- Linux-stable-mirror - lists.linaro.org

[PATCH] arm64: dts: marvell: cn913x-solidrun: fix sata ports status

by Josua Mayer

Commit "arm64: dts: marvell: only enable complete sata nodes" changed armada-cp11x.dtsi disabling all sata ports status by default. The author missed some dts which relied on the dtsi enabling all ports, and just disabled unused ones instead. Update dts for SolidRun cn913x based boards to enable the available ports, rather than disabling the unvavailable one. Further according to dt bindings the serdes phys are to be specified in the port node, not the controller node. Move those phys properties accordingly in clearfog base/pro/solidwan. Fixes: 30023876aef4 ("arm64: dts: marvell: only enable complete sata nodes") Cc: stable(a)vger.kernel.org Signed-off-by: Josua Mayer <josua(a)solid-run.com> --- arch/arm64/boot/dts/marvell/cn9130-cf.dtsi | 7 ++++--- arch/arm64/boot/dts/marvell/cn9131-cf-solidwan.dts | 6 ++++-- arch/arm64/boot/dts/marvell/cn9132-clearfog.dts | 6 ++---- 3 files changed, 10 insertions(+), 9 deletions(-) diff --git a/arch/arm64/boot/dts/marvell/cn9130-cf.dtsi b/arch/arm64/boot/dts/marvell/cn9130-cf.dtsi index ad0ab34b66028c53b8a18b3e8ee0c0aec869759f..bd42bfbe408bbe2a4d58dbd40204bcfb3c126312 100644 --- a/arch/arm64/boot/dts/marvell/cn9130-cf.dtsi +++ b/arch/arm64/boot/dts/marvell/cn9130-cf.dtsi @@ -152,11 +152,12 @@ expander0_pins: cp0-expander0-pins { /* SRDS #0 - SATA on M.2 connector */ &cp0_sata0 { - phys = <&cp0_comphy0 1>; status = "okay"; - /* only port 1 is available */ - /delete-node/ sata-port@0; + sata-port@1 { + phys = <&cp0_comphy0 1>; + status = "okay"; + }; }; /* microSD */ diff --git a/arch/arm64/boot/dts/marvell/cn9131-cf-solidwan.dts b/arch/arm64/boot/dts/marvell/cn9131-cf-solidwan.dts index 47234d0858dd2195bb1485f25768ad3c757b7ac2..338853d3b179bb5cb742e975bb830fdb9d62d4cc 100644 --- a/arch/arm64/boot/dts/marvell/cn9131-cf-solidwan.dts +++ b/arch/arm64/boot/dts/marvell/cn9131-cf-solidwan.dts @@ -563,11 +563,13 @@ &cp1_rtc { /* SRDS #1 - SATA on M.2 (J44) */ &cp1_sata0 { - phys = <&cp1_comphy1 0>; status = "okay"; /* only port 0 is available */ - /delete-node/ sata-port@1; + sata-port@0 { + phys = <&cp1_comphy1 0>; + status = "okay"; + }; }; &cp1_syscon0 { diff --git a/arch/arm64/boot/dts/marvell/cn9132-clearfog.dts b/arch/arm64/boot/dts/marvell/cn9132-clearfog.dts index 0f53745a6fa0d8cbd3ab9cdc28a972ed748c275f..115c55d73786e2b9265e1caa4c62ee26f498fb41 100644 --- a/arch/arm64/boot/dts/marvell/cn9132-clearfog.dts +++ b/arch/arm64/boot/dts/marvell/cn9132-clearfog.dts @@ -512,10 +512,9 @@ &cp1_sata0 { status = "okay"; /* only port 1 is available */ - /delete-node/ sata-port@0; - sata-port@1 { phys = <&cp1_comphy3 1>; + status = "okay"; }; }; @@ -631,9 +630,8 @@ &cp2_sata0 { status = "okay"; /* only port 1 is available */ - /delete-node/ sata-port@0; - sata-port@1 { + status = "okay"; phys = <&cp2_comphy3 1>; }; }; --- base-commit: 8f5ae30d69d7543eee0d70083daf4de8fe15d585 change-id: 20250911-cn913x-sr-fix-sata-5c737ebdb97f Best regards, -- Josua Mayer <josua(a)solid-run.com>

