August 2024 - Linux-stable-mirror

[PATCH V4] scsi: ufs: qcom: update MODE_MAX cfg_bw value

by Manish Pandey

The cfg_bw value for max mode is incorrect for the Qualcomm SoC. Update it to the correct value for cfg_bw max mode. Fixes: 03ce80a1bb86 ("scsi: ufs: qcom: Add support for scaling interconnects") Cc: stable(a)vger.kernel.org Signed-off-by: Manish Pandey <quic_mapa(a)quicinc.com> --- Changes from v3: - Cced stable(a)vger.kernel.org. Changes from v2: - Addressed Mani comment, added fixes tag. Changes from v1: - Updated commit message. --- drivers/ufs/host/ufs-qcom.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/ufs/host/ufs-qcom.c b/drivers/ufs/host/ufs-qcom.c index c87fdc849c62..ecdfff2456e3 100644 --- a/drivers/ufs/host/ufs-qcom.c +++ b/drivers/ufs/host/ufs-qcom.c @@ -93,7 +93,7 @@ static const struct __ufs_qcom_bw_table { [MODE_HS_RB][UFS_HS_G3][UFS_LANE_2] = { 1492582, 204800 }, [MODE_HS_RB][UFS_HS_G4][UFS_LANE_2] = { 2915200, 409600 }, [MODE_HS_RB][UFS_HS_G5][UFS_LANE_2] = { 5836800, 819200 }, - [MODE_MAX][0][0] = { 7643136, 307200 }, + [MODE_MAX][0][0] = { 7643136, 819200 }, }; static void ufs_qcom_get_default_testbus_cfg(struct ufs_qcom_host *host); -- 2.17.1

8 months, 3 weeks

2
1
0 0

[PATCH] sched/syscalls: Allow setting niceness using sched_param struct

by Michael Pratt

From userspace, spawning a new process with, for example, posix_spawn(), only allows the user to work with the scheduling priority value defined by POSIX in the sched_param struct. However, sched_setparam() and similar syscalls lead to __sched_setscheduler() which rejects any new value for the priority other than 0 for non-RT schedule classes, a behavior kept since Linux 2.6 or earlier. Linux translates the usage of the sched_param struct into it's own internal sched_attr struct during the syscall, but the user has no way to manage the other values within the sched_attr struct using only POSIX functions. The only other way to adjust niceness while using posix_spawn() would be to set the value after the process has started, but this introduces the risk of the process being dead before the next syscall can set the priority after the fact. To resolve this, allow the use of the priority value originally from the POSIX sched_param struct in order to set the niceness value instead of rejecting the priority value. Edit the sched_get_priority_*() POSIX syscalls in order to reflect the range of values accepted. Cc: stable(a)vger.kernel.org # Apply to kernel/sched/core.c Signed-off-by: Michael Pratt <mcpratt(a)pm.me> --- kernel/sched/syscalls.c | 21 +++++++++++++++++++-- 1 file changed, 19 insertions(+), 2 deletions(-) diff --git a/kernel/sched/syscalls.c b/kernel/sched/syscalls.c index ae1b42775ef9..52c02b80f037 100644 --- a/kernel/sched/syscalls.c +++ b/kernel/sched/syscalls.c @@ -853,6 +853,19 @@ static int _sched_setscheduler(struct task_struct *p, int policy, attr.sched_policy = policy; } + if (attr.sched_priority > MAX_PRIO-1) + return -EINVAL; + + /* + * If priority is set for SCHED_NORMAL or SCHED_BATCH, + * set the niceness instead, but only for user calls. + */ + if (check && attr.sched_priority > MAX_RT_PRIO-1 && + ((policy != SETPARAM_POLICY && fair_policy(policy)) || fair_policy(p->policy))) { + attr.sched_nice = PRIO_TO_NICE(attr.sched_priority); + attr.sched_priority = 0; + } + return __sched_setscheduler(p, &attr, check, true); } /** @@ -1598,9 +1611,11 @@ SYSCALL_DEFINE1(sched_get_priority_max, int, policy) case SCHED_RR: ret = MAX_RT_PRIO-1; break; - case SCHED_DEADLINE: case SCHED_NORMAL: case SCHED_BATCH: + ret = MAX_PRIO-1; + break; + case SCHED_DEADLINE: case SCHED_IDLE: ret = 0; break; @@ -1625,9 +1640,11 @@ SYSCALL_DEFINE1(sched_get_priority_min, int, policy) case SCHED_RR: ret = 1; break; - case SCHED_DEADLINE: case SCHED_NORMAL: case SCHED_BATCH: + ret = MAX_RT_PRIO; + break; + case SCHED_DEADLINE: case SCHED_IDLE: ret = 0; } base-commit: 5be63fc19fcaa4c236b307420483578a56986a37 -- 2.30.2

