmap_vmas(struct mmu_gather *tlb, { struct mmu_notifier_range range; struct zap_details details = { - .zap_flags = ZAP_FLAG_DROP_MARKER, + .zap_flags = ZAP_FLAG_DROP_MARKER | ZAP_FLAG_UNMAP, /* Careful - we need to zap private pages too! */ .even_cows = true, }; _ Patches currently in -mm which might be from mike.kravetz(a)oracle.com are ipc-shm-call-underlying-open-close-vm_ops.patch madvise-use-zap_page_range_single-for-madvise-dontneed.patch hugetlb-remove-duplicate-mmu-notifications.patch hugetlb-dont-delete-vma_lock-in-hugetlb-madv_dontneed-processing.patch selftests-vm-update-hugetlb-madvise.patch

2 years, 1 month

1
0
0 0

+ hugetlb-remove-duplicate-mmu-notifications.patch added to mm-hotfixes-unstable branch

by Andrew Morton

The patch titled Subject: hugetlb: remove duplicate mmu notifications has been added to the -mm mm-hotfixes-unstable branch. Its filename is hugetlb-remove-duplicate-mmu-notifications.patch This patch will shortly appear at https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patche… This patch will later appear in the mm-hotfixes-unstable branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm Before you just go and hit "reply", please: a) Consider who else should be cc'ed b) Prefer to cc a suitable mailing list as well c) Ideally: find the original patch on the mailing list and do a reply-to-all to that, adding suitable additional cc's *** Remember to use Documentation/process/submit-checklist.rst when testing your code *** The -mm tree is included into linux-next via the mm-everything branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm and is updated there every 2-3 working days ------------------------------------------------------ From: Mike Kravetz <mike.kravetz(a)oracle.com> Subject: hugetlb: remove duplicate mmu notifications Date: Fri, 11 Nov 2022 15:26:27 -0800 The common hugetlb unmap routine __unmap_hugepage_range performs mmu notification calls. However, in the case where __unmap_hugepage_range is called via __unmap_hugepage_range_final, mmu notification calls are performed earlier in other calling routines. Remove mmu notification calls from __unmap_hugepage_range. Add notification calls to the only other caller: unmap_hugepage_range. unmap_hugepage_range is called for truncation and hole punch, so change notification type from UNMAP to CLEAR as this is more appropriate. Link: https://lkml.kernel.org/r/20221111232628.290160-3-mike.kravetz@oracle.com Fixes: 90e7e7f5ef3f ("mm: enable MADV_DONTNEED for hugetlb mappings") Signed-off-by: Mike Kravetz <mike.kravetz(a)oracle.com> Reported-by: Wei Chen <harperchen1110(a)gmail.com> Cc: <stable(a)vger.kernel.org> Cc: Axel Rasmussen <axelrasmussen(a)google.com> Cc: David Hildenbrand <david(a)redhat.com> Cc: Matthew Wilcox <willy(a)infradead.org> Cc: Mina Almasry <almasrymina(a)google.com> Cc: Nadav Amit <nadav.amit(a)gmail.com> Cc: Naoya Horiguchi <naoya.horiguchi(a)linux.dev> Cc: Peter Xu <peterx(a)redhat.com> Cc: Rik van Riel <riel(a)surriel.com> Cc: Vlastimil Babka <vbabka(a)suse.cz> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> --- mm/hugetlb.c | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-) --- a/mm/hugetlb.c~hugetlb-remove-duplicate-mmu-notifications +++ a/mm/hugetlb.c @@ -5064,7 +5064,6 @@ static void __unmap_hugepage_range(struc struct page *page; struct hstate *h = hstate_vma(vma); unsigned long sz = huge_page_size(h); - struct mmu_notifier_range range; unsigned long last_addr_mask; bool force_flush = false; @@ -5079,13 +5078,6 @@ static void __unmap_hugepage_range(struc tlb_change_page_size(tlb, sz); tlb_start_vma(tlb, vma); - /* - * If sharing possible, alert mmu notifiers of worst case. - */ - mmu_notifier_range_init(&range, MMU_NOTIFY_UNMAP, 0, vma, mm, start, - end); - adjust_range_if_pmd_sharing_possible(vma, &range.start, &range.end); - mmu_notifier_invalidate_range_start(&range); last_addr_mask = hugetlb_mask_last_page(h); address = start; for (; address < end; address += sz) { @@ -5174,7 +5166,6 @@ static void __unmap_hugepage_range(struc if (ref_page) break; } - mmu_notifier_invalidate_range_end(&range); tlb_end_vma(tlb, vma); /* @@ -5202,6 +5193,7 @@ void __unmap_hugepage_range_final(struct hugetlb_vma_lock_write(vma); i_mmap_lock_write(vma->vm_file->f_mapping); + /* mmu notification performed in caller */ __unmap_hugepage_range(tlb, vma, start, end, ref_page, zap_flags); /* @@ -5221,10 +5213,18 @@ void unmap_hugepage_range(struct vm_area unsigned long end, struct page *ref_page, zap_flags_t zap_flags) { + struct mmu_notifier_range range; struct mmu_gather tlb; + mmu_notifier_range_init(&range, MMU_NOTIFY_CLEAR, 0, vma, vma->vm_mm, + start, end); + adjust_range_if_pmd_sharing_possible(vma, &range.start, &range.end); + mmu_notifier_invalidate_range_start(&range); tlb_gather_mmu(&tlb, vma->vm_mm); + __unmap_hugepage_range(&tlb, vma, start, end, ref_page, zap_flags); + + mmu_notifier_invalidate_range_end(&range); tlb_finish_mmu(&tlb); } _ Patches currently in -mm which might be from mike.kravetz(a)oracle.com are ipc-shm-call-underlying-open-close-vm_ops.patch madvise-use-zap_page_range_single-for-madvise-dontneed.patch hugetlb-remove-duplicate-mmu-notifications.patch hugetlb-dont-delete-vma_lock-in-hugetlb-madv_dontneed-processing.patch selftests-vm-update-hugetlb-madvise.patch