1 month, 2 weeks

2
2
0 0

[PATCH v2 6.12 1/3] drm/managed: Add DRM-managed alloc_ordered_workqueue

by Mikhail Dmitrichenko

From: Louis Chauvet <louis.chauvet(a)bootlin.com> commit c367b772e6d89d8c7b560c7df7e3803ce6b8bcea upstream. Add drmm_alloc_ordered_workqueue(), a helper that provides managed ordered workqueue cleanup. The workqueue will be destroyed with the final reference of the DRM device. Reviewed-by: Thomas Zimmermann <tzimmermann(a)suse.de> Reviewed-by: Maxime Ripard <mripard(a)kernel.org> Link: https://patchwork.freedesktop.org/patch/msgid/20250116-google-vkms-managed-… Signed-off-by: Louis Chauvet <louis.chauvet(a)bootlin.com> Signed-off-by: Mikhail Dmitrichenko <mdmitrichenko(a)astralinux.ru> --- drivers/gpu/drm/drm_managed.c | 8 ++++++++ include/drm/drm_managed.h | 12 ++++++++++++ 2 files changed, 20 insertions(+) diff --git a/drivers/gpu/drm/drm_managed.c b/drivers/gpu/drm/drm_managed.c index 79ce86a5bd67..cc4c463daae7 100644 --- a/drivers/gpu/drm/drm_managed.c +++ b/drivers/gpu/drm/drm_managed.c @@ -310,3 +310,11 @@ void __drmm_mutex_release(struct drm_device *dev, void *res) mutex_destroy(lock); } EXPORT_SYMBOL(__drmm_mutex_release); + +void __drmm_workqueue_release(struct drm_device *device, void *res) +{ + struct workqueue_struct *wq = res; + + destroy_workqueue(wq); +} +EXPORT_SYMBOL(__drmm_workqueue_release); diff --git a/include/drm/drm_managed.h b/include/drm/drm_managed.h index f547b09ca023..53017cc609ac 100644 --- a/include/drm/drm_managed.h +++ b/include/drm/drm_managed.h @@ -127,4 +127,16 @@ void __drmm_mutex_release(struct drm_device *dev, void *res); drmm_add_action_or_reset(dev, __drmm_mutex_release, lock); \ }) \ +void __drmm_workqueue_release(struct drm_device *device, void *wq); + +#define drmm_alloc_ordered_workqueue(dev, fmt, flags, args...) \ + ({ \ + struct workqueue_struct *wq = alloc_ordered_workqueue(fmt, flags, ##args); \ + wq ? ({ \ + int ret = drmm_add_action_or_reset(dev, __drmm_workqueue_release, wq); \ + ret ? ERR_PTR(ret) : wq; \ + }) : \ + wq; \ + }) + #endif -- 2.39.2