8 months, 3 weeks

1
0
0 0

[PATCH] media: i2c: ar0521: Use cansleep version of gpiod_set_value()

by Alexander Shiyan

If we use GPIO reset from I2C port expander, we must use *_cansleep() variant of GPIO functions. This was not done in ar0521_power_on()/ar0521_power_off() functions. Let's fix that. ------------[ cut here ]------------ WARNING: CPU: 0 PID: 11 at drivers/gpio/gpiolib.c:3496 gpiod_set_value+0x74/0x7c Modules linked in: CPU: 0 PID: 11 Comm: kworker/u16:0 Not tainted 6.10.0 #53 Hardware name: Diasom DS-RK3568-SOM-EVB (DT) Workqueue: events_unbound deferred_probe_work_func pstate: 80400009 (Nzcv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--) pc : gpiod_set_value+0x74/0x7c lr : ar0521_power_on+0xcc/0x290 sp : ffffff8001d7ab70 x29: ffffff8001d7ab70 x28: ffffff80027dcc90 x27: ffffff8003c82000 x26: ffffff8003ca9250 x25: ffffffc080a39c60 x24: ffffff8003ca9088 x23: ffffff8002402720 x22: ffffff8003ca9080 x21: ffffff8003ca9088 x20: 0000000000000000 x19: ffffff8001eb2a00 x18: ffffff80efeeac80 x17: 756d2d6332692f30 x16: 0000000000000000 x15: 0000000000000000 x14: ffffff8001d91d40 x13: 0000000000000016 x12: ffffffc080e98930 x11: ffffff8001eb2880 x10: 0000000000000890 x9 : ffffff8001d7a9f0 x8 : ffffff8001d92570 x7 : ffffff80efeeac80 x6 : 000000003fc6e780 x5 : ffffff8001d91c80 x4 : 0000000000000002 x3 : 0000000000000000 x2 : 0000000000000000 x1 : 0000000000000000 x0 : 0000000000000001 Call trace: gpiod_set_value+0x74/0x7c ar0521_power_on+0xcc/0x290 ... Signed-off-by: Alexander Shiyan <eagle.alexander923(a)gmail.com> Fixes: 852b50aeed15 ("media: On Semi AR0521 sensor driver") Cc: stable(a)vger.kernel.org --- drivers/media/i2c/ar0521.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/media/i2c/ar0521.c b/drivers/media/i2c/ar0521.c index 09331cf95c62..d557f3b3de3d 100644 --- a/drivers/media/i2c/ar0521.c +++ b/drivers/media/i2c/ar0521.c @@ -844,7 +844,8 @@ static int ar0521_power_off(struct device *dev) clk_disable_unprepare(sensor->extclk); if (sensor->reset_gpio) - gpiod_set_value(sensor->reset_gpio, 1); /* assert RESET signal */ + /* assert RESET signal */ + gpiod_set_value_cansleep(sensor->reset_gpio, 1); for (i = ARRAY_SIZE(ar0521_supply_names) - 1; i >= 0; i--) { if (sensor->supplies[i]) @@ -878,7 +879,7 @@ static int ar0521_power_on(struct device *dev) if (sensor->reset_gpio) /* deassert RESET signal */ - gpiod_set_value(sensor->reset_gpio, 0); + gpiod_set_value_cansleep(sensor->reset_gpio, 0); usleep_range(4500, 5000); /* min 45000 clocks */ for (cnt = 0; cnt < ARRAY_SIZE(initial_regs); cnt++) { -- 2.39.1