2 years, 1 month

1
0
0 0

+ madvise-use-zap_page_range_single-for-madvise-dontneed.patch added to mm-hotfixes-unstable branch

by Andrew Morton

The patch titled Subject: madvise: use zap_page_range_single for madvise dontneed has been added to the -mm mm-hotfixes-unstable branch. Its filename is madvise-use-zap_page_range_single-for-madvise-dontneed.patch This patch will shortly appear at https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patche… This patch will later appear in the mm-hotfixes-unstable branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm Before you just go and hit "reply", please: a) Consider who else should be cc'ed b) Prefer to cc a suitable mailing list as well c) Ideally: find the original patch on the mailing list and do a reply-to-all to that, adding suitable additional cc's *** Remember to use Documentation/process/submit-checklist.rst when testing your code *** The -mm tree is included into linux-next via the mm-everything branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm and is updated there every 2-3 working days ------------------------------------------------------ From: Mike Kravetz <mike.kravetz(a)oracle.com> Subject: madvise: use zap_page_range_single for madvise dontneed Date: Fri, 11 Nov 2022 15:26:26 -0800 Patch series "fix hugetlb MADV_DONTNEED vma_lock handling", v9. This series addresses the issue first reported in [1], and fully described in patch 3. While exploring solutions to this issue, related problems with mmu notification calls were discovered. The first two patches address those issues. Previous discussions suggested further cleanup by removing the routine zap_page_range. This is possible because zap_page_range_single is now exported, and all callers of zap_page_range pass ranges entirely within a single vma. This work will be done in a later patch so as not to distract from this bug fix. [1] https://lore.kernel.org/lkml/CAO4mrfdLMXsao9RF4fUE8-Wfde8xmjsKrTNMNC9wjUb6J… This patch (of 3): Expose the routine zap_page_range_single to zap a range within a single vma. The madvise routine madvise_dontneed_single_vma can use this routine as it explicitly operates on a single vma. Also, update the mmu notification range in zap_page_range_single to take hugetlb pmd sharing into account. This is required as MADV_DONTNEED supports hugetlb vmas. Link: https://lkml.kernel.org/r/20221111232628.290160-1-mike.kravetz@oracle.com Link: https://lkml.kernel.org/r/20221111232628.290160-2-mike.kravetz@oracle.com Fixes: 90e7e7f5ef3f ("mm: enable MADV_DONTNEED for hugetlb mappings") Signed-off-by: Mike Kravetz <mike.kravetz(a)oracle.com> Reported-by: Wei Chen <harperchen1110(a)gmail.com> Cc: Axel Rasmussen <axelrasmussen(a)google.com> Cc: David Hildenbrand <david(a)redhat.com> Cc: Matthew Wilcox <willy(a)infradead.org> Cc: Mike Kravetz <mike.kravetz(a)oracle.com> Cc: Mina Almasry <almasrymina(a)google.com> Cc: Nadav Amit <nadav.amit(a)gmail.com> Cc: Naoya Horiguchi <naoya.horiguchi(a)linux.dev> Cc: Peter Xu <peterx(a)redhat.com> Cc: Rik van Riel <riel(a)surriel.com> Cc: Vlastimil Babka <vbabka(a)suse.cz> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> --- include/linux/mm.h | 27 +++++++++++++++++++-------- mm/madvise.c | 6 +++--- mm/memory.c | 23 +++++++++++------------ 3 files changed, 33 insertions(+), 23 deletions(-) --- a/include/linux/mm.h~madvise-use-zap_page_range_single-for-madvise-dontneed +++ a/include/linux/mm.h @@ -1852,6 +1852,23 @@ static void __maybe_unused show_free_are __show_free_areas(flags, nodemask, MAX_NR_ZONES - 1); } +/* + * Parameter block passed down to zap_pte_range in exceptional cases. + */ +struct zap_details { + struct folio *single_folio; /* Locked folio to be unmapped */ + bool even_cows; /* Zap COWed private pages too? */ + zap_flags_t zap_flags; /* Extra flags for zapping */ +}; + +/* + * Whether to drop the pte markers, for example, the uffd-wp information for + * file-backed memory. This should only be specified when we will completely + * drop the page in the mm, either by truncation or unmapping of the vma. By + * default, the flag is not set. + */ +#define ZAP_FLAG_DROP_MARKER ((__force zap_flags_t) BIT(0)) + #ifdef CONFIG_MMU extern bool can_do_mlock(void); #else @@ -1869,6 +1886,8 @@ void zap_vma_ptes(struct vm_area_struct unsigned long size); void zap_page_range(struct vm_area_struct *vma, unsigned long address, unsigned long size); +void zap_page_range_single(struct vm_area_struct *vma, unsigned long address, + unsigned long size, struct zap_details *details); void unmap_vmas(struct mmu_gather *tlb, struct maple_tree *mt, struct vm_area_struct *start_vma, unsigned long start, unsigned long end); @@ -3467,12 +3486,4 @@ madvise_set_anon_name(struct mm_struct * } #endif -/* - * Whether to drop the pte markers, for example, the uffd-wp information for - * file-backed memory. This should only be specified when we will completely - * drop the page in the mm, either by truncation or unmapping of the vma. By - * default, the flag is not set. - */ -#define ZAP_FLAG_DROP_MARKER ((__force zap_flags_t) BIT(0)) - #endif /* _LINUX_MM_H */ --- a/mm/madvise.c~madvise-use-zap_page_range_single-for-madvise-dontneed +++ a/mm/madvise.c @@ -772,8 +772,8 @@ static int madvise_free_single_vma(struc * Application no longer needs these pages. If the pages are dirty, * it's OK to just throw them away. The app will be more careful about * data it wants to keep. Be sure to free swap resources too. The - * zap_page_range call sets things up for shrink_active_list to actually free - * these pages later if no one else has touched them in the meantime, + * zap_page_range_single call sets things up for shrink_active_list to actually + * free these pages later if no one else has touched them in the meantime, * although we could add these pages to a global reuse list for * shrink_active_list to pick up before reclaiming other pages. * @@ -790,7 +790,7 @@ static int madvise_free_single_vma(struc static long madvise_dontneed_single_vma(struct vm_area_struct *vma, unsigned long start, unsigned long end) { - zap_page_range(vma, start, end - start); + zap_page_range_single(vma, start, end - start, NULL); return 0; } --- a/mm/memory.c~madvise-use-zap_page_range_single-for-madvise-dontneed +++ a/mm/memory.c @@ -1341,15 +1341,6 @@ copy_page_range(struct vm_area_struct *d return ret; } -/* - * Parameter block passed down to zap_pte_range in exceptional cases. - */ -struct zap_details { - struct folio *single_folio; /* Locked folio to be unmapped */ - bool even_cows; /* Zap COWed private pages too? */ - zap_flags_t zap_flags; /* Extra flags for zapping */ -}; - /* Whether we should zap all COWed (private) pages too */ static inline bool should_zap_cows(struct zap_details *details) { @@ -1774,19 +1765,27 @@ void zap_page_range(struct vm_area_struc * * The range must fit into one VMA. */ -static void zap_page_range_single(struct vm_area_struct *vma, unsigned long address, +void zap_page_range_single(struct vm_area_struct *vma, unsigned long address, unsigned long size, struct zap_details *details) { + unsigned long end = address + size; struct mmu_notifier_range range; struct mmu_gather tlb; lru_add_drain(); mmu_notifier_range_init(&range, MMU_NOTIFY_CLEAR, 0, vma, vma->vm_mm, - address, address + size); + address, end); + if (is_vm_hugetlb_page(vma)) + adjust_range_if_pmd_sharing_possible(vma, &range.start, + &range.end); tlb_gather_mmu(&tlb, vma->vm_mm); update_hiwater_rss(vma->vm_mm); mmu_notifier_invalidate_range_start(&range); - unmap_single_vma(&tlb, vma, address, range.end, details); + /* + * unmap 'address-end' not 'range.start-range.end' as range + * could have been expanded for hugetlb pmd sharing. + */ + unmap_single_vma(&tlb, vma, address, end, details); mmu_notifier_invalidate_range_end(&range); tlb_finish_mmu(&tlb); } _ Patches currently in -mm which might be from mike.kravetz(a)oracle.com are ipc-shm-call-underlying-open-close-vm_ops.patch madvise-use-zap_page_range_single-for-madvise-dontneed.patch hugetlb-remove-duplicate-mmu-notifications.patch hugetlb-dont-delete-vma_lock-in-hugetlb-madv_dontneed-processing.patch selftests-vm-update-hugetlb-madvise.patch

2 years, 1 month

1
0
0 0

[PATCH v9 3/3] hugetlb: don't delete vma_lock in hugetlb MADV_DONTNEED processing

by Mike Kravetz

madvise(MADV_DONTNEED) ends up calling zap_page_range() to clear page tables associated with the address range. For hugetlb vmas, zap_page_range will call __unmap_hugepage_range_final. However, __unmap_hugepage_range_final assumes the passed vma is about to be removed and deletes the vma_lock to prevent pmd sharing as the vma is on the way out. In the case of madvise(MADV_DONTNEED) the vma remains, but the missing vma_lock prevents pmd sharing and could potentially lead to issues with truncation/fault races. This issue was originally reported here [1] as a BUG triggered in page_try_dup_anon_rmap. Prior to the introduction of the hugetlb vma_lock, __unmap_hugepage_range_final cleared the VM_MAYSHARE flag to prevent pmd sharing. Subsequent faults on this vma were confused as VM_MAYSHARE indicates a sharable vma, but was not set so page_mapping was not set in new pages added to the page table. This resulted in pages that appeared anonymous in a VM_SHARED vma and triggered the BUG. Address issue by adding a new zap flag ZAP_FLAG_UNMAP to indicate an unmap call from unmap_vmas(). This is used to indicate the 'final' unmapping of a hugetlb vma. When called via MADV_DONTNEED, this flag is not set and the vm_lock is not deleted. [1] https://lore.kernel.org/lkml/CAO4mrfdLMXsao9RF4fUE8-Wfde8xmjsKrTNMNC9wjUb6J… Fixes: 90e7e7f5ef3f ("mm: enable MADV_DONTNEED for hugetlb mappings") Signed-off-by: Mike Kravetz <mike.kravetz(a)oracle.com> Reported-by: Wei Chen <harperchen1110(a)gmail.com> Cc: <stable(a)vger.kernel.org> --- include/linux/mm.h | 2 ++ mm/hugetlb.c | 27 ++++++++++++++++----------- mm/memory.c | 2 +- 3 files changed, 19 insertions(+), 12 deletions(-) diff --git a/include/linux/mm.h b/include/linux/mm.h index 9e7cad65dfde..d0e8d5007343 100644 --- a/include/linux/mm.h +++ b/include/linux/mm.h @@ -1897,6 +1897,8 @@ struct zap_details { * default, the flag is not set. */ #define ZAP_FLAG_DROP_MARKER ((__force zap_flags_t) BIT(0)) +/* Set in unmap_vmas() to indicate a final unmap call. Only used by hugetlb */ +#define ZAP_FLAG_UNMAP ((__force zap_flags_t) BIT(1)) #ifdef CONFIG_MMU extern bool can_do_mlock(void); diff --git a/mm/hugetlb.c b/mm/hugetlb.c index 205c67c6787a..0cdefa63f474 100644 --- a/mm/hugetlb.c +++ b/mm/hugetlb.c @@ -5202,17 +5202,22 @@ void __unmap_hugepage_range_final(struct mmu_gather *tlb, /* mmu notification performed in caller */ __unmap_hugepage_range(tlb, vma, start, end, ref_page, zap_flags); - /* - * Unlock and free the vma lock before releasing i_mmap_rwsem. When - * the vma_lock is freed, this makes the vma ineligible for pmd - * sharing. And, i_mmap_rwsem is required to set up pmd sharing. - * This is important as page tables for this unmapped range will - * be asynchrously deleted. If the page tables are shared, there - * will be issues when accessed by someone else. - */ - __hugetlb_vma_unlock_write_free(vma); - - i_mmap_unlock_write(vma->vm_file->f_mapping); + if (zap_flags & ZAP_FLAG_UNMAP) { /* final unmap */ + /* + * Unlock and free the vma lock before releasing i_mmap_rwsem. + * When the vma_lock is freed, this makes the vma ineligible + * for pmd sharing. And, i_mmap_rwsem is required to set up + * pmd sharing. This is important as page tables for this + * unmapped range will be asynchrously deleted. If the page + * tables are shared, there will be issues when accessed by + * someone else. + */ + __hugetlb_vma_unlock_write_free(vma); + i_mmap_unlock_write(vma->vm_file->f_mapping); + } else { + i_mmap_unlock_write(vma->vm_file->f_mapping); + hugetlb_vma_unlock_write(vma); + } } void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start, diff --git a/mm/memory.c b/mm/memory.c index ebdbd395cfad..2d453736f87c 100644 --- a/mm/memory.c +++ b/mm/memory.c @@ -1673,7 +1673,7 @@ void unmap_vmas(struct mmu_gather *tlb, struct maple_tree *mt, { struct mmu_notifier_range range; struct zap_details details = { - .zap_flags = ZAP_FLAG_DROP_MARKER, + .zap_flags = ZAP_FLAG_DROP_MARKER | ZAP_FLAG_UNMAP, /* Careful - we need to zap private pages too! */ .even_cows = true, }; -- 2.37.3

2 years, 1 month

1
0
0 0

+ gcov-clang-fix-the-buffer-overflow-issue.patch added to mm-hotfixes-unstable branch

by Andrew Morton

The patch titled Subject: gcov: clang: fix the buffer overflow issue has been added to the -mm mm-hotfixes-unstable branch. Its filename is gcov-clang-fix-the-buffer-overflow-issue.patch This patch will shortly appear at https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patche… This patch will later appear in the mm-hotfixes-unstable branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm Before you just go and hit "reply", please: a) Consider who else should be cc'ed b) Prefer to cc a suitable mailing list as well c) Ideally: find the original patch on the mailing list and do a reply-to-all to that, adding suitable additional cc's *** Remember to use Documentation/process/submit-checklist.rst when testing your code *** The -mm tree is included into linux-next via the mm-everything branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm and is updated there every 2-3 working days ------------------------------------------------------ From: Mukesh Ojha <quic_mojha(a)quicinc.com> Subject: gcov: clang: fix the buffer overflow issue Date: Thu, 10 Nov 2022 00:31:37 +0530 Currently, in clang version of gcov code when module is getting removed gcov_info_add() incorrectly adds the sfn_ptr->counter to all the dst->functions and it result in the kernel panic in below crash report. Fix this by properly handling it. [ 8.899094][ T599] Unable to handle kernel write to read-only memory at virtual address ffffff80461cc000 [ 8.899100][ T599] Mem abort info: [ 8.899102][ T599] ESR = 0x9600004f [ 8.899103][ T599] EC = 0x25: DABT (current EL), IL = 32 bits [ 8.899105][ T599] SET = 0, FnV = 0 [ 8.899107][ T599] EA = 0, S1PTW = 0 [ 8.899108][ T599] FSC = 0x0f: level 3 permission fault [ 8.899110][ T599] Data abort info: [ 8.899111][ T599] ISV = 0, ISS = 0x0000004f [ 8.899113][ T599] CM = 0, WnR = 1 [ 8.899114][ T599] swapper pgtable: 4k pages, 39-bit VAs, pgdp=00000000ab8de000 [ 8.899116][ T599] [ffffff80461cc000] pgd=18000009ffcde003, p4d=18000009ffcde003, pud=18000009ffcde003, pmd=18000009ffcad003, pte=00600000c61cc787 [ 8.899124][ T599] Internal error: Oops: 9600004f [#1] PREEMPT SMP [ 8.899265][ T599] Skip md ftrace buffer dump for: 0x1609e0 .... .., [ 8.899544][ T599] CPU: 7 PID: 599 Comm: modprobe Tainted: G S OE 5.15.41-android13-8-g38e9b1af6bce #1 [ 8.899547][ T599] Hardware name: XXX (DT) [ 8.899549][ T599] pstate: 82400005 (Nzcv daif +PAN -UAO +TCO -DIT -SSBS BTYPE=--) [ 8.899551][ T599] pc : gcov_info_add+0x9c/0xb8 [ 8.899557][ T599] lr : gcov_event+0x28c/0x6b8 [ 8.899559][ T599] sp : ffffffc00e733b00 [ 8.899560][ T599] x29: ffffffc00e733b00 x28: ffffffc00e733d30 x27: ffffffe8dc297470 [ 8.899563][ T599] x26: ffffffe8dc297000 x25: ffffffe8dc297000 x24: ffffffe8dc297000 [ 8.899566][ T599] x23: ffffffe8dc0a6200 x22: ffffff880f68bf20 x21: 0000000000000000 [ 8.899569][ T599] x20: ffffff880f68bf00 x19: ffffff8801babc00 x18: ffffffc00d7f9058 [ 8.899572][ T599] x17: 0000000000088793 x16: ffffff80461cbe00 x15: 9100052952800785 [ 8.899575][ T599] x14: 0000000000000200 x13: 0000000000000041 x12: 9100052952800785 [ 8.899577][ T599] x11: ffffffe8dc297000 x10: ffffffe8dc297000 x9 : ffffff80461cbc80 [ 8.899580][ T599] x8 : ffffff8801babe80 x7 : ffffffe8dc2ec000 x6 : ffffffe8dc2ed000 [ 8.899583][ T599] x5 : 000000008020001f x4 : fffffffe2006eae0 x3 : 000000008020001f [ 8.899586][ T599] x2 : ffffff8027c49200 x1 : ffffff8801babc20 x0 : ffffff80461cb3a0 [ 8.899589][ T599] Call trace: [ 8.899590][ T599] gcov_info_add+0x9c/0xb8 [ 8.899592][ T599] gcov_module_notifier+0xbc/0x120 [ 8.899595][ T599] blocking_notifier_call_chain+0xa0/0x11c [ 8.899598][ T599] do_init_module+0x2a8/0x33c [ 8.899600][ T599] load_module+0x23cc/0x261c [ 8.899602][ T599] __arm64_sys_finit_module+0x158/0x194 [ 8.899604][ T599] invoke_syscall+0x94/0x2bc [ 8.899607][ T599] el0_svc_common+0x1d8/0x34c [ 8.899609][ T599] do_el0_svc+0x40/0x54 [ 8.899611][ T599] el0_svc+0x94/0x2f0 [ 8.899613][ T599] el0t_64_sync_handler+0x88/0xec [ 8.899615][ T599] el0t_64_sync+0x1b4/0x1b8 [ 8.899618][ T599] Code: f905f56c f86e69ec f86e6a0f 8b0c01ec (f82e6a0c) [ 8.899620][ T599] ---[ end trace ed5218e9e5b6e2e6 ]--- Link: https://lkml.kernel.org/r/1668020497-13142-1-git-send-email-quic_mojha@quic… Fixes: e178a5beb369 ("gcov: clang support") Signed-off-by: Mukesh Ojha <quic_mojha(a)quicinc.com> Reviewed-by: Peter Oberparleiter <oberpar(a)linux.ibm.com> Tested-by: Peter Oberparleiter <oberpar(a)linux.ibm.com> Cc: Nathan Chancellor <nathan(a)kernel.org> Cc: Nick Desaulniers <ndesaulniers(a)google.com> Cc: Tom Rix <trix(a)redhat.com> Cc: <stable(a)vger.kernel.org> [5.2+] Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> --- kernel/gcov/clang.c | 2 ++ 1 file changed, 2 insertions(+) --- a/kernel/gcov/clang.c~gcov-clang-fix-the-buffer-overflow-issue +++ a/kernel/gcov/clang.c @@ -280,6 +280,8 @@ void gcov_info_add(struct gcov_info *dst for (i = 0; i < sfn_ptr->num_counters; i++) dfn_ptr->counters[i] += sfn_ptr->counters[i]; + + sfn_ptr = list_next_entry(sfn_ptr, head); } } _ Patches currently in -mm which might be from quic_mojha(a)quicinc.com are gcov-clang-fix-the-buffer-overflow-issue.patch

2 years, 1 month

1
0
0 0

[to-be-updated] hugetlb-dont-delete-vma_lock-in-hugetlb-madv_dontneed-processing.patch removed from -mm tree

by Andrew Morton

The quilt patch titled Subject: hugetlb: don't delete vma_lock in hugetlb MADV_DONTNEED processing has been removed from the -mm tree. Its filename was hugetlb-dont-delete-vma_lock-in-hugetlb-madv_dontneed-processing.patch This patch was dropped because an updated version will be merged ------------------------------------------------------ From: Mike Kravetz <mike.kravetz(a)oracle.com> Subject: hugetlb: don't delete vma_lock in hugetlb MADV_DONTNEED processing Date: Tue, 1 Nov 2022 18:31:00 -0700 madvise(MADV_DONTNEED) ends up calling zap_page_range() to clear page tables associated with the address range. For hugetlb vmas, zap_page_range will call __unmap_hugepage_range_final. However, __unmap_hugepage_range_final assumes the passed vma is about to be removed and deletes the vma_lock to prevent pmd sharing as the vma is on the way out. In the case of madvise(MADV_DONTNEED) the vma remains, but the missing vma_lock prevents pmd sharing and could potentially lead to issues with truncation/fault races. This issue was originally reported here [1] as a BUG triggered in page_try_dup_anon_rmap. Prior to the introduction of the hugetlb vma_lock, __unmap_hugepage_range_final cleared the VM_MAYSHARE flag to prevent pmd sharing. Subsequent faults on this vma were confused as VM_MAYSHARE indicates a sharable vma, but was not set so page_mapping was not set in new pages added to the page table. This resulted in pages that appeared anonymous in a VM_SHARED vma and triggered the BUG. Address issue by: - Add a new zap flag ZAP_FLAG_UNMAP to indicate an unmap call from unmap_vmas(). This is used to indicate the 'final' unmapping of a vma. When called via MADV_DONTNEED, this flag is not set and the vm_lock is not deleted. - mmu notification is removed from __unmap_hugepage_range to avoid duplication, and notification is added to the other calling routine (unmap_hugepage_range). - notification calls are updated in zap_page range to take into account the possibility of multiple vmas. [1] https://lore.kernel.org/lkml/CAO4mrfdLMXsao9RF4fUE8-Wfde8xmjsKrTNMNC9wjUb6J… Link: https://lkml.kernel.org/r/20221102013100.455139-1-mike.kravetz@oracle.com Fixes: 90e7e7f5ef3f ("mm: enable MADV_DONTNEED for hugetlb mappings") Signed-off-by: Mike Kravetz <mike.kravetz(a)oracle.com> Reported-by: Wei Chen <harperchen1110(a)gmail.com> Cc: Axel Rasmussen <axelrasmussen(a)google.com> Cc: David Hildenbrand <david(a)redhat.com> Cc: Matthew Wilcox (Oracle) <willy(a)infradead.org> Cc: Mike Kravetz <mike.kravetz(a)oracle.com> Cc: Mina Almasry <almasrymina(a)google.com> Cc: Nadav Amit <nadav.amit(a)gmail.com> Cc: Naoya Horiguchi <naoya.horiguchi(a)linux.dev> Cc: Peter Xu <peterx(a)redhat.com> Cc: Rik van Riel <riel(a)surriel.com> Cc: Vlastimil Babka <vbabka(a)suse.cz> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> --- include/linux/mm.h | 3 ++ mm/hugetlb.c | 45 +++++++++++++++++++++++++------------------ mm/memory.c | 21 ++++++++++++++------ 3 files changed, 45 insertions(+), 24 deletions(-) --- a/include/linux/mm.h~hugetlb-dont-delete-vma_lock-in-hugetlb-madv_dontneed-processing +++ a/include/linux/mm.h @@ -3475,4 +3475,7 @@ madvise_set_anon_name(struct mm_struct * */ #define ZAP_FLAG_DROP_MARKER ((__force zap_flags_t) BIT(0)) +/* Set in unmap_vmas() to indicate an unmap call. Only used by hugetlb */ +#define ZAP_FLAG_UNMAP ((__force zap_flags_t) BIT(1)) + #endif /* _LINUX_MM_H */ --- a/mm/hugetlb.c~hugetlb-dont-delete-vma_lock-in-hugetlb-madv_dontneed-processing +++ a/mm/hugetlb.c @@ -5064,7 +5064,6 @@ static void __unmap_hugepage_range(struc struct page *page; struct hstate *h = hstate_vma(vma); unsigned long sz = huge_page_size(h); - struct mmu_notifier_range range; unsigned long last_addr_mask; bool force_flush = false; @@ -5079,13 +5078,6 @@ static void __unmap_hugepage_range(struc tlb_change_page_size(tlb, sz); tlb_start_vma(tlb, vma); - /* - * If sharing possible, alert mmu notifiers of worst case. - */ - mmu_notifier_range_init(&range, MMU_NOTIFY_UNMAP, 0, vma, mm, start, - end); - adjust_range_if_pmd_sharing_possible(vma, &range.start, &range.end); - mmu_notifier_invalidate_range_start(&range); last_addr_mask = hugetlb_mask_last_page(h); address = start; for (; address < end; address += sz) { @@ -5174,7 +5166,6 @@ static void __unmap_hugepage_range(struc if (ref_page) break; } - mmu_notifier_invalidate_range_end(&range); tlb_end_vma(tlb, vma); /* @@ -5199,32 +5190,50 @@ void __unmap_hugepage_range_final(struct unsigned long end, struct page *ref_page, zap_flags_t zap_flags) { + bool final = zap_flags & ZAP_FLAG_UNMAP; + hugetlb_vma_lock_write(vma); i_mmap_lock_write(vma->vm_file->f_mapping); __unmap_hugepage_range(tlb, vma, start, end, ref_page, zap_flags); /* - * Unlock and free the vma lock before releasing i_mmap_rwsem. When - * the vma_lock is freed, this makes the vma ineligible for pmd - * sharing. And, i_mmap_rwsem is required to set up pmd sharing. - * This is important as page tables for this unmapped range will - * be asynchrously deleted. If the page tables are shared, there - * will be issues when accessed by someone else. + * When called via zap_page_range (MADV_DONTNEED), this is not the + * final unmap of the vma, and we do not want to delete the vma_lock. */ - __hugetlb_vma_unlock_write_free(vma); - - i_mmap_unlock_write(vma->vm_file->f_mapping); + if (final) { + /* + * Unlock and free the vma lock before releasing i_mmap_rwsem. + * When the vma_lock is freed, this makes the vma ineligible + * for pmd sharing. And, i_mmap_rwsem is required to set up + * pmd sharing. This is important as page tables for this + * unmapped range will be asynchrously deleted. If the page + * tables are shared, there will be issues when accessed by + * someone else. + */ + __hugetlb_vma_unlock_write_free(vma); + i_mmap_unlock_write(vma->vm_file->f_mapping); + } else { + i_mmap_unlock_write(vma->vm_file->f_mapping); + hugetlb_vma_unlock_write(vma); + } } void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start, unsigned long end, struct page *ref_page, zap_flags_t zap_flags) { + struct mmu_notifier_range range; struct mmu_gather tlb; + mmu_notifier_range_init(&range, MMU_NOTIFY_UNMAP, 0, vma, vma->vm_mm, + start, end); + adjust_range_if_pmd_sharing_possible(vma, &range.start, &range.end); tlb_gather_mmu(&tlb, vma->vm_mm); + __unmap_hugepage_range(&tlb, vma, start, end, ref_page, zap_flags); + + mmu_notifier_invalidate_range_end(&range); tlb_finish_mmu(&tlb); } --- a/mm/memory.c~hugetlb-dont-delete-vma_lock-in-hugetlb-madv_dontneed-processing +++ a/mm/memory.c @@ -1720,7 +1720,7 @@ void unmap_vmas(struct mmu_gather *tlb, { struct mmu_notifier_range range; struct zap_details details = { - .zap_flags = ZAP_FLAG_DROP_MARKER, + .zap_flags = ZAP_FLAG_DROP_MARKER | ZAP_FLAG_UNMAP, /* Careful - we need to zap private pages too! */ .even_cows = true, }; @@ -1753,15 +1753,24 @@ void zap_page_range(struct vm_area_struc MA_STATE(mas, mt, vma->vm_end, vma->vm_end); lru_add_drain(); - mmu_notifier_range_init(&range, MMU_NOTIFY_CLEAR, 0, vma, vma->vm_mm, - start, start + size); tlb_gather_mmu(&tlb, vma->vm_mm); update_hiwater_rss(vma->vm_mm); - mmu_notifier_invalidate_range_start(&range); do { - unmap_single_vma(&tlb, vma, start, range.end, NULL); + mmu_notifier_range_init(&range, MMU_NOTIFY_CLEAR, 0, vma, + vma->vm_mm, + max(start, vma->vm_start), + min(end, vma->vm_end)); + if (is_vm_hugetlb_page(vma)) + adjust_range_if_pmd_sharing_possible(vma, + &range.start, &range.end); + mmu_notifier_invalidate_range_start(&range); + /* + * unmap 'start-end' not 'range.start-range.end' as range + * could have been expanded for pmd sharing. + */ + unmap_single_vma(&tlb, vma, start, end, NULL); + mmu_notifier_invalidate_range_end(&range); } while ((vma = mas_find(&mas, end - 1)) != NULL); - mmu_notifier_invalidate_range_end(&range); tlb_finish_mmu(&tlb); } _ Patches currently in -mm which might be from mike.kravetz(a)oracle.com are ipc-shm-call-underlying-open-close-vm_ops.patch selftests-vm-update-hugetlb-madvise.patch

2 years, 1 month

1
0
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror November 2022