1 month, 2 weeks

1
2
0 0

[PATCH v3] bus: mhi: ep: Fix chained transfer handling in read path

by Sumit Kumar

The mhi_ep_read_channel function incorrectly assumes the End of Transfer (EOT) bit is received with the doorbell in chained transactions, causing it to advance mhi_chan->rd_offset beyond wr_offset during host-to-device transfers when EOT has not yet arrived, leading to access of unmapped host memory that causes IOMMU faults and processing of stale TREs. Modify the loop condition to ensure mhi_queue is not empty, allowing the function to process only valid TREs up to the current write pointer to prevent premature reads and ensure safe traversal of chained TREs. Remove buf_left from the while loop condition to avoid exiting prematurely before reading the ring completely, and remove write_offset since it will always be zero because the new cache buffer is allocated every time. Fixes: 5301258899773 ("bus: mhi: ep: Add support for reading from the host") Cc: stable(a)vger.kernel.org Co-developed-by: Akhil Vinod <akhil.vinod(a)oss.qualcomm.com> Signed-off-by: Akhil Vinod <akhil.vinod(a)oss.qualcomm.com> Signed-off-by: Sumit Kumar <sumit.kumar(a)oss.qualcomm.com> --- Changes in v3: - Update commit message - Migrated to new mail - Link to v2: https://lore.kernel.org/r/20250822-chained_transfer-v2-1-7aeb5ac215b6@quici… Changes in v2: - Use mhi_ep_queue_is_empty in while loop (Mani). - Remove do while loop in mhi_ep_process_ch_ring (Mani). - Remove buf_left, wr_offset, tr_done. - Haven't added Reviewed-by as there is change in logic. - Link to v1: https://lore.kernel.org/r/20250709-chained_transfer-v1-1-2326a4605c9c@quici… --- drivers/bus/mhi/ep/main.c | 37 ++++++++++++------------------------- 1 file changed, 12 insertions(+), 25 deletions(-) diff --git a/drivers/bus/mhi/ep/main.c b/drivers/bus/mhi/ep/main.c index b3eafcf2a2c50d95e3efd3afb27038ecf55552a5..cdea24e9291959ae0a92487c1b9698dc8164d2f1 100644 --- a/drivers/bus/mhi/ep/main.c +++ b/drivers/bus/mhi/ep/main.c @@ -403,17 +403,13 @@ static int mhi_ep_read_channel(struct mhi_ep_cntrl *mhi_cntrl, { struct mhi_ep_chan *mhi_chan = &mhi_cntrl->mhi_chan[ring->ch_id]; struct device *dev = &mhi_cntrl->mhi_dev->dev; - size_t tr_len, read_offset, write_offset; + size_t tr_len, read_offset; struct mhi_ep_buf_info buf_info = {}; u32 len = MHI_EP_DEFAULT_MTU; struct mhi_ring_element *el; - bool tr_done = false; void *buf_addr; - u32 buf_left; int ret; - buf_left = len; - do { /* Don't process the transfer ring if the channel is not in RUNNING state */ if (mhi_chan->state != MHI_CH_STATE_RUNNING) { @@ -426,24 +422,23 @@ static int mhi_ep_read_channel(struct mhi_ep_cntrl *mhi_cntrl, /* Check if there is data pending to be read from previous read operation */ if (mhi_chan->tre_bytes_left) { dev_dbg(dev, "TRE bytes remaining: %u\n", mhi_chan->tre_bytes_left); - tr_len = min(buf_left, mhi_chan->tre_bytes_left); + tr_len = min(len, mhi_chan->tre_bytes_left); } else { mhi_chan->tre_loc = MHI_TRE_DATA_GET_PTR(el); mhi_chan->tre_size = MHI_TRE_DATA_GET_LEN(el); mhi_chan->tre_bytes_left = mhi_chan->tre_size; - tr_len = min(buf_left, mhi_chan->tre_size); + tr_len = min(len, mhi_chan->tre_size); } read_offset = mhi_chan->tre_size - mhi_chan->tre_bytes_left; - write_offset = len - buf_left; buf_addr = kmem_cache_zalloc(mhi_cntrl->tre_buf_cache, GFP_KERNEL); if (!buf_addr) return -ENOMEM; buf_info.host_addr = mhi_chan->tre_loc + read_offset; - buf_info.dev_addr = buf_addr + write_offset; + buf_info.dev_addr = buf_addr; buf_info.size = tr_len; buf_info.cb = mhi_ep_read_completion; buf_info.cb_buf = buf_addr; @@ -459,16 +454,12 @@ static int mhi_ep_read_channel(struct mhi_ep_cntrl *mhi_cntrl, goto err_free_buf_addr; } - buf_left -= tr_len; mhi_chan->tre_bytes_left -= tr_len; - if (!mhi_chan->tre_bytes_left) { - if (MHI_TRE_DATA_GET_IEOT(el)) - tr_done = true; - + if (!mhi_chan->tre_bytes_left) mhi_chan->rd_offset = (mhi_chan->rd_offset + 1) % ring->ring_size; - } - } while (buf_left && !tr_done); + /* Read until the some buffer is left or the ring becomes not empty */ + } while (!mhi_ep_queue_is_empty(mhi_chan->mhi_dev, DMA_TO_DEVICE)); return 0; @@ -502,15 +493,11 @@ static int mhi_ep_process_ch_ring(struct mhi_ep_ring *ring) mhi_chan->xfer_cb(mhi_chan->mhi_dev, &result); } else { /* UL channel */ - do { - ret = mhi_ep_read_channel(mhi_cntrl, ring); - if (ret < 0) { - dev_err(&mhi_chan->mhi_dev->dev, "Failed to read channel\n"); - return ret; - } - - /* Read until the ring becomes empty */ - } while (!mhi_ep_queue_is_empty(mhi_chan->mhi_dev, DMA_TO_DEVICE)); + ret = mhi_ep_read_channel(mhi_cntrl, ring); + if (ret < 0) { + dev_err(&mhi_chan->mhi_dev->dev, "Failed to read channel\n"); + return ret; + } } return 0; --- base-commit: 4c06e63b92038fadb566b652ec3ec04e228931e8 change-id: 20250910-final_chained-750c213725f1 Best regards, -- Sumit Kumar <sumit.kumar(a)oss.qualcomm.com>