8 months, 3 weeks

2
1
0 0

[PATCH net] RDMA/mana_ib: fix a bug in calculating the wrong page table index when 64kb page table is used

by longli＠linuxonhyperv.com

From: Long Li <longli(a)microsoft.com> MANA hardware uses 4k page size. When calculating the page table index, it should use the hardware page size, not the system page size. Cc: stable(a)vger.kernel.org Fixes: 0266a177631d ("RDMA/mana_ib: Add a driver for Microsoft Azure Network Adapter") Signed-off-by: Long Li <longli(a)microsoft.com> --- drivers/infiniband/hw/mana/main.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/infiniband/hw/mana/main.c b/drivers/infiniband/hw/mana/main.c index d13abc954d2a..f68f54aea820 100644 --- a/drivers/infiniband/hw/mana/main.c +++ b/drivers/infiniband/hw/mana/main.c @@ -383,7 +383,7 @@ static int mana_ib_gd_create_dma_region(struct mana_ib_dev *dev, struct ib_umem create_req->length = umem->length; create_req->offset_in_page = ib_umem_dma_offset(umem, page_sz); - create_req->gdma_page_type = order_base_2(page_sz) - PAGE_SHIFT; + create_req->gdma_page_type = order_base_2(page_sz) - MANA_PAGE_SHIFT; create_req->page_count = num_pages_total; ibdev_dbg(&dev->ib_dev, "size_dma_region %lu num_pages_total %lu\n", -- 2.17.1

8 months, 3 weeks

4
5
0 0

[PATCH v2] drm/vmwgfx: Cleanup kms setup without 3d

by Zack Rusin

Do not validate format equality for the non 3d cases to allow xrgb to argb copies and make sure the dx binding flags are only used on dx compatible surfaces. Fixes basic 2d kms setup on configurations without 3d. There's little practical benefit to it because kms framebuffer coherence is disabled on configurations without 3d but with those changes the code actually makes sense. v2: Remove the now unused format variable Signed-off-by: Zack Rusin <zack.rusin(a)broadcom.com> Fixes: d6667f0ddf46 ("drm/vmwgfx: Fix handling of dumb buffers") Cc: Broadcom internal kernel review list <bcm-kernel-feedback-list(a)broadcom.com> Cc: dri-devel(a)lists.freedesktop.org Cc: <stable(a)vger.kernel.org> # v6.9+ Cc: Maaz Mombasawala <maaz.mombasawala(a)broadcom.com> Cc: Martin Krastev <martin.krastev(a)broadcom.com> --- drivers/gpu/drm/vmwgfx/vmwgfx_kms.c | 29 ------------------------- drivers/gpu/drm/vmwgfx/vmwgfx_surface.c | 9 +++++--- 2 files changed, 6 insertions(+), 32 deletions(-) diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c b/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c index 288ed0bb75cb..282b6153bcdd 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c @@ -1283,7 +1283,6 @@ static int vmw_kms_new_framebuffer_surface(struct vmw_private *dev_priv, { struct drm_device *dev = &dev_priv->drm; struct vmw_framebuffer_surface *vfbs; - enum SVGA3dSurfaceFormat format; struct vmw_surface *surface; int ret; @@ -1320,34 +1319,6 @@ static int vmw_kms_new_framebuffer_surface(struct vmw_private *dev_priv, return -EINVAL; } - switch (mode_cmd->pixel_format) { - case DRM_FORMAT_ARGB8888: - format = SVGA3D_A8R8G8B8; - break; - case DRM_FORMAT_XRGB8888: - format = SVGA3D_X8R8G8B8; - break; - case DRM_FORMAT_RGB565: - format = SVGA3D_R5G6B5; - break; - case DRM_FORMAT_XRGB1555: - format = SVGA3D_A1R5G5B5; - break; - default: - DRM_ERROR("Invalid pixel format: %p4cc\n", - &mode_cmd->pixel_format); - return -EINVAL; - } - - /* - * For DX, surface format validation is done when surface->scanout - * is set. - */ - if (!has_sm4_context(dev_priv) && format != surface->metadata.format) { - DRM_ERROR("Invalid surface format for requested mode.\n"); - return -EINVAL; - } - vfbs = kzalloc(sizeof(*vfbs), GFP_KERNEL); if (!vfbs) { ret = -ENOMEM; diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_surface.c b/drivers/gpu/drm/vmwgfx/vmwgfx_surface.c index 1625b30d9970..5721c74da3e0 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_surface.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_surface.c @@ -2276,9 +2276,12 @@ int vmw_dumb_create(struct drm_file *file_priv, const struct SVGA3dSurfaceDesc *desc = vmw_surface_get_desc(format); SVGA3dSurfaceAllFlags flags = SVGA3D_SURFACE_HINT_TEXTURE | SVGA3D_SURFACE_HINT_RENDERTARGET | - SVGA3D_SURFACE_SCREENTARGET | - SVGA3D_SURFACE_BIND_SHADER_RESOURCE | - SVGA3D_SURFACE_BIND_RENDER_TARGET; + SVGA3D_SURFACE_SCREENTARGET; + + if (vmw_surface_is_dx_screen_target_format(format)) { + flags |= SVGA3D_SURFACE_BIND_SHADER_RESOURCE | + SVGA3D_SURFACE_BIND_RENDER_TARGET; + } /* * Without mob support we're just going to use raw memory buffer -- 2.43.0