1 month, 2 weeks

2
1
0 0

[PATCH 1/2] staging: axis-fifo: fix TX handling on copy_from_user() failure

by Ovidiu Panait

If copy_from_user() fails, write() currently returns -EFAULT, but any partially written data leaves the TX FIFO in an inconsistent state. Subsequent write() calls then fail with "transmit length mismatch" errors. Once partial data is written to the hardware FIFO, it cannot be removed without a TX reset. Commit c6e8d85fafa7 ("staging: axis-fifo: Remove hardware resets for user errors") removed a full FIFO reset for this case, which fixed a potential RX data loss, but introduced this TX issue. Fix this by introducing a bounce buffer: copy the full packet from userspace first, and write to the hardware FIFO only if the copy was successful. Fixes: c6e8d85fafa7 ("staging: axis-fifo: Remove hardware resets for user errors") Cc: stable(a)vger.kernel.org Signed-off-by: Ovidiu Panait <ovidiu.panait.oss(a)gmail.com> --- drivers/staging/axis-fifo/axis-fifo.c | 36 ++++++++------------------- 1 file changed, 10 insertions(+), 26 deletions(-) diff --git a/drivers/staging/axis-fifo/axis-fifo.c b/drivers/staging/axis-fifo/axis-fifo.c index f54614ba1aa8..c47c6a022402 100644 --- a/drivers/staging/axis-fifo/axis-fifo.c +++ b/drivers/staging/axis-fifo/axis-fifo.c @@ -43,7 +43,6 @@ #define DRIVER_NAME "axis_fifo" #define READ_BUF_SIZE 128U /* read buffer length in words */ -#define WRITE_BUF_SIZE 128U /* write buffer length in words */ #define AXIS_FIFO_DEBUG_REG_NAME_MAX_LEN 4 @@ -305,11 +304,8 @@ static ssize_t axis_fifo_write(struct file *f, const char __user *buf, { struct axis_fifo *fifo = (struct axis_fifo *)f->private_data; unsigned int words_to_write; - unsigned int copied; - unsigned int copy; - unsigned int i; + u32 *txbuf; int ret; - u32 tmp_buf[WRITE_BUF_SIZE]; if (len % sizeof(u32)) { dev_err(fifo->dt_device, @@ -374,32 +370,20 @@ static ssize_t axis_fifo_write(struct file *f, const char __user *buf, } } - /* write data from an intermediate buffer into the fifo IP, refilling - * the buffer with userspace data as needed - */ - copied = 0; - while (words_to_write > 0) { - copy = min(words_to_write, WRITE_BUF_SIZE); - - if (copy_from_user(tmp_buf, buf + copied * sizeof(u32), - copy * sizeof(u32))) { - ret = -EFAULT; - goto end_unlock; - } - - for (i = 0; i < copy; i++) - iowrite32(tmp_buf[i], fifo->base_addr + - XLLF_TDFD_OFFSET); - - copied += copy; - words_to_write -= copy; + txbuf = vmemdup_user(buf, len); + if (IS_ERR(txbuf)) { + ret = PTR_ERR(txbuf); + goto end_unlock; } - ret = copied * sizeof(u32); + for (int i = 0; i < words_to_write; ++i) + iowrite32(txbuf[i], fifo->base_addr + XLLF_TDFD_OFFSET); /* write packet size to fifo */ - iowrite32(ret, fifo->base_addr + XLLF_TLR_OFFSET); + iowrite32(len, fifo->base_addr + XLLF_TLR_OFFSET); + ret = len; + kvfree(txbuf); end_unlock: mutex_unlock(&fifo->write_lock); -- 2.50.0

1 month, 2 weeks

1
1
0 0

[PATCH] drm/omap: Mark dispc_save_context() with noinline_for_stack

by Nathan Chancellor

A recent innocuous internal optimization change in LLVM [1] causes the same issue that necessitated commit 660942f2441d ("drm: omapdrm: reduce clang stack usage") to occur in dispc_runtime_suspend() from inlinling dispc_save_context(). drivers/gpu/drm/omapdrm/dss/dispc.c:4720:27: error: stack frame size (2272) exceeds limit (2048) in 'dispc_runtime_suspend' [-Werror,-Wframe-larger-than] 4720 | static __maybe_unused int dispc_runtime_suspend(struct device *dev) | ^ There is an unfortunate interaction between the inner loops of dispc_save_context() getting unrolled and the calculation of the index into the ctx array being spilled to the stack when sanitizers are enabled [2]. While this should obviously be addressed on the LLVM side, such a fix may not be easy to craft and it is simple enough to work around the issue in the same manner as before by marking dispc_save_context() with noinline_for_stack, which makes it use the same amount of stack as dispc_restore_context() does after the same change. Cc: stable(a)vger.kernel.org Link: https://github.com/llvm/llvm-project/commit/055bfc027141bbfafd51fb43f5ab81b… [1] Link: https://llvm.org/pr143908 [2] Signed-off-by: Nathan Chancellor <nathan(a)kernel.org> --- drivers/gpu/drm/omapdrm/dss/dispc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/omapdrm/dss/dispc.c b/drivers/gpu/drm/omapdrm/dss/dispc.c index cf055815077cffad554a4ae58cfd7b81edcbb0d4..d079f557c8f24d1afd0bc182edd13165cb9c356c 100644 --- a/drivers/gpu/drm/omapdrm/dss/dispc.c +++ b/drivers/gpu/drm/omapdrm/dss/dispc.c @@ -417,7 +417,7 @@ static bool dispc_has_feature(struct dispc_device *dispc, #define RR(dispc, reg) \ dispc_write_reg(dispc, DISPC_##reg, dispc->ctx[DISPC_##reg / sizeof(u32)]) -static void dispc_save_context(struct dispc_device *dispc) +static noinline_for_stack void dispc_save_context(struct dispc_device *dispc) { int i, j; --- base-commit: 76eeb9b8de9880ca38696b2fb56ac45ac0a25c6c change-id: 20250911-omapdrm-reduce-clang-stack-usage-pt-2-9a9ae9263b91 Best regards, -- Nathan Chancellor <nathan(a)kernel.org>

1 month, 2 weeks

2
1
0 0

[PATCH wireless-next] wifi: mt76: mt7921u: Add VID/PID for Netgear A7500

by Nick

Add VID/PID 0846/9065 for Netgear A7500. Reported-by: Autumn Dececco <autumndececco(a)gmail.com> Tested-by: Autumn Dececco <autumndececco(a)gmail.com> Signed-off-by: Nick Morrow <morrownr(a)gmail.com> Cc: stable(a)vger.kernel.org --- drivers/net/wireless/mediatek/mt76/mt7921/usb.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/drivers/net/wireless/mediatek/mt76/mt7921/usb.c b/drivers/net/wireless/mediatek/mt76/mt7921/usb.c index fe9751851ff7..100bdba32ba5 100644 --- a/drivers/net/wireless/mediatek/mt76/mt7921/usb.c +++ b/drivers/net/wireless/mediatek/mt76/mt7921/usb.c @@ -21,6 +21,9 @@ static const struct usb_device_id mt7921u_device_table[] = { /* Netgear, Inc. [A8000,AXE3000] */ { USB_DEVICE_AND_INTERFACE_INFO(0x0846, 0x9060, 0xff, 0xff, 0xff), .driver_info = (kernel_ulong_t)MT7921_FIRMWARE_WM }, + /* Netgear, Inc. A7500 */ + { USB_DEVICE_AND_INTERFACE_INFO(0x0846, 0x9065, 0xff, 0xff, 0xff), + .driver_info = (kernel_ulong_t)MT7921_FIRMWARE_WM }, /* TP-Link TXE50UH */ { USB_DEVICE_AND_INTERFACE_INFO(0x35bc, 0x0107, 0xff, 0xff, 0xff), .driver_info = (kernel_ulong_t)MT7921_FIRMWARE_WM }, -- 2.47.3

1 month, 2 weeks

2
1
0 0

[PATCH v2] arm64: dts: qcom: sdm845-shift-axolotl: Fix typo of compatible

by Tamura Dai

The bug is a typo in the compatible string for the touchscreen node. According to Documentation/devicetree/bindings/input/touchscreen/edt-ft5x06.yaml, the correct compatible is "focaltech,ft8719", but the device tree used "focaltech,fts8719". Fixes: 45882459159de (arm64: dts: qcom: sdm845: add device tree for SHIFT6mq) Cc: stable(a)vger.kernel.org Signed-off-by: Tamura Dai <kirinode0(a)gmail.com> --- arch/arm64/boot/dts/qcom/sdm845-shift-axolotl.dts | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/arm64/boot/dts/qcom/sdm845-shift-axolotl.dts b/arch/arm64/boot/dts/qcom/sdm845-shift-axolotl.dts index 2cf7b5e1243c..a0b288d6162f 100644 --- a/arch/arm64/boot/dts/qcom/sdm845-shift-axolotl.dts +++ b/arch/arm64/boot/dts/qcom/sdm845-shift-axolotl.dts @@ -432,7 +432,7 @@ &i2c5 { status = "okay"; touchscreen@38 { - compatible = "focaltech,fts8719"; + compatible = "focaltech,ft8719"; reg = <0x38>; wakeup-source; interrupt-parent = <&tlmm>; -- 2.34.1

1 month, 2 weeks

3
2
0 0

[PATCH] arm64: dts: qcom: sdm845-shift-axolotl: Fix typo of compatible

by Tamura Dai

Merge a hibernation regression fix and an fix related to energy model The bug is a typo in the compatible string for the touchscreen node. According to Documentation/devicetree/bindings/input/touchscreen/edt-ft5x06.yaml, the correct compatible is "focaltech,ft8719", but the device tree used "focaltech,fts8719". Fixes: 45882459159de (arm64: dts: qcom: sdm845: add device tree for SHIFT6mq) Cc: stable(a)vger.kernel.org Signed-off-by: Tamura Dai <kirinode0(a)gmail.com> --- arch/arm64/boot/dts/qcom/sdm845-shift-axolotl.dts | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/arm64/boot/dts/qcom/sdm845-shift-axolotl.dts b/arch/arm64/boot/dts/qcom/sdm845-shift-axolotl.dts index 2cf7b5e1243c..a0b288d6162f 100644 --- a/arch/arm64/boot/dts/qcom/sdm845-shift-axolotl.dts +++ b/arch/arm64/boot/dts/qcom/sdm845-shift-axolotl.dts @@ -432,7 +432,7 @@ &i2c5 { status = "okay"; touchscreen@38 { - compatible = "focaltech,fts8719"; + compatible = "focaltech,ft8719"; reg = <0x38>; wakeup-source; interrupt-parent = <&tlmm>; -- 2.34.1