8 months, 3 weeks

3
2
0 0

[PATCH v5] x86/entry_32: Use stack segment selector for VERW operand

by Pawan Gupta

Robert Gill reported below #GP when dosemu software was executing vm86() system call: general protection fault: 0000 [#1] PREEMPT SMP CPU: 4 PID: 4610 Comm: dosemu.bin Not tainted 6.6.21-gentoo-x86 #1 Hardware name: Dell Inc. PowerEdge 1950/0H723K, BIOS 2.7.0 10/30/2010 EIP: restore_all_switch_stack+0xbe/0xcf EAX: 00000000 EBX: 00000000 ECX: 00000000 EDX: 00000000 ESI: 00000000 EDI: 00000000 EBP: 00000000 ESP: ff8affdc DS: 0000 ES: 0000 FS: 0000 GS: 0033 SS: 0068 EFLAGS: 00010046 CR0: 80050033 CR2: 00c2101c CR3: 04b6d000 CR4: 000406d0 Call Trace: show_regs+0x70/0x78 die_addr+0x29/0x70 exc_general_protection+0x13c/0x348 exc_bounds+0x98/0x98 handle_exception+0x14d/0x14d exc_bounds+0x98/0x98 restore_all_switch_stack+0xbe/0xcf exc_bounds+0x98/0x98 restore_all_switch_stack+0xbe/0xcf This only happens when VERW based mitigations like MDS/RFDS are enabled. This is because segment registers with an arbitrary user value can result in #GP when executing VERW. Intel SDM vol. 2C documents the following behavior for VERW instruction: #GP(0) - If a memory operand effective address is outside the CS, DS, ES, FS, or GS segment limit. CLEAR_CPU_BUFFERS macro executes VERW instruction before returning to user space. Replace CLEAR_CPU_BUFFERS with a safer version that uses %ss to refer VERW operand mds_verw_sel. This ensures VERW will not #GP for an arbitrary user %ds. Also, in NMI return path, move VERW to after RESTORE_ALL_NMI that touches GPRs. For clarity, below are the locations where the new CLEAR_CPU_BUFFERS_SAFE version is being used: * entry_INT80_32(), entry_SYSENTER_32() and interrupts (via handle_exception_return) do: restore_all_switch_stack: [...] mov %esi,%esi verw %ss:0xc0fc92c0 <------------- iret * Opportunistic SYSEXIT: [...] verw %ss:0xc0fc92c0 <------------- btrl $0x9,(%esp) popf pop %eax sti sysexit * nmi_return and nmi_from_espfix: mov %esi,%esi verw %ss:0xc0fc92c0 <------------- jmp .Lirq_return Fixes: a0e2dab44d22 ("x86/entry_32: Add VERW just before userspace transition") Cc: stable(a)vger.kernel.org # 5.10+ Reported-by: Robert Gill <rtgill82(a)gmail.com> Closes: https://bugzilla.kernel.org/show_bug.cgi?id=218707 Closes: https://lore.kernel.org/all/8c77ccfd-d561-45a1-8ed5-6b75212c7a58@leemhuis.i… Suggested-by: Dave Hansen <dave.hansen(a)linux.intel.com> Suggested-by: Brian Gerst <brgerst(a)gmail.com> # Use %ss Signed-off-by: Pawan Gupta <pawan.kumar.gupta(a)linux.intel.com> --- v5: - Simplify the use of ALTERNATIVE construct (Uros/Jiri/Peter). v4: https://lore.kernel.org/r/20240710-fix-dosemu-vm86-v4-1-aa6464e1de6f@linux.… - Further simplify the patch by using %ss for all VERW calls in 32-bit mode (Brian). - In NMI exit path move VERW after RESTORE_ALL_NMI that touches GPRs (Dave). v3: https://lore.kernel.org/r/20240701-fix-dosemu-vm86-v3-1-b1969532c75a@linux.… - Simplify CLEAR_CPU_BUFFERS_SAFE by using %ss instead of %ds (Brian). - Do verw before popf in SYSEXIT path (Jari). v2: https://lore.kernel.org/r/20240627-fix-dosemu-vm86-v2-1-d5579f698e77@linux.… - Safe guard against any other system calls like vm86() that might change %ds (Dave). v1: https://lore.kernel.org/r/20240426-fix-dosemu-vm86-v1-1-88c826a3f378@linux.… --- --- arch/x86/entry/entry_32.S | 16 +++++++++++++--- 1 file changed, 13 insertions(+), 3 deletions(-) diff --git a/arch/x86/entry/entry_32.S b/arch/x86/entry/entry_32.S index d3a814efbff6..25c942149fb5 100644 --- a/arch/x86/entry/entry_32.S +++ b/arch/x86/entry/entry_32.S @@ -253,6 +253,14 @@ .Lend_\@: .endm +/* + * Safer version of CLEAR_CPU_BUFFERS that uses %ss to reference VERW operand + * mds_verw_sel. This ensures VERW will not #GP for an arbitrary user %ds. + */ +.macro CLEAR_CPU_BUFFERS_SAFE + ALTERNATIVE "", __stringify(verw %ss:_ASM_RIP(mds_verw_sel)), X86_FEATURE_CLEAR_CPU_BUF +.endm + .macro RESTORE_INT_REGS popl %ebx popl %ecx @@ -871,6 +879,8 @@ SYM_FUNC_START(entry_SYSENTER_32) /* Now ready to switch the cr3 */ SWITCH_TO_USER_CR3 scratch_reg=%eax + /* Clobbers ZF */ + CLEAR_CPU_BUFFERS_SAFE /* * Restore all flags except IF. (We restore IF separately because @@ -881,7 +891,6 @@ SYM_FUNC_START(entry_SYSENTER_32) BUG_IF_WRONG_CR3 no_user_check=1 popfl popl %eax - CLEAR_CPU_BUFFERS /* * Return back to the vDSO, which will pop ecx and edx. @@ -951,7 +960,7 @@ restore_all_switch_stack: /* Restore user state */ RESTORE_REGS pop=4 # skip orig_eax/error_code - CLEAR_CPU_BUFFERS + CLEAR_CPU_BUFFERS_SAFE .Lirq_return: /* * ARCH_HAS_MEMBARRIER_SYNC_CORE rely on IRET core serialization @@ -1144,7 +1153,6 @@ SYM_CODE_START(asm_exc_nmi) /* Not on SYSENTER stack. */ call exc_nmi - CLEAR_CPU_BUFFERS jmp .Lnmi_return .Lnmi_from_sysenter_stack: @@ -1165,6 +1173,7 @@ SYM_CODE_START(asm_exc_nmi) CHECK_AND_APPLY_ESPFIX RESTORE_ALL_NMI cr3_reg=%edi pop=4 + CLEAR_CPU_BUFFERS_SAFE jmp .Lirq_return #ifdef CONFIG_X86_ESPFIX32 @@ -1206,6 +1215,7 @@ SYM_CODE_START(asm_exc_nmi) * 1 - orig_ax */ lss (1+5+6)*4(%esp), %esp # back to espfix stack + CLEAR_CPU_BUFFERS_SAFE jmp .Lirq_return #endif SYM_CODE_END(asm_exc_nmi) --- base-commit: f2661062f16b2de5d7b6a5c42a9a5c96326b8454 change-id: 20240426-fix-dosemu-vm86-dd111a01737e

8 months, 3 weeks

2
2
0 0

[PATCH v2 0/5] soc: qcom: fix rpm_requests module probing

by Dmitry Baryshkov

The GLINK RPMSG channels get modalias based on the compatible string rather than the channel type, however the smd-rpm module uses rpmsg ID instead. Thus if the smd-rpm is built as a module, it doesn't get automatically loaded. Add generic compatible to such devices and fix module's ID table. Module loading worked before the commit bcabe1e09135 ("soc: qcom: smd-rpm: Match rpmsg channel instead of compatible"), because the driver listed all compatible strings, but the mentioned commit changed ID table. Revert the offending commit and add generic compatible strings instead. Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov(a)linaro.org> --- Changes in v2: - Separate fix from the improvements (Krzysztof - Split the qcom,glink-smd-rpm compat from the qcom,smd-rpm as they use different channels underneath. - Link to v1: https://lore.kernel.org/r/20240729-fix-smd-rpm-v1-0-99a96133cc65@linaro.org --- Dmitry Baryshkov (5): Revert "soc: qcom: smd-rpm: Match rpmsg channel instead of compatible" dt-bindings: soc: qcom: smd-rpm: add generic compatibles soc: qcom: smd-rpm: add qcom,smd-rpm compatible ARM: dts: qcom: add generic compat string to RPM glink channels arm64: dts: qcom: add generic compat string to RPM glink channels .../devicetree/bindings/clock/qcom,rpmcc.yaml | 2 +- .../bindings/remoteproc/qcom,glink-rpm-edge.yaml | 2 +- .../bindings/remoteproc/qcom,rpm-proc.yaml | 4 +- .../devicetree/bindings/soc/qcom/qcom,smd-rpm.yaml | 74 ++++++++++------------ .../devicetree/bindings/soc/qcom/qcom,smd.yaml | 2 +- arch/arm/boot/dts/qcom/qcom-apq8084.dtsi | 2 +- arch/arm/boot/dts/qcom/qcom-msm8226.dtsi | 2 +- arch/arm/boot/dts/qcom/qcom-msm8974.dtsi | 2 +- arch/arm64/boot/dts/qcom/ipq6018.dtsi | 2 +- arch/arm64/boot/dts/qcom/ipq9574.dtsi | 2 +- arch/arm64/boot/dts/qcom/msm8916.dtsi | 2 +- arch/arm64/boot/dts/qcom/msm8939.dtsi | 2 +- arch/arm64/boot/dts/qcom/msm8953.dtsi | 2 +- arch/arm64/boot/dts/qcom/msm8976.dtsi | 2 +- arch/arm64/boot/dts/qcom/msm8994.dtsi | 2 +- arch/arm64/boot/dts/qcom/msm8996.dtsi | 2 +- arch/arm64/boot/dts/qcom/msm8998.dtsi | 2 +- arch/arm64/boot/dts/qcom/qcm2290.dtsi | 2 +- arch/arm64/boot/dts/qcom/qcs404.dtsi | 2 +- arch/arm64/boot/dts/qcom/sdm630.dtsi | 2 +- arch/arm64/boot/dts/qcom/sm6115.dtsi | 2 +- arch/arm64/boot/dts/qcom/sm6125.dtsi | 2 +- arch/arm64/boot/dts/qcom/sm6375.dtsi | 2 +- drivers/soc/qcom/smd-rpm.c | 41 +++++++++--- 24 files changed, 88 insertions(+), 73 deletions(-) --- base-commit: 668d33c9ff922c4590c58754ab064aaf53c387dd change-id: 20240729-fix-smd-rpm-9651e87a9bb0 Best regards, -- Dmitry Baryshkov <dmitry.baryshkov(a)linaro.org>

8 months, 3 weeks

3
3
0 0

[PATCH net 12/13] can: mcp251x: fix deadlock if an interrupt occurs during mcp251x_open

by Marc Kleine-Budde

From: Simon Arlott <simon(a)octiron.net> The mcp251x_hw_wake() function is called with the mpc_lock mutex held and disables the interrupt handler so that no interrupts can be processed while waking the device. If an interrupt has already occurred then waiting for the interrupt handler to complete will deadlock because it will be trying to acquire the same mutex. CPU0 CPU1 ---- ---- mcp251x_open() mutex_lock(&priv->mcp_lock) request_threaded_irq() <interrupt> mcp251x_can_ist() mutex_lock(&priv->mcp_lock) mcp251x_hw_wake() disable_irq() <-- deadlock Use disable_irq_nosync() instead because the interrupt handler does everything while holding the mutex so it doesn't matter if it's still running. Fixes: 8ce8c0abcba3 ("can: mcp251x: only reset hardware as required") Signed-off-by: Simon Arlott <simon(a)octiron.net> Reviewed-by: Przemek Kitszel <przemyslaw.kitszel(a)intel.com> Cc: stable(a)vger.kernel.org Link: https://lore.kernel.org/all/4fc08687-1d80-43fe-9f0d-8ef8475e75f6@0882a8b5-c… Signed-off-by: Marc Kleine-Budde <mkl(a)pengutronix.de> --- drivers/net/can/spi/mcp251x.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/net/can/spi/mcp251x.c b/drivers/net/can/spi/mcp251x.c index 3b8736ff0345..ec5c64006a16 100644 --- a/drivers/net/can/spi/mcp251x.c +++ b/drivers/net/can/spi/mcp251x.c @@ -752,7 +752,7 @@ static int mcp251x_hw_wake(struct spi_device *spi) int ret; /* Force wakeup interrupt to wake device, but don't execute IST */ - disable_irq(spi->irq); + disable_irq_nosync(spi->irq); mcp251x_write_2regs(spi, CANINTE, CANINTE_WAKIE, CANINTF_WAKIF); /* Wait for oscillator startup timer after wake up */ -- 2.45.2

8 months, 3 weeks

1
0
0 0

[PATCH 4.19 0/1] Fix CVE-2021-3493

by hsimeliere.opensource＠witekio.com

https://nvd.nist.gov/vuln/detail/CVE-2021-3493

8 months, 3 weeks

2
5
0 0

FAILED: patch "[PATCH] smb/client: fix rdma usage in smb2_async_writev()" failed to apply to 6.6-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 6.6-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. To reproduce the conflict and resubmit, you may use the following commands: git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-6.6.y git checkout FETCH_HEAD git cherry-pick -x 017d1701743657fbfaea74397727a9d2b81846b7 # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2024082910-depraved-thing-8c02@gregkh' --subject-prefix 'PATCH 6.6.y' HEAD^.. Possible dependencies: 017d17017436 ("smb/client: fix rdma usage in smb2_async_writev()") b608e2c31878 ("smb/client: remove unused rq_iter_size from struct smb_rqst") dc5939de82f1 ("cifs: Replace the writedata replay bool with a netfs sreq flag") ab58fbdeebc7 ("cifs: Use more fields from netfs_io_subrequest") a975a2f22cdc ("cifs: Replace cifs_writedata with a wrapper around netfs_io_subrequest") 753b67eb630d ("cifs: Replace cifs_readdata with a wrapper around netfs_io_subrequest") f3dc1bdb6b0b ("cifs: Fix writeback data corruption") d1bba17e20d5 ("Merge tag '6.8-rc1-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6") thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 017d1701743657fbfaea74397727a9d2b81846b7 Mon Sep 17 00:00:00 2001 From: Stefan Metzmacher <metze(a)samba.org> Date: Wed, 21 Aug 2024 16:31:39 +0200 Subject: [PATCH] smb/client: fix rdma usage in smb2_async_writev() rqst.rq_iter needs to be truncated otherwise we'll also send the bytes into the stream socket... This is the logic behind rqst.rq_npages = 0, which was removed in "cifs: Change the I/O paths to use an iterator rather than a page list" (d08089f649a0cfb2099c8551ac47eef0cc23fdf2). Cc: stable(a)vger.kernel.org Fixes: d08089f649a0 ("cifs: Change the I/O paths to use an iterator rather than a page list") Reviewed-by: David Howells <dhowells(a)redhat.com> Signed-off-by: Stefan Metzmacher <metze(a)samba.org> Signed-off-by: Steve French <stfrench(a)microsoft.com> diff --git a/fs/smb/client/smb2pdu.c b/fs/smb/client/smb2pdu.c index 63a2541d4a05..2d7e6c42cf18 100644 --- a/fs/smb/client/smb2pdu.c +++ b/fs/smb/client/smb2pdu.c @@ -4913,6 +4913,13 @@ smb2_async_writev(struct cifs_io_subrequest *wdata) if (rc) goto out; + rqst.rq_iov = iov; + rqst.rq_iter = wdata->subreq.io_iter; + + rqst.rq_iov[0].iov_len = total_len - 1; + rqst.rq_iov[0].iov_base = (char *)req; + rqst.rq_nvec += 1; + if (smb3_encryption_required(tcon)) flags |= CIFS_TRANSFORM_REQ; @@ -4924,6 +4931,7 @@ smb2_async_writev(struct cifs_io_subrequest *wdata) req->WriteChannelInfoOffset = 0; req->WriteChannelInfoLength = 0; req->Channel = SMB2_CHANNEL_NONE; + req->Length = cpu_to_le32(io_parms->length); req->Offset = cpu_to_le64(io_parms->offset); req->DataOffset = cpu_to_le16( offsetof(struct smb2_write_req, Buffer)); @@ -4943,7 +4951,6 @@ smb2_async_writev(struct cifs_io_subrequest *wdata) */ if (smb3_use_rdma_offload(io_parms)) { struct smbd_buffer_descriptor_v1 *v1; - size_t data_size = iov_iter_count(&wdata->subreq.io_iter); bool need_invalidate = server->dialect == SMB30_PROT_ID; wdata->mr = smbd_register_mr(server->smbd_conn, &wdata->subreq.io_iter, @@ -4952,9 +4959,10 @@ smb2_async_writev(struct cifs_io_subrequest *wdata) rc = -EAGAIN; goto async_writev_out; } + /* For RDMA read, I/O size is in RemainingBytes not in Length */ + req->RemainingBytes = req->Length; req->Length = 0; req->DataOffset = 0; - req->RemainingBytes = cpu_to_le32(data_size); req->Channel = SMB2_CHANNEL_RDMA_V1_INVALIDATE; if (need_invalidate) req->Channel = SMB2_CHANNEL_RDMA_V1; @@ -4966,30 +4974,22 @@ smb2_async_writev(struct cifs_io_subrequest *wdata) v1->offset = cpu_to_le64(wdata->mr->mr->iova); v1->token = cpu_to_le32(wdata->mr->mr->rkey); v1->length = cpu_to_le32(wdata->mr->mr->length); + + rqst.rq_iov[0].iov_len += sizeof(*v1); + + /* + * We keep wdata->subreq.io_iter, + * but we have to truncate rqst.rq_iter + */ + iov_iter_truncate(&rqst.rq_iter, 0); } #endif - iov[0].iov_len = total_len - 1; - iov[0].iov_base = (char *)req; - rqst.rq_iov = iov; - rqst.rq_nvec = 1; - rqst.rq_iter = wdata->subreq.io_iter; if (test_bit(NETFS_SREQ_RETRYING, &wdata->subreq.flags)) smb2_set_replay(server, &rqst); -#ifdef CONFIG_CIFS_SMB_DIRECT - if (wdata->mr) - iov[0].iov_len += sizeof(struct smbd_buffer_descriptor_v1); -#endif - cifs_dbg(FYI, "async write at %llu %u bytes iter=%zx\n", - io_parms->offset, io_parms->length, iov_iter_count(&rqst.rq_iter)); -#ifdef CONFIG_CIFS_SMB_DIRECT - /* For RDMA read, I/O size is in RemainingBytes not in Length */ - if (!wdata->mr) - req->Length = cpu_to_le32(io_parms->length); -#else - req->Length = cpu_to_le32(io_parms->length); -#endif + cifs_dbg(FYI, "async write at %llu %u bytes iter=%zx\n", + io_parms->offset, io_parms->length, iov_iter_count(&wdata->subreq.io_iter)); if (wdata->credits.value > 0) { shdr->CreditCharge = cpu_to_le16(DIV_ROUND_UP(wdata->subreq.len,

8 months, 3 weeks

1
0
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror August 2024