1 month, 2 weeks

1
0
0 0

[PATCH] arm64 defconfig: remove CONFIG_SCHED_DEBUG reference

by Mikko Rapeli

It has been completely removed since v6.14-rc6 by commit dd5bdaf2b72da81d57f4f99e518af80002b6562e Author: Ingo Molnar <mingo(a)kernel.org> AuthorDate: Mon Mar 17 11:42:54 2025 +0100 Commit: Ingo Molnar <mingo(a)kernel.org> CommitDate: Wed Mar 19 22:20:53 2025 +0100 sched/debug: Make CONFIG_SCHED_DEBUG functionality unconditional Fixes yocto meta-arm sbsa-ref kernel config warning which uses kernel.org arm64 defconfig: DEBUG: Executing python function do_kernel_configcheck WARNING: [kernel config]: This BSP contains fragments with warnings: [INFO]: the following symbols were not found in the active configuration: - CONFIG_SCHED_DEBUG DEBUG: Python function do_kernel_configcheck finished Fixes: dd5bdaf2b72d ("sched/debug: Make CONFIG_SCHED_DEBUG functionality unconditional") Cc: <stable(a)vger.kernel.org> Cc: Jon Mason <jon.mason(a)arm.com> Cc: Ross Burton <ross.burton(a)arm.com> Cc: bruce.ashfield(a)gmail.com Signed-off-by: Mikko Rapeli <mikko.rapeli(a)linaro.org> --- arch/arm64/configs/defconfig | 1 - 1 file changed, 1 deletion(-) diff --git a/arch/arm64/configs/defconfig b/arch/arm64/configs/defconfig index 58f87d09366cd..4126281665bf2 100644 --- a/arch/arm64/configs/defconfig +++ b/arch/arm64/configs/defconfig @@ -1798,7 +1798,6 @@ CONFIG_DEBUG_INFO_DWARF_TOOLCHAIN_DEFAULT=y CONFIG_DEBUG_INFO_REDUCED=y CONFIG_MAGIC_SYSRQ=y CONFIG_DEBUG_FS=y -# CONFIG_SCHED_DEBUG is not set # CONFIG_DEBUG_PREEMPT is not set # CONFIG_FTRACE is not set CONFIG_CORESIGHT=m -- 2.34.1

1 month, 2 weeks

2
2
0 0

Fwd: [PATCH] fs/netfs: fix reference leak

by Max Kellermann

Hi Greg, sorry I mistyped the "stable" email address, so I'm forwarding this patch to you. Here's the original email: https://lore.kernel.org/lkml/20250911222501.1417765-1-max.kellermann@ionos.… ---------- Forwarded message --------- From: Max Kellermann <max.kellermann(a)ionos.com> Date: Fri, Sep 12, 2025 at 12:25 AM Subject: [PATCH] fs/netfs: fix reference leak To: David Howells <dhowells(a)redhat.com>, Paulo Alcantara <pc(a)manguebit.org>, Christian Brauner <brauner(a)kernel.org>, <netfs(a)lists.linux.dev>, <linux-fsdevel(a)vger.kernel.org>, <linux-kernel(a)vger.kernel.org> Cc: Max Kellermann <max.kellermann(a)ionos.com>, <linux-stable(a)vger.kernel.org> Commit 20d72b00ca81 ("netfs: Fix the request's work item to not require a ref") modified netfs_alloc_request() to initialize the reference counter to 2 instead of 1. The rationale was that the requet's "work" would release the second reference after completion (via netfs_{read,write}_collection_worker()). That works most of the time if all goes well. However, it leaks this additional reference if the request is released before the I/O operation has been submitted: the error code path only decrements the reference counter once and the work item will never be queued because there will never be a completion. This has caused outages of our whole server cluster today because tasks were blocked in netfs_wait_for_outstanding_io(), leading to deadlocks in Ceph (another bug that I will address soon in another patch). This was caused by a netfs_pgpriv2_begin_copy_to_cache() call which failed in fscache_begin_write_operation(). The leaked netfs_io_request was never completed, leaving `netfs_inode.io_count` with a positive value forever. All of this is super-fragile code. Finding out which code paths will lead to an eventual completion and which do not is hard to see: - Some functions like netfs_create_write_req() allocate a request, but will never submit any I/O. - netfs_unbuffered_read_iter_locked() calls netfs_unbuffered_read() and then netfs_put_request(); however, netfs_unbuffered_read() can also fail early before submitting the I/O request, therefore another netfs_put_request() call must be added there. A rule of thumb is that functions that return a `netfs_io_request` do not submit I/O, and all of their callers must be checked. For my taste, the whole netfs code needs an overhaul to make reference counting easier to understand and less fragile & obscure. But to fix this bug here and now and produce a patch that is adequate for a stable backport, I tried a minimal approach that quickly frees the request object upon early failure. I decided against adding a second netfs_put_request() each time because that would cause code duplication which obscures the code further. Instead, I added the function netfs_put_failed_request() which frees such a failed request synchronously under the assumption that the reference count is exactly 2 (as initially set by netfs_alloc_request() and never touched), verified by a WARN_ON_ONCE(). It then deinitializes the request object (without going through the "cleanup_work" indirection) and frees the allocation (without the "call_rcu" indirection). This should be safe because this is the same context that allocated/initialized the request and nobody else has a pointer to this object. All code paths that fail early have been changed to call netfs_put_failed_request() instead of netfs_put_request(). Additionally, I have added a netfs_put_request() call to netfs_unbuffered_read() as explained above because the netfs_put_failed_request() approach does not work there. Fixes: 20d72b00ca81 ("netfs: Fix the request's work item to not require a ref") Cc: linux-stable(a)vger.kernel.org Signed-off-by: Max Kellermann <max.kellermann(a)ionos.com> --- fs/netfs/buffered_read.c | 10 +++++----- fs/netfs/direct_read.c | 7 ++++++- fs/netfs/direct_write.c | 6 +++++- fs/netfs/internal.h | 1 + fs/netfs/objects.c | 32 +++++++++++++++++++++++++++++--- fs/netfs/read_pgpriv2.c | 2 +- fs/netfs/read_single.c | 2 +- fs/netfs/write_issue.c | 3 +-- 8 files changed, 49 insertions(+), 14 deletions(-) diff --git a/fs/netfs/buffered_read.c b/fs/netfs/buffered_read.c index 18b3dc74c70e..37ab6f28b5ad 100644 --- a/fs/netfs/buffered_read.c +++ b/fs/netfs/buffered_read.c @@ -369,7 +369,7 @@ void netfs_readahead(struct readahead_control *ractl) return netfs_put_request(rreq, netfs_rreq_trace_put_return); cleanup_free: - return netfs_put_request(rreq, netfs_rreq_trace_put_failed); + return netfs_put_failed_request(rreq); } EXPORT_SYMBOL(netfs_readahead); @@ -472,7 +472,7 @@ static int netfs_read_gaps(struct file *file, struct folio *folio) return ret < 0 ? ret : 0; discard: - netfs_put_request(rreq, netfs_rreq_trace_put_discard); + netfs_put_failed_request(rreq); alloc_error: folio_unlock(folio); return ret; @@ -532,7 +532,7 @@ int netfs_read_folio(struct file *file, struct folio *folio) return ret < 0 ? ret : 0; discard: - netfs_put_request(rreq, netfs_rreq_trace_put_discard); + netfs_put_failed_request(rreq); alloc_error: folio_unlock(folio); return ret; @@ -699,7 +699,7 @@ int netfs_write_begin(struct netfs_inode *ctx, return 0; error_put: - netfs_put_request(rreq, netfs_rreq_trace_put_failed); + netfs_put_failed_request(rreq); error: if (folio) { folio_unlock(folio); @@ -754,7 +754,7 @@ int netfs_prefetch_for_write(struct file *file, struct folio *folio, return ret < 0 ? ret : 0; error_put: - netfs_put_request(rreq, netfs_rreq_trace_put_discard); + netfs_put_failed_request(rreq); error: _leave(" = %d", ret); return ret; diff --git a/fs/netfs/direct_read.c b/fs/netfs/direct_read.c index a05e13472baf..a498ee8d6674 100644 --- a/fs/netfs/direct_read.c +++ b/fs/netfs/direct_read.c @@ -131,6 +131,7 @@ static ssize_t netfs_unbuffered_read(struct netfs_io_request *rreq, bool sync) if (rreq->len == 0) { pr_err("Zero-sized read [R=%x]\n", rreq->debug_id); + netfs_put_request(rreq, netfs_rreq_trace_put_discard); return -EIO; } @@ -205,7 +206,7 @@ ssize_t netfs_unbuffered_read_iter_locked(struct kiocb *iocb, struct iov_iter *i if (user_backed_iter(iter)) { ret = netfs_extract_user_iter(iter, rreq->len, &rreq->buffer.iter, 0); if (ret < 0) - goto out; + goto error_put; rreq->direct_bv = (struct bio_vec *)rreq->buffer.iter.bvec; rreq->direct_bv_count = ret; rreq->direct_bv_unpin = iov_iter_extract_will_pin(iter); @@ -238,6 +239,10 @@ ssize_t netfs_unbuffered_read_iter_locked(struct kiocb *iocb, struct iov_iter *i if (ret > 0) orig_count -= ret; return ret; + +error_put: + netfs_put_failed_request(rreq); + return ret; } EXPORT_SYMBOL(netfs_unbuffered_read_iter_locked); diff --git a/fs/netfs/direct_write.c b/fs/netfs/direct_write.c index a16660ab7f83..a9d1c3b2c084 100644 --- a/fs/netfs/direct_write.c +++ b/fs/netfs/direct_write.c @@ -57,7 +57,7 @@ ssize_t netfs_unbuffered_write_iter_locked(struct kiocb *iocb, struct iov_iter * n = netfs_extract_user_iter(iter, len, &wreq->buffer.iter, 0); if (n < 0) { ret = n; - goto out; + goto error_put; } wreq->direct_bv = (struct bio_vec *)wreq->buffer.iter.bvec; wreq->direct_bv_count = n; @@ -101,6 +101,10 @@ ssize_t netfs_unbuffered_write_iter_locked(struct kiocb *iocb, struct iov_iter * out: netfs_put_request(wreq, netfs_rreq_trace_put_return); return ret; + +error_put: + netfs_put_failed_request(wreq); + return ret; } EXPORT_SYMBOL(netfs_unbuffered_write_iter_locked); diff --git a/fs/netfs/internal.h b/fs/netfs/internal.h index d4f16fefd965..4319611f5354 100644 --- a/fs/netfs/internal.h +++ b/fs/netfs/internal.h @@ -87,6 +87,7 @@ struct netfs_io_request *netfs_alloc_request(struct address_space *mapping, void netfs_get_request(struct netfs_io_request *rreq, enum netfs_rreq_ref_trace what); void netfs_clear_subrequests(struct netfs_io_request *rreq); void netfs_put_request(struct netfs_io_request *rreq, enum netfs_rreq_ref_trace what); +void netfs_put_failed_request(struct netfs_io_request *rreq); struct netfs_io_subrequest *netfs_alloc_subrequest(struct netfs_io_request *rreq); static inline void netfs_see_request(struct netfs_io_request *rreq, diff --git a/fs/netfs/objects.c b/fs/netfs/objects.c index e8c99738b5bb..9a3fbb73325e 100644 --- a/fs/netfs/objects.c +++ b/fs/netfs/objects.c @@ -116,10 +116,8 @@ static void netfs_free_request_rcu(struct rcu_head *rcu) netfs_stat_d(&netfs_n_rh_rreq); } -static void netfs_free_request(struct work_struct *work) +static void netfs_deinit_request(struct netfs_io_request *rreq) { - struct netfs_io_request *rreq = - container_of(work, struct netfs_io_request, cleanup_work); struct netfs_inode *ictx = netfs_inode(rreq->inode); unsigned int i; @@ -149,6 +147,14 @@ static void netfs_free_request(struct work_struct *work) if (atomic_dec_and_test(&ictx->io_count)) wake_up_var(&ictx->io_count); +} + +static void netfs_free_request(struct work_struct *work) +{ + struct netfs_io_request *rreq = + container_of(work, struct netfs_io_request, cleanup_work); + + netfs_deinit_request(rreq); call_rcu(&rreq->rcu, netfs_free_request_rcu); } @@ -167,6 +173,26 @@ void netfs_put_request(struct netfs_io_request *rreq, enum netfs_rreq_ref_trace } } +/* + * Free a request (synchronously) that was just allocated but has + * failed before it could be submitted. + */ +void netfs_put_failed_request(struct netfs_io_request *rreq) +{ + /* new requests have two references (see + * netfs_alloc_request(), and this function is only allowed on + * new request objects + */ + WARN_ON_ONCE(refcount_read(&rreq->ref) != 2); + + trace_netfs_rreq_ref(rreq->debug_id, 0, netfs_rreq_trace_put_failed); + + netfs_deinit_request(rreq); + + mempool_free(rreq, rreq->netfs_ops->request_pool ?: &netfs_request_pool); + netfs_stat_d(&netfs_n_rh_rreq); +} + /* * Allocate and partially initialise an I/O request structure. */ diff --git a/fs/netfs/read_pgpriv2.c b/fs/netfs/read_pgpriv2.c index 8097bc069c1d..a1489aa29f78 100644 --- a/fs/netfs/read_pgpriv2.c +++ b/fs/netfs/read_pgpriv2.c @@ -118,7 +118,7 @@ static struct netfs_io_request *netfs_pgpriv2_begin_copy_to_cache( return creq; cancel_put: - netfs_put_request(creq, netfs_rreq_trace_put_return); + netfs_put_failed_request(creq); cancel: rreq->copy_to_cache = ERR_PTR(-ENOBUFS); clear_bit(NETFS_RREQ_FOLIO_COPY_TO_CACHE, &rreq->flags); diff --git a/fs/netfs/read_single.c b/fs/netfs/read_single.c index fa622a6cd56d..5c0dc4efc792 100644 --- a/fs/netfs/read_single.c +++ b/fs/netfs/read_single.c @@ -189,7 +189,7 @@ ssize_t netfs_read_single(struct inode *inode, struct file *file, struct iov_ite return ret; cleanup_free: - netfs_put_request(rreq, netfs_rreq_trace_put_failed); + netfs_put_failed_request(rreq); return ret; } EXPORT_SYMBOL(netfs_read_single); diff --git a/fs/netfs/write_issue.c b/fs/netfs/write_issue.c index 0584cba1a043..dd8743bc8d7f 100644 --- a/fs/netfs/write_issue.c +++ b/fs/netfs/write_issue.c @@ -133,8 +133,7 @@ struct netfs_io_request *netfs_create_write_req(struct address_space *mapping, return wreq; nomem: - wreq->error = -ENOMEM; - netfs_put_request(wreq, netfs_rreq_trace_put_failed); + netfs_put_failed_request(wreq); return ERR_PTR(-ENOMEM); } -- 2.47.3

1 month, 2 weeks

2
1
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror