April 2025 - Linux-stable-mirror

[PATCH] asm-generic/io.h: Skip trace helpers if rwmmio events are disabled

by Varad Gautam

With `CONFIG_TRACE_MMIO_ACCESS=y`, the `{read,write}{b,w,l,q}{_relaxed}()` mmio accessors unconditionally call `log_{post_}{read,write}_mmio()` helpers, which in turn call the ftrace ops for `rwmmio` trace events This adds a performance penalty per mmio accessor call, even when `rwmmio` events are disabled at runtime (~80% overhead on local measurement). Guard these with `tracepoint_enabled()`. Signed-off-by: Varad Gautam <varadgautam(a)google.com> Fixes: 210031971cdd ("asm-generic/io: Add logging support for MMIO accessors") Cc: <stable(a)vger.kernel.org> --- include/asm-generic/io.h | 98 +++++++++++++++++++++++++++------------- 1 file changed, 66 insertions(+), 32 deletions(-) diff --git a/include/asm-generic/io.h b/include/asm-generic/io.h index 3c61c29ff6ab..a9b5da547523 100644 --- a/include/asm-generic/io.h +++ b/include/asm-generic/io.h @@ -75,6 +75,7 @@ #if IS_ENABLED(CONFIG_TRACE_MMIO_ACCESS) && !(defined(__DISABLE_TRACE_MMIO__)) #include <linux/tracepoint-defs.h> +#define rwmmio_tracepoint_enabled(tracepoint) tracepoint_enabled(tracepoint) DECLARE_TRACEPOINT(rwmmio_write); DECLARE_TRACEPOINT(rwmmio_post_write); DECLARE_TRACEPOINT(rwmmio_read); @@ -91,6 +92,7 @@ void log_post_read_mmio(u64 val, u8 width, const volatile void __iomem *addr, #else +#define rwmmio_tracepoint_enabled(tracepoint) false static inline void log_write_mmio(u64 val, u8 width, volatile void __iomem *addr, unsigned long caller_addr, unsigned long caller_addr0) {} static inline void log_post_write_mmio(u64 val, u8 width, volatile void __iomem *addr, @@ -189,11 +191,13 @@ static inline u8 readb(const volatile void __iomem *addr) { u8 val; - log_read_mmio(8, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_read)) + log_read_mmio(8, addr, _THIS_IP_, _RET_IP_); __io_br(); val = __raw_readb(addr); __io_ar(val); - log_post_read_mmio(val, 8, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_post_read)) + log_post_read_mmio(val, 8, addr, _THIS_IP_, _RET_IP_); return val; } #endif @@ -204,11 +208,13 @@ static inline u16 readw(const volatile void __iomem *addr) { u16 val; - log_read_mmio(16, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_read)) + log_read_mmio(16, addr, _THIS_IP_, _RET_IP_); __io_br(); val = __le16_to_cpu((__le16 __force)__raw_readw(addr)); __io_ar(val); - log_post_read_mmio(val, 16, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_post_read)) + log_post_read_mmio(val, 16, addr, _THIS_IP_, _RET_IP_); return val; } #endif @@ -219,11 +225,13 @@ static inline u32 readl(const volatile void __iomem *addr) { u32 val; - log_read_mmio(32, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_read)) + log_read_mmio(32, addr, _THIS_IP_, _RET_IP_); __io_br(); val = __le32_to_cpu((__le32 __force)__raw_readl(addr)); __io_ar(val); - log_post_read_mmio(val, 32, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_post_read)) + log_post_read_mmio(val, 32, addr, _THIS_IP_, _RET_IP_); return val; } #endif @@ -235,11 +243,13 @@ static inline u64 readq(const volatile void __iomem *addr) { u64 val; - log_read_mmio(64, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_read)) + log_read_mmio(64, addr, _THIS_IP_, _RET_IP_); __io_br(); val = __le64_to_cpu((__le64 __force)__raw_readq(addr)); __io_ar(val); - log_post_read_mmio(val, 64, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_post_read)) + log_post_read_mmio(val, 64, addr, _THIS_IP_, _RET_IP_); return val; } #endif @@ -249,11 +259,13 @@ static inline u64 readq(const volatile void __iomem *addr) #define writeb writeb static inline void writeb(u8 value, volatile void __iomem *addr) { - log_write_mmio(value, 8, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_write)) + log_write_mmio(value, 8, addr, _THIS_IP_, _RET_IP_); __io_bw(); __raw_writeb(value, addr); __io_aw(); - log_post_write_mmio(value, 8, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_post_write)) + log_post_write_mmio(value, 8, addr, _THIS_IP_, _RET_IP_); } #endif @@ -261,11 +273,13 @@ static inline void writeb(u8 value, volatile void __iomem *addr) #define writew writew static inline void writew(u16 value, volatile void __iomem *addr) { - log_write_mmio(value, 16, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_write)) + log_write_mmio(value, 16, addr, _THIS_IP_, _RET_IP_); __io_bw(); __raw_writew((u16 __force)cpu_to_le16(value), addr); __io_aw(); - log_post_write_mmio(value, 16, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_post_write)) + log_post_write_mmio(value, 16, addr, _THIS_IP_, _RET_IP_); } #endif @@ -273,11 +287,13 @@ static inline void writew(u16 value, volatile void __iomem *addr) #define writel writel static inline void writel(u32 value, volatile void __iomem *addr) { - log_write_mmio(value, 32, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_write)) + log_write_mmio(value, 32, addr, _THIS_IP_, _RET_IP_); __io_bw(); __raw_writel((u32 __force)__cpu_to_le32(value), addr); __io_aw(); - log_post_write_mmio(value, 32, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_post_write)) + log_post_write_mmio(value, 32, addr, _THIS_IP_, _RET_IP_); } #endif @@ -286,11 +302,13 @@ static inline void writel(u32 value, volatile void __iomem *addr) #define writeq writeq static inline void writeq(u64 value, volatile void __iomem *addr) { - log_write_mmio(value, 64, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_write)) + log_write_mmio(value, 64, addr, _THIS_IP_, _RET_IP_); __io_bw(); __raw_writeq((u64 __force)__cpu_to_le64(value), addr); __io_aw(); - log_post_write_mmio(value, 64, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_post_write)) + log_post_write_mmio(value, 64, addr, _THIS_IP_, _RET_IP_); } #endif #endif /* CONFIG_64BIT */ @@ -306,9 +324,11 @@ static inline u8 readb_relaxed(const volatile void __iomem *addr) { u8 val; - log_read_mmio(8, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_read)) + log_read_mmio(8, addr, _THIS_IP_, _RET_IP_); val = __raw_readb(addr); - log_post_read_mmio(val, 8, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_post_read)) + log_post_read_mmio(val, 8, addr, _THIS_IP_, _RET_IP_); return val; } #endif @@ -319,9 +339,11 @@ static inline u16 readw_relaxed(const volatile void __iomem *addr) { u16 val; - log_read_mmio(16, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_read)) + log_read_mmio(16, addr, _THIS_IP_, _RET_IP_); val = __le16_to_cpu((__le16 __force)__raw_readw(addr)); - log_post_read_mmio(val, 16, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_post_read)) + log_post_read_mmio(val, 16, addr, _THIS_IP_, _RET_IP_); return val; } #endif @@ -332,9 +354,11 @@ static inline u32 readl_relaxed(const volatile void __iomem *addr) { u32 val; - log_read_mmio(32, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_read)) + log_read_mmio(32, addr, _THIS_IP_, _RET_IP_); val = __le32_to_cpu((__le32 __force)__raw_readl(addr)); - log_post_read_mmio(val, 32, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_post_read)) + log_post_read_mmio(val, 32, addr, _THIS_IP_, _RET_IP_); return val; } #endif @@ -345,9 +369,11 @@ static inline u64 readq_relaxed(const volatile void __iomem *addr) { u64 val; - log_read_mmio(64, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_read)) + log_read_mmio(64, addr, _THIS_IP_, _RET_IP_); val = __le64_to_cpu((__le64 __force)__raw_readq(addr)); - log_post_read_mmio(val, 64, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_post_read)) + log_post_read_mmio(val, 64, addr, _THIS_IP_, _RET_IP_); return val; } #endif @@ -356,9 +382,11 @@ static inline u64 readq_relaxed(const volatile void __iomem *addr) #define writeb_relaxed writeb_relaxed static inline void writeb_relaxed(u8 value, volatile void __iomem *addr) { - log_write_mmio(value, 8, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_write)) + log_write_mmio(value, 8, addr, _THIS_IP_, _RET_IP_); __raw_writeb(value, addr); - log_post_write_mmio(value, 8, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_post_write)) + log_post_write_mmio(value, 8, addr, _THIS_IP_, _RET_IP_); } #endif @@ -366,9 +394,11 @@ static inline void writeb_relaxed(u8 value, volatile void __iomem *addr) #define writew_relaxed writew_relaxed static inline void writew_relaxed(u16 value, volatile void __iomem *addr) { - log_write_mmio(value, 16, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_write)) + log_write_mmio(value, 16, addr, _THIS_IP_, _RET_IP_); __raw_writew((u16 __force)cpu_to_le16(value), addr); - log_post_write_mmio(value, 16, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_post_write)) + log_post_write_mmio(value, 16, addr, _THIS_IP_, _RET_IP_); } #endif @@ -376,9 +406,11 @@ static inline void writew_relaxed(u16 value, volatile void __iomem *addr) #define writel_relaxed writel_relaxed static inline void writel_relaxed(u32 value, volatile void __iomem *addr) { - log_write_mmio(value, 32, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_write)) + log_write_mmio(value, 32, addr, _THIS_IP_, _RET_IP_); __raw_writel((u32 __force)__cpu_to_le32(value), addr); - log_post_write_mmio(value, 32, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_post_write)) + log_post_write_mmio(value, 32, addr, _THIS_IP_, _RET_IP_); } #endif @@ -386,9 +418,11 @@ static inline void writel_relaxed(u32 value, volatile void __iomem *addr) #define writeq_relaxed writeq_relaxed static inline void writeq_relaxed(u64 value, volatile void __iomem *addr) { - log_write_mmio(value, 64, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_write)) + log_write_mmio(value, 64, addr, _THIS_IP_, _RET_IP_); __raw_writeq((u64 __force)__cpu_to_le64(value), addr); - log_post_write_mmio(value, 64, addr, _THIS_IP_, _RET_IP_); + if (rwmmio_tracepoint_enabled(rwmmio_post_write)) + log_post_write_mmio(value, 64, addr, _THIS_IP_, _RET_IP_); } #endif -- 2.49.0.472.ge94155a9ec-goog

1 month, 1 week

1
3
0 0

[PATCH v4] usb: dwc3: Abort suspend on soft disconnect failure

by Kuen-Han Tsai

When dwc3_gadget_soft_disconnect() fails, dwc3_suspend_common() keeps going with the suspend, resulting in a period where the power domain is off, but the gadget driver remains connected. Within this time frame, invoking vbus_event_work() will cause an error as it attempts to access DWC3 registers for endpoint disabling after the power domain has been completely shut down. Abort the suspend sequence when dwc3_gadget_suspend() cannot halt the controller and proceeds with a soft connect. Fixes: 9f8a67b65a49 ("usb: dwc3: gadget: fix gadget suspend/resume") CC: stable(a)vger.kernel.org Signed-off-by: Kuen-Han Tsai <khtsai(a)google.com> --- Kernel panic - not syncing: Asynchronous SError Interrupt Workqueue: events vbus_event_work Call trace: dump_backtrace+0xf4/0x118 show_stack+0x18/0x24 dump_stack_lvl+0x60/0x7c dump_stack+0x18/0x3c panic+0x16c/0x390 nmi_panic+0xa4/0xa8 arm64_serror_panic+0x6c/0x94 do_serror+0xc4/0xd0 el1h_64_error_handler+0x34/0x48 el1h_64_error+0x68/0x6c readl+0x4c/0x8c __dwc3_gadget_ep_disable+0x48/0x230 dwc3_gadget_ep_disable+0x50/0xc0 usb_ep_disable+0x44/0xe4 ffs_func_eps_disable+0x64/0xc8 ffs_func_set_alt+0x74/0x368 ffs_func_disable+0x18/0x28 composite_disconnect+0x90/0xec configfs_composite_disconnect+0x64/0x88 usb_gadget_disconnect_locked+0xc0/0x168 vbus_event_work+0x3c/0x58 process_one_work+0x1e4/0x43c worker_thread+0x25c/0x430 kthread+0x104/0x1d4 ret_from_fork+0x10/0x20 --- Changelog: v4: - correct the mistake where semicolon was forgotten - return -EAGAIN upon dwc3_gadget_suspend() failure v3: - change the Fixes tag v2: - move declarations in separate lines - add the Fixes tag --- drivers/usb/dwc3/core.c | 9 +++++++-- drivers/usb/dwc3/gadget.c | 22 +++++++++------------- 2 files changed, 16 insertions(+), 15 deletions(-) diff --git a/drivers/usb/dwc3/core.c b/drivers/usb/dwc3/core.c index 66a08b527165..f36bc933c55b 100644 --- a/drivers/usb/dwc3/core.c +++ b/drivers/usb/dwc3/core.c @@ -2388,6 +2388,7 @@ static int dwc3_suspend_common(struct dwc3 *dwc, pm_message_t msg) { u32 reg; int i; + int ret; if (!pm_runtime_suspended(dwc->dev) && !PMSG_IS_AUTO(msg)) { dwc->susphy_state = (dwc3_readl(dwc->regs, DWC3_GUSB2PHYCFG(0)) & @@ -2406,7 +2407,9 @@ static int dwc3_suspend_common(struct dwc3 *dwc, pm_message_t msg) case DWC3_GCTL_PRTCAP_DEVICE: if (pm_runtime_suspended(dwc->dev)) break; - dwc3_gadget_suspend(dwc); + ret = dwc3_gadget_suspend(dwc); + if (ret) + return ret; synchronize_irq(dwc->irq_gadget); dwc3_core_exit(dwc); break; @@ -2441,7 +2444,9 @@ static int dwc3_suspend_common(struct dwc3 *dwc, pm_message_t msg) break; if (dwc->current_otg_role == DWC3_OTG_ROLE_DEVICE) { - dwc3_gadget_suspend(dwc); + ret = dwc3_gadget_suspend(dwc); + if (ret) + return ret; synchronize_irq(dwc->irq_gadget); } diff --git a/drivers/usb/dwc3/gadget.c b/drivers/usb/dwc3/gadget.c index 89a4dc8ebf94..630fd5f0ce97 100644 --- a/drivers/usb/dwc3/gadget.c +++ b/drivers/usb/dwc3/gadget.c @@ -4776,26 +4776,22 @@ int dwc3_gadget_suspend(struct dwc3 *dwc) int ret; ret = dwc3_gadget_soft_disconnect(dwc); - if (ret) - goto err; - - spin_lock_irqsave(&dwc->lock, flags); - if (dwc->gadget_driver) - dwc3_disconnect_gadget(dwc); - spin_unlock_irqrestore(&dwc->lock, flags); - - return 0; - -err: /* * Attempt to reset the controller's state. Likely no * communication can be established until the host * performs a port reset. */ - if (dwc->softconnect) + if (ret && dwc->softconnect) { dwc3_gadget_soft_connect(dwc); + return -EAGAIN; + } - return ret; + spin_lock_irqsave(&dwc->lock, flags); + if (dwc->gadget_driver) + dwc3_disconnect_gadget(dwc); + spin_unlock_irqrestore(&dwc->lock, flags); + + return 0; } int dwc3_gadget_resume(struct dwc3 *dwc) -- 2.49.0.604.gff1f9ca942-goog

1 month, 1 week

3
8
0 0

[PATCH v5 0/4] media: uvcvideo: Introduce V4L2_META_FMT_UVC_MSXU_1_5 + other meta fixes

by Ricardo Ribalda

This series introduces a new metadata format for UVC cameras and adds a couple of improvements to the UVC metadata handling. Signed-off-by: Ricardo Ribalda <ribalda(a)chromium.org> --- Changes in v5: - Fix codestyle and kerneldoc warnings reported by media-ci - Link to v4: https://lore.kernel.org/r/20250403-uvc-meta-v4-0-877aa6475975@chromium.org Changes in v4: - Rename format to V4L2_META_FMT_UVC_MSXU_1_5 (Thanks Mauro) - Flag the new format with a quirk. - Autodetect MSXU devices. - Link to v3: https://lore.kernel.org/linux-media/20250313-uvc-metadata-v3-0-c467af869c60… Changes in v3: - Fix doc syntax errors. - Link to v2: https://lore.kernel.org/r/20250306-uvc-metadata-v2-0-7e939857cad5@chromium.… Changes in v2: - Add metadata invalid fix - Move doc note to a separate patch - Introuce V4L2_META_FMT_UVC_CUSTOM (thanks HdG!). - Link to v1: https://lore.kernel.org/r/20250226-uvc-metadata-v1-1-6cd6fe5ec2cb@chromium.… --- Ricardo Ribalda (4): media: uvcvideo: Do not mark valid metadata as invalid media: Documentation: Add note about UVCH length field media: uvcvideo: Introduce V4L2_META_FMT_UVC_MSXU_1_5 media: uvcvideo: Auto-set UVC_QUIRK_MSXU_META .../userspace-api/media/v4l/meta-formats.rst | 1 + .../media/v4l/metafmt-uvc-msxu-1-5.rst | 23 +++++ .../userspace-api/media/v4l/metafmt-uvc.rst | 4 +- MAINTAINERS | 1 + drivers/media/usb/uvc/uvc_metadata.c | 97 ++++++++++++++++++++-- drivers/media/usb/uvc/uvc_video.c | 12 +-- drivers/media/usb/uvc/uvcvideo.h | 1 + drivers/media/v4l2-core/v4l2-ioctl.c | 1 + include/linux/usb/uvc.h | 3 + include/uapi/linux/videodev2.h | 1 + 10 files changed, 131 insertions(+), 13 deletions(-) --- base-commit: 4e82c87058f45e79eeaa4d5bcc3b38dd3dce7209 change-id: 20250403-uvc-meta-e556773d12ae Best regards, -- Ricardo Ribalda <ribalda(a)chromium.org>

1 month, 2 weeks

3
4
0 0

IPC drop down on AMD epyc 7702P

by Jean-Baptiste Roquefere

Hi, We (Ateme, a video encoding company) may have found an unwanted behavior in the scheduler since 5.10 (commit 16b0a7a1a0af), then 5.16 (commit c5b0a7eefc70), then 5.19 (commit not found yet), then maybe some other commits from 5.19 to 6.12, with a consequence of IPC decrease. Problem still appears on lasts 6.12, 6.13 and 6.14 We have reverted both 16b0a7a1a0af and c5b0a7eefc70 commits that reduce our performances (see fair.patch attached, applicable on 6.12.17). Performances increase but still doesnt reach our reference on 5.4.152. Instead of trying to find every single commits from 5.18 to 6.12 that could decrease our performance, I chosed to bench 5.4.152 versus 6.12.17 with and without fair.patch. The problem appeared clear : a lot of CPU migrations go out of CCX, then L3 miss, then IPC decrease. Context of our bench: video decoder which work at a regulated speed, 1 process, 21 main threads, everyone of them creates 10 threads, 8 of them have a fine granularity, meaning they go to sleep quite often, giving the scheduler a lot of opportunities to act). Hardware is an AMD Epyc 7702P, 128 cores, grouped by shared LLC 4 cores +4 hyperthreaded cores. NUMA topology is set by the BIOS to 1 node per socket. Every pthread are created with default attributes. I use AMDuProf (-C -A system -a -m ipc,l1,l2,l3,memory) for CPU Utilization (%), CPU effective freq, IPC, L2 access (pti), L2 miss (pti), L3 miss (absolute) and Mem (GB/s, and perf (stat -d -d -d -a) for Context switches, CPU migrations and Real time (s). We noted that upgrade 5.4.152 to 6.12.17 without any special preempt configuration : Two fold increase in CPU migration 30% memory bandwidth increase 20% L3 cache misses increase 10% IPC decrease With the attached fair.patch applied to 6.12.17 (reminder : this patch reverts one commit appeared in 5.10 and another in 5.16) we managed to reduce CPU migrations and increase IPC but not as much as we had on 5.4.152. Our goal is to keep kernel "clean" without any patch (we don't want to apply and maintain fair.patch) then for the rest of my email we will consider stock kernel 6.12.17. I've reduced the "sub threads count" to stays below 128 threads. Then still 21 main threads and instead of 10 worker per main thread I set 5 workers (4 of them with fine granularity) giving 105 pthreads -> everything goes fine in 6.12.17, no extra CPU migration, no extra memory bandwidth... But as soon as we increase worker threads count (10 instead of 5) the problem appears. We know our decoder may have too many threads but that's out of our scope, it has been designed like that some years ago and moving from "lot of small threads to few of big thread" is for now not possible. We have a work around : we group threads using pthread affinities. Every main thread (and by inheritance of affinities every worker threads) on a single CCX so we reduce the L3 miss for them, then decrease memory bandwidth, then finally increasing IPC. With that solution, we go above our original performances, for both kernels, and they perform at the same level. However, it is impractical to productize as such. I've tried many kernel build configurations (CONFIG_PREMPT_*, CONFIG_SCHEDULER_*, tuning of fair.c:sysctl_sched_migration_cost) on 6.12.17, 6.12.21 (longterm), 6.13.9 (mainline), and 6.14.0 Nothing changes. Q: Is there anyway to tune the kernel so we can get our performance back without using the pthread affinities work around ? Feel free to ask an archive containing binaries and payload. I first posted on https://bugzilla.kernel.org/show_bug.cgi?id=220000 but one told me the best way to get answers where these mailing lists Regards, Jean-Baptiste Roquefere, Ateme Attached bench.tar.gz : * bench/fair.patch * bench/bench.ods with 2 sheets : - regulated : decoder speed is regulated to keep real time constant - no regul : decoder speed is not regulated and uses from 1 to 76 main threads with 10 worker per main thread * bench/regulated.csv : bench.ods:regulated exported in csv format * bench/not-regulated : bench.ods:no regul exported in csv format

1 month, 2 weeks

5
15
0 0

FAILED: patch "[PATCH] Input: iqs7222 - preserve system status register" failed to apply to 6.1-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 6.1-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-6.1.y git checkout FETCH_HEAD git cherry-pick -x a2add513311b48cc924a699a8174db2c61ed5e8a # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025031607-striking-creasing-c69c@gregkh' --subject-prefix 'PATCH 6.1.y' HEAD^.. Possible dependencies: thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From a2add513311b48cc924a699a8174db2c61ed5e8a Mon Sep 17 00:00:00 2001 From: Jeff LaBundy <jeff(a)labundy.com> Date: Sun, 9 Mar 2025 20:29:59 -0500 Subject: [PATCH] Input: iqs7222 - preserve system status register Some register groups reserve a byte at the end of their continuous address space. Depending on the variant of silicon, this field may share the same memory space as the lower byte of the system status register (0x10). In these cases, caching the reserved byte and writing it later may effectively reset the device depending on what happened in between the read and write operations. Solve this problem by avoiding any access to this last byte within offending register groups. This method replaces a workaround which attempted to write the reserved byte with up-to-date contents, but left a small window in which updates by the device could have been clobbered. Now that the driver does not touch these reserved bytes, the order in which the device's registers are written no longer matters, and they can be written in their natural order. The new method is also much more generic, and can be more easily extended to new variants of silicon with different register maps. As part of this change, the register read and write functions must be gently updated to support byte access instead of word access. Fixes: 2e70ef525b73 ("Input: iqs7222 - acknowledge reset before writing registers") Signed-off-by: Jeff LaBundy <jeff(a)labundy.com> Link: https://lore.kernel.org/r/Z85Alw+d9EHKXx2e@nixie71 Cc: stable(a)vger.kernel.org Signed-off-by: Dmitry Torokhov <dmitry.torokhov(a)gmail.com> diff --git a/drivers/input/misc/iqs7222.c b/drivers/input/misc/iqs7222.c index 22022d11470d..80b917944b51 100644 --- a/drivers/input/misc/iqs7222.c +++ b/drivers/input/misc/iqs7222.c @@ -100,11 +100,11 @@ enum iqs7222_reg_key_id { enum iqs7222_reg_grp_id { IQS7222_REG_GRP_STAT, - IQS7222_REG_GRP_FILT, IQS7222_REG_GRP_CYCLE, IQS7222_REG_GRP_GLBL, IQS7222_REG_GRP_BTN, IQS7222_REG_GRP_CHAN, + IQS7222_REG_GRP_FILT, IQS7222_REG_GRP_SLDR, IQS7222_REG_GRP_TPAD, IQS7222_REG_GRP_GPIO, @@ -286,6 +286,7 @@ static const struct iqs7222_event_desc iqs7222_tp_events[] = { struct iqs7222_reg_grp_desc { u16 base; + u16 val_len; int num_row; int num_col; }; @@ -342,6 +343,7 @@ static const struct iqs7222_dev_desc iqs7222_devs[] = { }, [IQS7222_REG_GRP_FILT] = { .base = 0xAC00, + .val_len = 3, .num_row = 1, .num_col = 2, }, @@ -400,6 +402,7 @@ static const struct iqs7222_dev_desc iqs7222_devs[] = { }, [IQS7222_REG_GRP_FILT] = { .base = 0xAC00, + .val_len = 3, .num_row = 1, .num_col = 2, }, @@ -454,6 +457,7 @@ static const struct iqs7222_dev_desc iqs7222_devs[] = { }, [IQS7222_REG_GRP_FILT] = { .base = 0xC400, + .val_len = 3, .num_row = 1, .num_col = 2, }, @@ -496,6 +500,7 @@ static const struct iqs7222_dev_desc iqs7222_devs[] = { }, [IQS7222_REG_GRP_FILT] = { .base = 0xC400, + .val_len = 3, .num_row = 1, .num_col = 2, }, @@ -543,6 +548,7 @@ static const struct iqs7222_dev_desc iqs7222_devs[] = { }, [IQS7222_REG_GRP_FILT] = { .base = 0xAA00, + .val_len = 3, .num_row = 1, .num_col = 2, }, @@ -600,6 +606,7 @@ static const struct iqs7222_dev_desc iqs7222_devs[] = { }, [IQS7222_REG_GRP_FILT] = { .base = 0xAA00, + .val_len = 3, .num_row = 1, .num_col = 2, }, @@ -656,6 +663,7 @@ static const struct iqs7222_dev_desc iqs7222_devs[] = { }, [IQS7222_REG_GRP_FILT] = { .base = 0xAE00, + .val_len = 3, .num_row = 1, .num_col = 2, }, @@ -712,6 +720,7 @@ static const struct iqs7222_dev_desc iqs7222_devs[] = { }, [IQS7222_REG_GRP_FILT] = { .base = 0xAE00, + .val_len = 3, .num_row = 1, .num_col = 2, }, @@ -768,6 +777,7 @@ static const struct iqs7222_dev_desc iqs7222_devs[] = { }, [IQS7222_REG_GRP_FILT] = { .base = 0xAE00, + .val_len = 3, .num_row = 1, .num_col = 2, }, @@ -1604,7 +1614,7 @@ static int iqs7222_force_comms(struct iqs7222_private *iqs7222) } static int iqs7222_read_burst(struct iqs7222_private *iqs7222, - u16 reg, void *val, u16 num_val) + u16 reg, void *val, u16 val_len) { u8 reg_buf[sizeof(__be16)]; int ret, i; @@ -1619,7 +1629,7 @@ static int iqs7222_read_burst(struct iqs7222_private *iqs7222, { .addr = client->addr, .flags = I2C_M_RD, - .len = num_val * sizeof(__le16), + .len = val_len, .buf = (u8 *)val, }, }; @@ -1675,7 +1685,7 @@ static int iqs7222_read_word(struct iqs7222_private *iqs7222, u16 reg, u16 *val) __le16 val_buf; int error; - error = iqs7222_read_burst(iqs7222, reg, &val_buf, 1); + error = iqs7222_read_burst(iqs7222, reg, &val_buf, sizeof(val_buf)); if (error) return error; @@ -1685,10 +1695,9 @@ static int iqs7222_read_word(struct iqs7222_private *iqs7222, u16 reg, u16 *val) } static int iqs7222_write_burst(struct iqs7222_private *iqs7222, - u16 reg, const void *val, u16 num_val) + u16 reg, const void *val, u16 val_len) { int reg_len = reg > U8_MAX ? sizeof(reg) : sizeof(u8); - int val_len = num_val * sizeof(__le16); int msg_len = reg_len + val_len; int ret, i; struct i2c_client *client = iqs7222->client; @@ -1747,7 +1756,7 @@ static int iqs7222_write_word(struct iqs7222_private *iqs7222, u16 reg, u16 val) { __le16 val_buf = cpu_to_le16(val); - return iqs7222_write_burst(iqs7222, reg, &val_buf, 1); + return iqs7222_write_burst(iqs7222, reg, &val_buf, sizeof(val_buf)); } static int iqs7222_ati_trigger(struct iqs7222_private *iqs7222) @@ -1831,30 +1840,14 @@ static int iqs7222_dev_init(struct iqs7222_private *iqs7222, int dir) /* * Acknowledge reset before writing any registers in case the device - * suffers a spurious reset during initialization. Because this step - * may change the reserved fields of the second filter beta register, - * its cache must be updated. - * - * Writing the second filter beta register, in turn, may clobber the - * system status register. As such, the filter beta register pair is - * written first to protect against this hazard. + * suffers a spurious reset during initialization. */ if (dir == WRITE) { - u16 reg = dev_desc->reg_grps[IQS7222_REG_GRP_FILT].base + 1; - u16 filt_setup; - error = iqs7222_write_word(iqs7222, IQS7222_SYS_SETUP, iqs7222->sys_setup[0] | IQS7222_SYS_SETUP_ACK_RESET); if (error) return error; - - error = iqs7222_read_word(iqs7222, reg, &filt_setup); - if (error) - return error; - - iqs7222->filt_setup[1] &= GENMASK(7, 0); - iqs7222->filt_setup[1] |= (filt_setup & ~GENMASK(7, 0)); } /* @@ -1883,6 +1876,7 @@ static int iqs7222_dev_init(struct iqs7222_private *iqs7222, int dir) int num_col = dev_desc->reg_grps[i].num_col; u16 reg = dev_desc->reg_grps[i].base; __le16 *val_buf; + u16 val_len = dev_desc->reg_grps[i].val_len ? : num_col * sizeof(*val_buf); u16 *val; if (!num_col) @@ -1900,7 +1894,7 @@ static int iqs7222_dev_init(struct iqs7222_private *iqs7222, int dir) switch (dir) { case READ: error = iqs7222_read_burst(iqs7222, reg, - val_buf, num_col); + val_buf, val_len); for (k = 0; k < num_col; k++) val[k] = le16_to_cpu(val_buf[k]); break; @@ -1909,7 +1903,7 @@ static int iqs7222_dev_init(struct iqs7222_private *iqs7222, int dir) for (k = 0; k < num_col; k++) val_buf[k] = cpu_to_le16(val[k]); error = iqs7222_write_burst(iqs7222, reg, - val_buf, num_col); + val_buf, val_len); break; default: @@ -1962,7 +1956,7 @@ static int iqs7222_dev_info(struct iqs7222_private *iqs7222) int error, i; error = iqs7222_read_burst(iqs7222, IQS7222_PROD_NUM, dev_id, - ARRAY_SIZE(dev_id)); + sizeof(dev_id)); if (error) return error; @@ -2915,7 +2909,7 @@ static int iqs7222_report(struct iqs7222_private *iqs7222) __le16 status[IQS7222_MAX_COLS_STAT]; error = iqs7222_read_burst(iqs7222, IQS7222_SYS_STATUS, status, - num_stat); + num_stat * sizeof(*status)); if (error) return error;

1 month, 2 weeks

3
3
0 0

[PATCH 6.12 000/122] 6.12.11-rc1 review

by Greg Kroah-Hartman

This is the start of the stable review cycle for the 6.12.11 release. There are 122 patches in this series, all will be posted as a response to this one. If anyone has any issues with these being applied, please let me know. Responses should be made by Thu, 23 Jan 2025 17:45:02 +0000. Anything received after that time might be too late. The whole patch series can be found in one patch at: https://www.kernel.org/pub/linux/kernel/v6.x/stable-review/patch-6.12.11-rc… or in the git tree and branch at: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable-rc.git linux-6.12.y and the diffstat can be found below. thanks, greg k-h ------------- Pseudo-Shortlog of commits: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> Linux 6.12.11-rc1 Ryan Lee <ryan.lee(a)canonical.com> apparmor: allocate xmatch for nullpdb inside aa_alloc_null Wayne Lin <Wayne.Lin(a)amd.com> drm/amd/display: Validate mdoe under MST LCT=1 case as well Nicholas Susanto <Nicholas.Susanto(a)amd.com> Revert "drm/amd/display: Enable urgent latency adjustments for DCN35" Leo Li <sunpeng.li(a)amd.com> drm/amd/display: Do not wait for PSR disable on vbl enable Tom Chung <chiahsuan.chung(a)amd.com> drm/amd/display: Disable replay and psr while VRR is enabled Tom Chung <chiahsuan.chung(a)amd.com> drm/amd/display: Fix PSR-SU not support but still call the amdgpu_dm_psr_enable Christian König <christian.koenig(a)amd.com> drm/amdgpu: always sync the GFX pipe on ctx switch Kenneth Feng <kenneth.feng(a)amd.com> drm/amdgpu: disable gfxoff with the compute workload on gfx12 Gui Chengming <Jack.Gui(a)amd.com> drm/amdgpu: fix fw attestation for MP0_14_0_{2/3} Alex Deucher <alexander.deucher(a)amd.com> drm/amdgpu/smu13: update powersave optimizations Ashutosh Dixit <ashutosh.dixit(a)intel.com> drm/xe/oa: Add missing VISACTL mux registers Matthew Brost <matthew.brost(a)intel.com> drm/xe: Mark ComputeCS read mode as UC on iGPU Ville Syrjälä <ville.syrjala(a)linux.intel.com> drm/i915/fb: Relax clear color alignment to 64 bytes Xin Li (Intel) <xin(a)zytor.com> x86/fred: Fix the FRED RSP0 MSR out of sync with its per-CPU cache Frederic Weisbecker <frederic(a)kernel.org> timers/migration: Enforce group initialization visibility to tree walkers Frederic Weisbecker <frederic(a)kernel.org> timers/migration: Fix another race between hotplug and idle entry/exit Koichiro Den <koichiro.den(a)canonical.com> hrtimers: Handle CPU state correctly on hotplug Tomas Krcka <krckatom(a)amazon.de> irqchip/gic-v3-its: Don't enable interrupts in its_irq_set_vcpu_affinity() Yogesh Lal <quic_ylal(a)quicinc.com> irqchip/gic-v3: Handle CPU_PM_ENTER_FAILED correctly Joe Hattori <joe(a)pf.is.s.u-tokyo.ac.jp> irqchip: Plug a OF node reference leak in platform_irqchip_probe() Steven Rostedt <rostedt(a)goodmis.org> tracing: gfp: Fix the GFP enum values shown for user space tracing tools Donet Tom <donettom(a)linux.ibm.com> mm: vmscan : pgdemote vmstat is not getting updated when MGLRU is enabled. Ryan Roberts <ryan.roberts(a)arm.com> mm: clear uffd-wp PTE/PMD state on mremap() Leo Li <sunpeng.li(a)amd.com> drm/amd/display: Do not elevate mem_type change to full update Ryan Roberts <ryan.roberts(a)arm.com> selftests/mm: set allocated memory to non-zero content in cow test Guo Weikang <guoweikang.kernel(a)gmail.com> mm/kmemleak: fix percpu memory leak detection failure Xiaolei Wang <xiaolei.wang(a)windriver.com> pmdomain: imx8mp-blk-ctrl: add missing loop break condition Suren Baghdasaryan <surenb(a)google.com> tools: fix atomic_set() definition to set the value correctly Sean Anderson <sean.anderson(a)linux.dev> gpio: xilinx: Convert gpio_lock to raw spinlock Rik van Riel <riel(a)surriel.com> fs/proc: fix softlockup in __read_vmcore (part 2) Marco Nelissen <marco.nelissen(a)gmail.com> filemap: avoid truncating 64-bit offset to 32 bits Paul Fertser <fercerpav(a)gmail.com> net/ncsi: fix locking in Get MAC Address handling Takashi Iwai <tiwai(a)suse.de> drm/nouveau/disp: Fix missing backlight control on Macbook 5,1 Dave Airlie <airlied(a)redhat.com> nouveau/fence: handle cross device fences properly Stefano Garzarella <sgarzare(a)redhat.com> vsock: prevent null-ptr-deref in vsock_*[has_data|has_space] Stefano Garzarella <sgarzare(a)redhat.com> vsock: reset socket state when de-assigning the transport Stefano Garzarella <sgarzare(a)redhat.com> vsock/virtio: cancel close work in the destructor Stefano Garzarella <sgarzare(a)redhat.com> vsock/virtio: discard packets if the transport changes Stefano Garzarella <sgarzare(a)redhat.com> vsock/bpf: return early if transport is not assigned Heiner Kallweit <hkallweit1(a)gmail.com> net: ethernet: xgbe: re-add aneg to supported features in PHY quirks Paolo Abeni <pabeni(a)redhat.com> selftests: mptcp: avoid spurious errors on disconnect Paolo Abeni <pabeni(a)redhat.com> mptcp: fix spurious wake-up on under memory pressure Paolo Abeni <pabeni(a)redhat.com> mptcp: be sure to send ack when mptcp-level window re-opens Tomi Valkeinen <tomi.valkeinen+renesas(a)ideasonboard.com> i2c: atr: Fix client detach Kairui Song <kasong(a)tencent.com> zram: fix potential UAF of zram table Luke D. Jones <luke(a)ljones.dev> ALSA: hda/realtek: fixup ASUS H7606W Luke D. Jones <luke(a)ljones.dev> ALSA: hda/realtek: fixup ASUS GA605W Stefan Binding <sbinding(a)opensource.cirrus.com> ALSA: hda/realtek: Add support for Ayaneo System using CS35L41 HDA Juergen Gross <jgross(a)suse.com> x86/asm: Make serialize() always_inline Peter Zijlstra <peterz(a)infradead.org> sched/fair: Fix update_cfs_group() vs DELAY_DEQUEUE Peter Zijlstra <peterz(a)infradead.org> sched/fair: Fix EEVDF entity placement bug causing scheduling lag Luis Chamberlain <mcgrof(a)kernel.org> nvmet: propagate npwg topology Tejun Heo <tj(a)kernel.org> sched_ext: Fix dsq_local_on selftest Hongguang Gao <hongguang.gao(a)broadcom.com> RDMA/bnxt_re: Fix to export port num to ib_query_qp David Vernet <void(a)manifault.com> scx: Fix maximal BPF selftest prog Ihor Solodrai <ihor.solodrai(a)pm.me> selftests/sched_ext: fix build after renames in sched_ext API Oleg Nesterov <oleg(a)redhat.com> poll_wait: add mb() to fix theoretical race between waitqueue_active() and .poll() Lizhi Xu <lizhi.xu(a)windriver.com> afs: Fix merge preference rule failure condition Marco Nelissen <marco.nelissen(a)gmail.com> iomap: avoid avoid truncating 64-bit offset to 32 bits Henry Huang <henry.hj(a)antgroup.com> sched_ext: keep running prev when prev->scx.slice != 0 Hans de Goede <hdegoede(a)redhat.com> ACPI: resource: acpi_dev_irq_override(): Check DMI match last Srinivas Pandruvada <srinivas.pandruvada(a)linux.intel.com> platform/x86: ISST: Add Clearwater Forest to support list Srinivas Pandruvada <srinivas.pandruvada(a)linux.intel.com> platform/x86/intel: power-domains: Add Clearwater Forest support Jakub Kicinski <kuba(a)kernel.org> selftests: tc-testing: reduce rshift value Koichiro Den <koichiro.den(a)canonical.com> gpio: sim: lock up configfs that an instantiated device depends on Koichiro Den <koichiro.den(a)canonical.com> gpio: virtuser: lock up configfs that an instantiated device depends on Manivannan Sadhasivam <manivannan.sadhasivam(a)linaro.org> scsi: ufs: core: Honor runtime/system PM levels if set by host controller drivers Max Kellermann <max.kellermann(a)ionos.com> cachefiles: Parse the "secctx" immediately David Howells <dhowells(a)redhat.com> netfs: Fix non-contiguous donation between completed reads David Howells <dhowells(a)redhat.com> kheaders: Ignore silly-rename files Zhang Kunbo <zhangkunbo(a)huawei.com> fs: fix missing declaration of init_files Brahmajit Das <brahmajit.xyz(a)gmail.com> fs/qnx6: Fix building with GCC 15 Leo Stone <leocstone(a)gmail.com> hfs: Sanity check the root record Lizhi Xu <lizhi.xu(a)windriver.com> mac802154: check local interfaces before deleting sdata list Paulo Alcantara <pc(a)manguebit.com> smb: client: fix double free of TCP_Server_Info::hostname David Lechner <dlechner(a)baylibre.com> hwmon: (ltc2991) Fix mixed signed/unsigned in DIV_ROUND_CLOSEST Wolfram Sang <wsa+renesas(a)sang-engineering.com> i2c: testunit: on errors, repeat NACK until STOP Wolfram Sang <wsa+renesas(a)sang-engineering.com> i2c: rcar: fix NACK handling when being a target Wolfram Sang <wsa+renesas(a)sang-engineering.com> i2c: mux: demux-pinctrl: check initial mux selection, too Pratyush Yadav <pratyush(a)kernel.org> Revert "mtd: spi-nor: core: replace dummy buswidth from addr to data" David Lechner <dlechner(a)baylibre.com> hwmon: (tmp513) Fix division of negative numbers Chenyuan Yang <chenyuan0y(a)gmail.com> platform/x86: lenovo-yoga-tab2-pro-1380-fastcharger: fix serdev race Chenyuan Yang <chenyuan0y(a)gmail.com> platform/x86: dell-uart-backlight: fix serdev race Joe Hattori <joe(a)pf.is.s.u-tokyo.ac.jp> i2c: core: fix reference leak in i2c_register_adapter() MD Danish Anwar <danishanwar(a)ti.com> soc: ti: pruss: Fix pruss APIs Claudiu Beznea <claudiu.beznea.uj(a)bp.renesas.com> reset: rzg2l-usbphy-ctrl: Assign proper of node to the allocated device Maíra Canal <mcanal(a)igalia.com> drm/v3d: Ensure job pointer is set to NULL after job completion Ian Forbes <ian.forbes(a)broadcom.com> drm/vmwgfx: Add new keep_resv BO param Ian Forbes <ian.forbes(a)broadcom.com> drm/vmwgfx: Unreserve BO on error Yu-Chun Lin <eleanor15x(a)gmail.com> drm/tests: helpers: Fix compiler warning Jakub Kicinski <kuba(a)kernel.org> netdev: avoid CFI problems with sock priv helpers Leon Romanovsky <leon(a)kernel.org> net/mlx5e: Always start IPsec sequence number from 1 Leon Romanovsky <leon(a)kernel.org> net/mlx5e: Rely on reqid in IPsec tunnel mode Leon Romanovsky <leon(a)kernel.org> net/mlx5e: Fix inversion dependency warning while enabling IPsec tunnel Mark Zhang <markzhang(a)nvidia.com> net/mlx5: Clear port select structure when fail to create Chris Mi <cmi(a)nvidia.com> net/mlx5: SF, Fix add port error handling Yishai Hadas <yishaih(a)nvidia.com> net/mlx5: Fix a lockdep warning as part of the write combining test Patrisious Haddad <phaddad(a)nvidia.com> net/mlx5: Fix RDMA TX steering prio Pavel Begunkov <asml.silence(a)gmail.com> net: make page_pool_ref_netmem work with net iovs Kevin Groeneveld <kgroeneveld(a)lenbrook.com> net: fec: handle page_pool_dev_alloc_pages error Sean Anderson <sean.anderson(a)linux.dev> net: xilinx: axienet: Fix IRQ coalescing packet count overflow Dan Carpenter <dan.carpenter(a)linaro.org> nfp: bpf: prevent integer overflow in nfp_bpf_event_output() Viresh Kumar <viresh.kumar(a)linaro.org> cpufreq: Move endif to the end of Kconfig file Kuniyuki Iwashima <kuniyu(a)amazon.com> pfcp: Destroy device along with udp socket's netns dismantle. Kuniyuki Iwashima <kuniyu(a)amazon.com> gtp: Destroy device along with udp socket's netns dismantle. Kuniyuki Iwashima <kuniyu(a)amazon.com> gtp: Use for_each_netdev_rcu() in gtp_genl_dump_pdp(). Qu Wenruo <wqu(a)suse.com> btrfs: add the missing error handling inside get_canonical_dev_path Rafael J. Wysocki <rafael.j.wysocki(a)intel.com> cpuidle: teo: Update documentation after previous changes Karol Kolacinski <karol.kolacinski(a)intel.com> ice: Add correct PHY lane assignment Sergey Temerkhanov <sergey.temerkhanov(a)intel.com> ice: Use ice_adapter for PTP shared data instead of auxdev Sergey Temerkhanov <sergey.temerkhanov(a)intel.com> ice: Add ice_get_ctrl_ptp() wrapper to simplify the code Sergey Temerkhanov <sergey.temerkhanov(a)intel.com> ice: Introduce ice_get_phy_model() wrapper Karol Kolacinski <karol.kolacinski(a)intel.com> ice: Fix ETH56G FC-FEC Rx offset value Karol Kolacinski <karol.kolacinski(a)intel.com> ice: Fix quad registers read on E825 Karol Kolacinski <karol.kolacinski(a)intel.com> ice: Fix E825 initialization Artem Chernyshev <artem.chernyshev(a)red-soft.ru> pktgen: Avoid out-of-bounds access in get_imix_entries Ilya Maximets <i.maximets(a)ovn.org> openvswitch: fix lockup on tx to unregistering netdev with carrier Paul Barker <paul.barker.ct(a)bp.renesas.com> net: ravb: Fix max TX frame size for RZ/V2M Jakub Kicinski <kuba(a)kernel.org> eth: bnxt: always recalculate features after XDP clearing, fix null-deref Michal Luczaj <mhal(a)rbox.co> bpf: Fix bpf_sk_select_reuseport() memory leak Sudheer Kumar Doredla <s-doredla(a)ti.com> net: ethernet: ti: cpsw_ale: Fix cpsw_ale_get_field() Ard Biesheuvel <ardb(a)kernel.org> efi/zboot: Limit compression options to GZIP and ZSTD ------------- Diffstat: Makefile | 4 +- arch/x86/include/asm/special_insns.h | 2 +- arch/x86/kernel/fred.c | 8 +- drivers/acpi/resource.c | 6 +- drivers/block/zram/zram_drv.c | 1 + drivers/cpufreq/Kconfig | 4 +- drivers/cpuidle/governors/teo.c | 91 +++---- drivers/firmware/efi/Kconfig | 4 - drivers/firmware/efi/libstub/Makefile.zboot | 18 +- drivers/gpio/gpio-sim.c | 48 +++- drivers/gpio/gpio-virtuser.c | 49 +++- drivers/gpio/gpio-xilinx.c | 32 +-- drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c | 5 +- drivers/gpu/drm/amd/amdgpu/amdgpu_fw_attestation.c | 4 + drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c | 4 +- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 41 ++- .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c | 25 +- .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c | 4 +- .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.h | 2 +- .../drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c | 2 +- .../amd/display/amdgpu_dm/amdgpu_dm_mst_types.c | 14 +- .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_psr.c | 35 ++- .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_psr.h | 3 +- .../gpu/drm/amd/display/dc/dml/dcn35/dcn35_fpu.c | 4 +- .../gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c | 11 +- drivers/gpu/drm/i915/display/intel_fb.c | 2 +- drivers/gpu/drm/nouveau/nouveau_fence.c | 6 +- drivers/gpu/drm/nouveau/nvkm/engine/disp/mcp77.c | 1 + drivers/gpu/drm/tests/drm_kunit_helpers.c | 3 +- drivers/gpu/drm/v3d/v3d_irq.c | 4 + drivers/gpu/drm/vmwgfx/vmwgfx_bo.c | 3 +- drivers/gpu/drm/vmwgfx/vmwgfx_bo.h | 3 +- drivers/gpu/drm/vmwgfx/vmwgfx_drv.c | 7 +- drivers/gpu/drm/vmwgfx/vmwgfx_gem.c | 1 + drivers/gpu/drm/vmwgfx/vmwgfx_kms.c | 20 +- drivers/gpu/drm/vmwgfx/vmwgfx_shader.c | 7 +- drivers/gpu/drm/vmwgfx/vmwgfx_ttm_buffer.c | 5 +- drivers/gpu/drm/xe/xe_hw_engine.c | 2 +- drivers/gpu/drm/xe/xe_oa.c | 1 + drivers/hwmon/ltc2991.c | 2 +- drivers/hwmon/tmp513.c | 7 +- drivers/i2c/busses/i2c-rcar.c | 20 +- drivers/i2c/i2c-atr.c | 2 +- drivers/i2c/i2c-core-base.c | 1 + drivers/i2c/i2c-slave-testunit.c | 19 +- drivers/i2c/muxes/i2c-demux-pinctrl.c | 4 +- drivers/infiniband/hw/bnxt_re/ib_verbs.c | 1 + drivers/infiniband/hw/bnxt_re/ib_verbs.h | 4 + drivers/infiniband/hw/bnxt_re/qplib_fp.c | 1 + drivers/infiniband/hw/bnxt_re/qplib_fp.h | 1 + drivers/irqchip/irq-gic-v3-its.c | 2 +- drivers/irqchip/irq-gic-v3.c | 2 +- drivers/irqchip/irqchip.c | 4 +- drivers/mtd/spi-nor/core.c | 2 +- drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c | 19 +- drivers/net/ethernet/broadcom/bnxt/bnxt.c | 25 +- drivers/net/ethernet/broadcom/bnxt/bnxt.h | 2 +- drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c | 7 - drivers/net/ethernet/freescale/fec_main.c | 19 +- drivers/net/ethernet/intel/ice/ice.h | 5 + drivers/net/ethernet/intel/ice/ice_adapter.c | 6 + drivers/net/ethernet/intel/ice/ice_adapter.h | 22 +- drivers/net/ethernet/intel/ice/ice_adminq_cmd.h | 1 + drivers/net/ethernet/intel/ice/ice_common.c | 51 ++++ drivers/net/ethernet/intel/ice/ice_common.h | 1 + drivers/net/ethernet/intel/ice/ice_main.c | 6 +- drivers/net/ethernet/intel/ice/ice_ptp.c | 165 +++++++----- drivers/net/ethernet/intel/ice/ice_ptp.h | 9 +- drivers/net/ethernet/intel/ice/ice_ptp_consts.h | 2 +- drivers/net/ethernet/intel/ice/ice_ptp_hw.c | 285 +++++++++++---------- drivers/net/ethernet/intel/ice/ice_ptp_hw.h | 5 + drivers/net/ethernet/intel/ice/ice_type.h | 2 - .../ethernet/mellanox/mlx5/core/en_accel/ipsec.c | 22 +- .../mellanox/mlx5/core/en_accel/ipsec_fs.c | 12 +- .../mellanox/mlx5/core/en_accel/ipsec_offload.c | 11 +- drivers/net/ethernet/mellanox/mlx5/core/fs_core.c | 1 + .../net/ethernet/mellanox/mlx5/core/lag/port_sel.c | 4 +- .../net/ethernet/mellanox/mlx5/core/sf/devlink.c | 1 + drivers/net/ethernet/mellanox/mlx5/core/wc.c | 24 +- drivers/net/ethernet/netronome/nfp/bpf/offload.c | 3 +- drivers/net/ethernet/renesas/ravb_main.c | 1 + drivers/net/ethernet/ti/cpsw_ale.c | 14 +- drivers/net/ethernet/xilinx/xilinx_axienet_main.c | 6 + drivers/net/gtp.c | 26 +- drivers/net/pfcp.c | 15 +- drivers/nvme/target/io-cmd-bdev.c | 2 +- drivers/platform/x86/dell/dell-uart-backlight.c | 5 +- .../x86/intel/speed_select_if/isst_if_common.c | 1 + drivers/platform/x86/intel/tpmi_power_domains.c | 1 + .../x86/lenovo-yoga-tab2-pro-1380-fastcharger.c | 5 +- drivers/pmdomain/imx/imx8mp-blk-ctrl.c | 2 +- drivers/reset/reset-rzg2l-usbphy-ctrl.c | 1 + drivers/ufs/core/ufshcd.c | 9 +- fs/afs/addr_prefs.c | 6 +- fs/btrfs/volumes.c | 4 + fs/cachefiles/daemon.c | 14 +- fs/cachefiles/internal.h | 3 +- fs/cachefiles/security.c | 6 +- fs/file.c | 1 + fs/hfs/super.c | 4 +- fs/iomap/buffered-io.c | 2 +- fs/netfs/read_collect.c | 9 +- fs/proc/vmcore.c | 2 + fs/qnx6/inode.c | 11 +- fs/smb/client/connect.c | 3 +- include/linux/hrtimer.h | 1 + include/linux/poll.h | 10 +- include/linux/pruss_driver.h | 12 +- include/linux/userfaultfd_k.h | 12 + include/net/page_pool/helpers.h | 2 +- include/trace/events/mmflags.h | 63 +++++ kernel/cpu.c | 2 +- kernel/gen_kheaders.sh | 1 + kernel/sched/ext.c | 11 +- kernel/sched/fair.c | 151 ++--------- kernel/time/hrtimer.c | 11 +- kernel/time/timer_migration.c | 43 +++- mm/filemap.c | 2 +- mm/huge_memory.c | 12 + mm/hugetlb.c | 14 +- mm/kmemleak.c | 2 +- mm/mremap.c | 32 ++- mm/vmscan.c | 3 + net/core/filter.c | 30 ++- net/core/netdev-genl-gen.c | 14 +- net/core/pktgen.c | 6 +- net/mac802154/iface.c | 4 + net/mptcp/options.c | 6 +- net/mptcp/protocol.h | 9 +- net/ncsi/internal.h | 2 + net/ncsi/ncsi-manage.c | 16 +- net/ncsi/ncsi-rsp.c | 19 +- net/openvswitch/actions.c | 4 +- net/vmw_vsock/af_vsock.c | 18 ++ net/vmw_vsock/virtio_transport_common.c | 38 ++- net/vmw_vsock/vsock_bpf.c | 9 + security/apparmor/policy.c | 1 + sound/pci/hda/patch_realtek.c | 3 + tools/net/ynl/ynl-gen-c.py | 16 +- tools/testing/selftests/mm/cow.c | 8 +- tools/testing/selftests/net/mptcp/mptcp_connect.c | 43 +++- .../selftests/sched_ext/ddsp_bogus_dsq_fail.bpf.c | 2 +- .../selftests/sched_ext/ddsp_vtimelocal_fail.bpf.c | 4 +- .../testing/selftests/sched_ext/dsp_local_on.bpf.c | 7 +- tools/testing/selftests/sched_ext/dsp_local_on.c | 5 +- .../selftests/sched_ext/enq_select_cpu_fails.bpf.c | 2 +- tools/testing/selftests/sched_ext/exit.bpf.c | 4 +- tools/testing/selftests/sched_ext/maximal.bpf.c | 8 +- .../selftests/sched_ext/select_cpu_dfl.bpf.c | 2 +- .../sched_ext/select_cpu_dfl_nodispatch.bpf.c | 2 +- .../selftests/sched_ext/select_cpu_dispatch.bpf.c | 2 +- .../sched_ext/select_cpu_dispatch_bad_dsq.bpf.c | 2 +- .../sched_ext/select_cpu_dispatch_dbl_dsp.bpf.c | 4 +- .../selftests/sched_ext/select_cpu_vtime.bpf.c | 8 +- .../tc-testing/tc-tests/filters/flow.json | 4 +- tools/testing/shared/linux/maple_tree.h | 2 +- tools/testing/vma/linux/atomic.h | 2 +- 157 files changed, 1345 insertions(+), 766 deletions(-)

1 month, 2 weeks

18
145
0 0

[PATCH v2 1/2] drm/i915/gem: Allow EXEC_CAPTURE on recoverable contexts on DG1

by Ville Syrjala

From: Ville Syrjälä <ville.syrjala(a)linux.intel.com> The intel-media-driver is currently broken on DG1 because it uses EXEC_CAPTURE with recovarable contexts. Relax the check to allow that. I've also submitted a fix for the intel-media-driver: https://github.com/intel/media-driver/pull/1920 Cc: stable(a)vger.kernel.org Cc: Matthew Auld <matthew.auld(a)intel.com> Cc: Thomas Hellström <thomas.hellstrom(a)linux.intel.com> Testcase: igt/gem_exec_capture/capture-invisible Fixes: 71b1669ea9bd ("drm/i915/uapi: tweak error capture on recoverable contexts") Reviewed-by: Andi Shyti <andi.shyti(a)linux.intel.com> Signed-off-by: Ville Syrjälä <ville.syrjala(a)linux.intel.com> --- drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c index ca7e9216934a..ea9d5063ce78 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c @@ -2013,7 +2013,7 @@ static int eb_capture_stage(struct i915_execbuffer *eb) continue; if (i915_gem_context_is_recoverable(eb->gem_context) && - (IS_DGFX(eb->i915) || GRAPHICS_VER_FULL(eb->i915) > IP_VER(12, 0))) + GRAPHICS_VER_FULL(eb->i915) > IP_VER(12, 10)) return -EINVAL; for_each_batch_create_order(eb, j) { -- 2.49.0

1 month, 2 weeks

4
4
0 0

[PATCH v3] usb: hub: lack of clearing xHC resources

by Pawel Laszczak

The xHC resources allocated for USB devices are not released in correct order after resuming in case when while suspend device was reconnected. This issue has been detected during the fallowing scenario: - connect hub HS to root port - connect LS/FS device to hub port - wait for enumeration to finish - force host to suspend - reconnect hub attached to root port - wake host For this scenario during enumeration of USB LS/FS device the Cadence xHC reports completion error code for xHC commands because the xHC resources used for devices has not been properly released. XHCI specification doesn't mention that device can be reset in any order so, we should not treat this issue as Cadence xHC controller bug. Similar as during disconnecting in this case the device resources should be cleared starting form the last usb device in tree toward the root hub. To fix this issue usbcore driver should call hcd->driver->reset_device for all USB devices connected to hub which was reconnected while suspending. Fixes: 3d82904559f4 ("usb: cdnsp: cdns3 Add main part of Cadence USBSSP DRD Driver") cc: <stable(a)vger.kernel.org> Signed-off-by: Pawel Laszczak <pawell(a)cadence.com> --- Changelog: v3: - Changed patch title - Corrected typo - Moved hub_hc_release_resources above mutex_lock(hcd->address0_mutex) v2: - Replaced disconnection procedure with releasing only the xHC resources drivers/usb/core/hub.c | 33 +++++++++++++++++++++++++++++++++ 1 file changed, 33 insertions(+) diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c index a76bb50b6202..dcba4281ea48 100644 --- a/drivers/usb/core/hub.c +++ b/drivers/usb/core/hub.c @@ -6065,6 +6065,36 @@ void usb_hub_cleanup(void) usb_deregister(&hub_driver); } /* usb_hub_cleanup() */ +/** + * hub_hc_release_resources - clear resources used by host controller + * @udev: pointer to device being released + * + * Context: task context, might sleep + * + * Function releases the host controller resources in correct order before + * making any operation on resuming usb device. The host controller resources + * allocated for devices in tree should be released starting from the last + * usb device in tree toward the root hub. This function is used only during + * resuming device when usb device require reinitialization – that is, when + * flag udev->reset_resume is set. + * + * This call is synchronous, and may not be used in an interrupt context. + */ +static void hub_hc_release_resources(struct usb_device *udev) +{ + struct usb_hub *hub = usb_hub_to_struct_hub(udev); + struct usb_hcd *hcd = bus_to_hcd(udev->bus); + int i; + + /* Release up resources for all children before this device */ + for (i = 0; i < udev->maxchild; i++) + if (hub->ports[i]->child) + hub_hc_release_resources(hub->ports[i]->child); + + if (hcd->driver->reset_device) + hcd->driver->reset_device(hcd, udev); +} + /** * usb_reset_and_verify_device - perform a USB port reset to reinitialize a device * @udev: device to reset (not in SUSPENDED or NOTATTACHED state) @@ -6129,6 +6159,9 @@ static int usb_reset_and_verify_device(struct usb_device *udev) bos = udev->bos; udev->bos = NULL; + if (udev->reset_resume) + hub_hc_release_resources(udev); + mutex_lock(hcd->address0_mutex); for (i = 0; i < PORT_INIT_TRIES; ++i) { -- 2.43.0

1 month, 2 weeks

4
5
0 0

FAILED: patch "[PATCH] hrtimers: Force migrate away hrtimers queued after" 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 53dac345395c0d2493cbc2f4c85fe38aef5b63f5 # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025021052-avenging-aflutter-192c@gregkh' --subject-prefix 'PATCH 6.6.y' HEAD^.. Possible dependencies: thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 53dac345395c0d2493cbc2f4c85fe38aef5b63f5 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker <frederic(a)kernel.org> Date: Sat, 18 Jan 2025 00:24:33 +0100 Subject: [PATCH] hrtimers: Force migrate away hrtimers queued after CPUHP_AP_HRTIMERS_DYING hrtimers are migrated away from the dying CPU to any online target at the CPUHP_AP_HRTIMERS_DYING stage in order not to delay bandwidth timers handling tasks involved in the CPU hotplug forward progress. However wakeups can still be performed by the outgoing CPU after CPUHP_AP_HRTIMERS_DYING. Those can result again in bandwidth timers being armed. Depending on several considerations (crystal ball power management based election, earliest timer already enqueued, timer migration enabled or not), the target may eventually be the current CPU even if offline. If that happens, the timer is eventually ignored. The most notable example is RCU which had to deal with each and every of those wake-ups by deferring them to an online CPU, along with related workarounds: _ e787644caf76 (rcu: Defer RCU kthreads wakeup when CPU is dying) _ 9139f93209d1 (rcu/nocb: Fix RT throttling hrtimer armed from offline CPU) _ f7345ccc62a4 (rcu/nocb: Fix rcuog wake-up from offline softirq) The problem isn't confined to RCU though as the stop machine kthread (which runs CPUHP_AP_HRTIMERS_DYING) reports its completion at the end of its work through cpu_stop_signal_done() and performs a wake up that eventually arms the deadline server timer: WARNING: CPU: 94 PID: 588 at kernel/time/hrtimer.c:1086 hrtimer_start_range_ns+0x289/0x2d0 CPU: 94 UID: 0 PID: 588 Comm: migration/94 Not tainted Stopper: multi_cpu_stop+0x0/0x120 <- stop_machine_cpuslocked+0x66/0xc0 RIP: 0010:hrtimer_start_range_ns+0x289/0x2d0 Call Trace: <TASK> start_dl_timer enqueue_dl_entity dl_server_start enqueue_task_fair enqueue_task ttwu_do_activate try_to_wake_up complete cpu_stopper_thread Instead of providing yet another bandaid to work around the situation, fix it in the hrtimers infrastructure instead: always migrate away a timer to an online target whenever it is enqueued from an offline CPU. This will also allow to revert all the above RCU disgraceful hacks. Fixes: 5c0930ccaad5 ("hrtimers: Push pending hrtimers away from outgoing CPU earlier") Reported-by: Vlad Poenaru <vlad.wing(a)gmail.com> Reported-by: Usama Arif <usamaarif642(a)gmail.com> Signed-off-by: Frederic Weisbecker <frederic(a)kernel.org> Signed-off-by: Paul E. McKenney <paulmck(a)kernel.org> Signed-off-by: Thomas Gleixner <tglx(a)linutronix.de> Cc: stable(a)vger.kernel.org Tested-by: Paul E. McKenney <paulmck(a)kernel.org> Link: https://lore.kernel.org/all/20250117232433.24027-1-frederic@kernel.org Closes: 20241213203739.1519801-1-usamaarif642(a)gmail.com diff --git a/include/linux/hrtimer_defs.h b/include/linux/hrtimer_defs.h index c3b4b7ed7c16..84a5045f80f3 100644 --- a/include/linux/hrtimer_defs.h +++ b/include/linux/hrtimer_defs.h @@ -125,6 +125,7 @@ struct hrtimer_cpu_base { ktime_t softirq_expires_next; struct hrtimer *softirq_next_timer; struct hrtimer_clock_base clock_base[HRTIMER_MAX_CLOCK_BASES]; + call_single_data_t csd; } ____cacheline_aligned; diff --git a/kernel/time/hrtimer.c b/kernel/time/hrtimer.c index 4fb81f8c6f1c..deb1aa32814e 100644 --- a/kernel/time/hrtimer.c +++ b/kernel/time/hrtimer.c @@ -58,6 +58,8 @@ #define HRTIMER_ACTIVE_SOFT (HRTIMER_ACTIVE_HARD << MASK_SHIFT) #define HRTIMER_ACTIVE_ALL (HRTIMER_ACTIVE_SOFT | HRTIMER_ACTIVE_HARD) +static void retrigger_next_event(void *arg); + /* * The timer bases: * @@ -111,7 +113,8 @@ DEFINE_PER_CPU(struct hrtimer_cpu_base, hrtimer_bases) = .clockid = CLOCK_TAI, .get_time = &ktime_get_clocktai, }, - } + }, + .csd = CSD_INIT(retrigger_next_event, NULL) }; static const int hrtimer_clock_to_base_table[MAX_CLOCKS] = { @@ -124,6 +127,14 @@ static const int hrtimer_clock_to_base_table[MAX_CLOCKS] = { [CLOCK_TAI] = HRTIMER_BASE_TAI, }; +static inline bool hrtimer_base_is_online(struct hrtimer_cpu_base *base) +{ + if (!IS_ENABLED(CONFIG_HOTPLUG_CPU)) + return true; + else + return likely(base->online); +} + /* * Functions and macros which are different for UP/SMP systems are kept in a * single place @@ -178,27 +189,54 @@ struct hrtimer_clock_base *lock_hrtimer_base(const struct hrtimer *timer, } /* - * We do not migrate the timer when it is expiring before the next - * event on the target cpu. When high resolution is enabled, we cannot - * reprogram the target cpu hardware and we would cause it to fire - * late. To keep it simple, we handle the high resolution enabled and - * disabled case similar. + * Check if the elected target is suitable considering its next + * event and the hotplug state of the current CPU. + * + * If the elected target is remote and its next event is after the timer + * to queue, then a remote reprogram is necessary. However there is no + * guarantee the IPI handling the operation would arrive in time to meet + * the high resolution deadline. In this case the local CPU becomes a + * preferred target, unless it is offline. + * + * High and low resolution modes are handled the same way for simplicity. * * Called with cpu_base->lock of target cpu held. */ -static int -hrtimer_check_target(struct hrtimer *timer, struct hrtimer_clock_base *new_base) +static bool hrtimer_suitable_target(struct hrtimer *timer, struct hrtimer_clock_base *new_base, + struct hrtimer_cpu_base *new_cpu_base, + struct hrtimer_cpu_base *this_cpu_base) { ktime_t expires; + /* + * The local CPU clockevent can be reprogrammed. Also get_target_base() + * guarantees it is online. + */ + if (new_cpu_base == this_cpu_base) + return true; + + /* + * The offline local CPU can't be the default target if the + * next remote target event is after this timer. Keep the + * elected new base. An IPI will we issued to reprogram + * it as a last resort. + */ + if (!hrtimer_base_is_online(this_cpu_base)) + return true; + expires = ktime_sub(hrtimer_get_expires(timer), new_base->offset); - return expires < new_base->cpu_base->expires_next; + + return expires >= new_base->cpu_base->expires_next; } -static inline -struct hrtimer_cpu_base *get_target_base(struct hrtimer_cpu_base *base, - int pinned) +static inline struct hrtimer_cpu_base *get_target_base(struct hrtimer_cpu_base *base, int pinned) { + if (!hrtimer_base_is_online(base)) { + int cpu = cpumask_any_and(cpu_online_mask, housekeeping_cpumask(HK_TYPE_TIMER)); + + return &per_cpu(hrtimer_bases, cpu); + } + #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) if (static_branch_likely(&timers_migration_enabled) && !pinned) return &per_cpu(hrtimer_bases, get_nohz_timer_target()); @@ -249,8 +287,8 @@ switch_hrtimer_base(struct hrtimer *timer, struct hrtimer_clock_base *base, raw_spin_unlock(&base->cpu_base->lock); raw_spin_lock(&new_base->cpu_base->lock); - if (new_cpu_base != this_cpu_base && - hrtimer_check_target(timer, new_base)) { + if (!hrtimer_suitable_target(timer, new_base, new_cpu_base, + this_cpu_base)) { raw_spin_unlock(&new_base->cpu_base->lock); raw_spin_lock(&base->cpu_base->lock); new_cpu_base = this_cpu_base; @@ -259,8 +297,7 @@ switch_hrtimer_base(struct hrtimer *timer, struct hrtimer_clock_base *base, } WRITE_ONCE(timer->base, new_base); } else { - if (new_cpu_base != this_cpu_base && - hrtimer_check_target(timer, new_base)) { + if (!hrtimer_suitable_target(timer, new_base, new_cpu_base, this_cpu_base)) { new_cpu_base = this_cpu_base; goto again; } @@ -706,8 +743,6 @@ static inline int hrtimer_is_hres_enabled(void) return hrtimer_hres_enabled; } -static void retrigger_next_event(void *arg); - /* * Switch to high resolution mode */ @@ -1195,6 +1230,7 @@ static int __hrtimer_start_range_ns(struct hrtimer *timer, ktime_t tim, u64 delta_ns, const enum hrtimer_mode mode, struct hrtimer_clock_base *base) { + struct hrtimer_cpu_base *this_cpu_base = this_cpu_ptr(&hrtimer_bases); struct hrtimer_clock_base *new_base; bool force_local, first; @@ -1206,9 +1242,15 @@ static int __hrtimer_start_range_ns(struct hrtimer *timer, ktime_t tim, * and enforce reprogramming after it is queued no matter whether * it is the new first expiring timer again or not. */ - force_local = base->cpu_base == this_cpu_ptr(&hrtimer_bases); + force_local = base->cpu_base == this_cpu_base; force_local &= base->cpu_base->next_timer == timer; + /* + * Don't force local queuing if this enqueue happens on a unplugged + * CPU after hrtimer_cpu_dying() has been invoked. + */ + force_local &= this_cpu_base->online; + /* * Remove an active timer from the queue. In case it is not queued * on the current CPU, make sure that remove_hrtimer() updates the @@ -1238,8 +1280,27 @@ static int __hrtimer_start_range_ns(struct hrtimer *timer, ktime_t tim, } first = enqueue_hrtimer(timer, new_base, mode); - if (!force_local) - return first; + if (!force_local) { + /* + * If the current CPU base is online, then the timer is + * never queued on a remote CPU if it would be the first + * expiring timer there. + */ + if (hrtimer_base_is_online(this_cpu_base)) + return first; + + /* + * Timer was enqueued remote because the current base is + * already offline. If the timer is the first to expire, + * kick the remote CPU to reprogram the clock event. + */ + if (first) { + struct hrtimer_cpu_base *new_cpu_base = new_base->cpu_base; + + smp_call_function_single_async(new_cpu_base->cpu, &new_cpu_base->csd); + } + return 0; + } /* * Timer was forced to stay on the current CPU to avoid

1 month, 2 weeks

3
2
0 0

FAILED: patch "[PATCH] hrtimers: Force migrate away hrtimers queued after" failed to apply to 6.1-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 6.1-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-6.1.y git checkout FETCH_HEAD git cherry-pick -x 53dac345395c0d2493cbc2f4c85fe38aef5b63f5 # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025021053-unranked-silt-0282@gregkh' --subject-prefix 'PATCH 6.1.y' HEAD^.. Possible dependencies: thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 53dac345395c0d2493cbc2f4c85fe38aef5b63f5 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker <frederic(a)kernel.org> Date: Sat, 18 Jan 2025 00:24:33 +0100 Subject: [PATCH] hrtimers: Force migrate away hrtimers queued after CPUHP_AP_HRTIMERS_DYING hrtimers are migrated away from the dying CPU to any online target at the CPUHP_AP_HRTIMERS_DYING stage in order not to delay bandwidth timers handling tasks involved in the CPU hotplug forward progress. However wakeups can still be performed by the outgoing CPU after CPUHP_AP_HRTIMERS_DYING. Those can result again in bandwidth timers being armed. Depending on several considerations (crystal ball power management based election, earliest timer already enqueued, timer migration enabled or not), the target may eventually be the current CPU even if offline. If that happens, the timer is eventually ignored. The most notable example is RCU which had to deal with each and every of those wake-ups by deferring them to an online CPU, along with related workarounds: _ e787644caf76 (rcu: Defer RCU kthreads wakeup when CPU is dying) _ 9139f93209d1 (rcu/nocb: Fix RT throttling hrtimer armed from offline CPU) _ f7345ccc62a4 (rcu/nocb: Fix rcuog wake-up from offline softirq) The problem isn't confined to RCU though as the stop machine kthread (which runs CPUHP_AP_HRTIMERS_DYING) reports its completion at the end of its work through cpu_stop_signal_done() and performs a wake up that eventually arms the deadline server timer: WARNING: CPU: 94 PID: 588 at kernel/time/hrtimer.c:1086 hrtimer_start_range_ns+0x289/0x2d0 CPU: 94 UID: 0 PID: 588 Comm: migration/94 Not tainted Stopper: multi_cpu_stop+0x0/0x120 <- stop_machine_cpuslocked+0x66/0xc0 RIP: 0010:hrtimer_start_range_ns+0x289/0x2d0 Call Trace: <TASK> start_dl_timer enqueue_dl_entity dl_server_start enqueue_task_fair enqueue_task ttwu_do_activate try_to_wake_up complete cpu_stopper_thread Instead of providing yet another bandaid to work around the situation, fix it in the hrtimers infrastructure instead: always migrate away a timer to an online target whenever it is enqueued from an offline CPU. This will also allow to revert all the above RCU disgraceful hacks. Fixes: 5c0930ccaad5 ("hrtimers: Push pending hrtimers away from outgoing CPU earlier") Reported-by: Vlad Poenaru <vlad.wing(a)gmail.com> Reported-by: Usama Arif <usamaarif642(a)gmail.com> Signed-off-by: Frederic Weisbecker <frederic(a)kernel.org> Signed-off-by: Paul E. McKenney <paulmck(a)kernel.org> Signed-off-by: Thomas Gleixner <tglx(a)linutronix.de> Cc: stable(a)vger.kernel.org Tested-by: Paul E. McKenney <paulmck(a)kernel.org> Link: https://lore.kernel.org/all/20250117232433.24027-1-frederic@kernel.org Closes: 20241213203739.1519801-1-usamaarif642(a)gmail.com diff --git a/include/linux/hrtimer_defs.h b/include/linux/hrtimer_defs.h index c3b4b7ed7c16..84a5045f80f3 100644 --- a/include/linux/hrtimer_defs.h +++ b/include/linux/hrtimer_defs.h @@ -125,6 +125,7 @@ struct hrtimer_cpu_base { ktime_t softirq_expires_next; struct hrtimer *softirq_next_timer; struct hrtimer_clock_base clock_base[HRTIMER_MAX_CLOCK_BASES]; + call_single_data_t csd; } ____cacheline_aligned; diff --git a/kernel/time/hrtimer.c b/kernel/time/hrtimer.c index 4fb81f8c6f1c..deb1aa32814e 100644 --- a/kernel/time/hrtimer.c +++ b/kernel/time/hrtimer.c @@ -58,6 +58,8 @@ #define HRTIMER_ACTIVE_SOFT (HRTIMER_ACTIVE_HARD << MASK_SHIFT) #define HRTIMER_ACTIVE_ALL (HRTIMER_ACTIVE_SOFT | HRTIMER_ACTIVE_HARD) +static void retrigger_next_event(void *arg); + /* * The timer bases: * @@ -111,7 +113,8 @@ DEFINE_PER_CPU(struct hrtimer_cpu_base, hrtimer_bases) = .clockid = CLOCK_TAI, .get_time = &ktime_get_clocktai, }, - } + }, + .csd = CSD_INIT(retrigger_next_event, NULL) }; static const int hrtimer_clock_to_base_table[MAX_CLOCKS] = { @@ -124,6 +127,14 @@ static const int hrtimer_clock_to_base_table[MAX_CLOCKS] = { [CLOCK_TAI] = HRTIMER_BASE_TAI, }; +static inline bool hrtimer_base_is_online(struct hrtimer_cpu_base *base) +{ + if (!IS_ENABLED(CONFIG_HOTPLUG_CPU)) + return true; + else + return likely(base->online); +} + /* * Functions and macros which are different for UP/SMP systems are kept in a * single place @@ -178,27 +189,54 @@ struct hrtimer_clock_base *lock_hrtimer_base(const struct hrtimer *timer, } /* - * We do not migrate the timer when it is expiring before the next - * event on the target cpu. When high resolution is enabled, we cannot - * reprogram the target cpu hardware and we would cause it to fire - * late. To keep it simple, we handle the high resolution enabled and - * disabled case similar. + * Check if the elected target is suitable considering its next + * event and the hotplug state of the current CPU. + * + * If the elected target is remote and its next event is after the timer + * to queue, then a remote reprogram is necessary. However there is no + * guarantee the IPI handling the operation would arrive in time to meet + * the high resolution deadline. In this case the local CPU becomes a + * preferred target, unless it is offline. + * + * High and low resolution modes are handled the same way for simplicity. * * Called with cpu_base->lock of target cpu held. */ -static int -hrtimer_check_target(struct hrtimer *timer, struct hrtimer_clock_base *new_base) +static bool hrtimer_suitable_target(struct hrtimer *timer, struct hrtimer_clock_base *new_base, + struct hrtimer_cpu_base *new_cpu_base, + struct hrtimer_cpu_base *this_cpu_base) { ktime_t expires; + /* + * The local CPU clockevent can be reprogrammed. Also get_target_base() + * guarantees it is online. + */ + if (new_cpu_base == this_cpu_base) + return true; + + /* + * The offline local CPU can't be the default target if the + * next remote target event is after this timer. Keep the + * elected new base. An IPI will we issued to reprogram + * it as a last resort. + */ + if (!hrtimer_base_is_online(this_cpu_base)) + return true; + expires = ktime_sub(hrtimer_get_expires(timer), new_base->offset); - return expires < new_base->cpu_base->expires_next; + + return expires >= new_base->cpu_base->expires_next; } -static inline -struct hrtimer_cpu_base *get_target_base(struct hrtimer_cpu_base *base, - int pinned) +static inline struct hrtimer_cpu_base *get_target_base(struct hrtimer_cpu_base *base, int pinned) { + if (!hrtimer_base_is_online(base)) { + int cpu = cpumask_any_and(cpu_online_mask, housekeeping_cpumask(HK_TYPE_TIMER)); + + return &per_cpu(hrtimer_bases, cpu); + } + #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) if (static_branch_likely(&timers_migration_enabled) && !pinned) return &per_cpu(hrtimer_bases, get_nohz_timer_target()); @@ -249,8 +287,8 @@ switch_hrtimer_base(struct hrtimer *timer, struct hrtimer_clock_base *base, raw_spin_unlock(&base->cpu_base->lock); raw_spin_lock(&new_base->cpu_base->lock); - if (new_cpu_base != this_cpu_base && - hrtimer_check_target(timer, new_base)) { + if (!hrtimer_suitable_target(timer, new_base, new_cpu_base, + this_cpu_base)) { raw_spin_unlock(&new_base->cpu_base->lock); raw_spin_lock(&base->cpu_base->lock); new_cpu_base = this_cpu_base; @@ -259,8 +297,7 @@ switch_hrtimer_base(struct hrtimer *timer, struct hrtimer_clock_base *base, } WRITE_ONCE(timer->base, new_base); } else { - if (new_cpu_base != this_cpu_base && - hrtimer_check_target(timer, new_base)) { + if (!hrtimer_suitable_target(timer, new_base, new_cpu_base, this_cpu_base)) { new_cpu_base = this_cpu_base; goto again; } @@ -706,8 +743,6 @@ static inline int hrtimer_is_hres_enabled(void) return hrtimer_hres_enabled; } -static void retrigger_next_event(void *arg); - /* * Switch to high resolution mode */ @@ -1195,6 +1230,7 @@ static int __hrtimer_start_range_ns(struct hrtimer *timer, ktime_t tim, u64 delta_ns, const enum hrtimer_mode mode, struct hrtimer_clock_base *base) { + struct hrtimer_cpu_base *this_cpu_base = this_cpu_ptr(&hrtimer_bases); struct hrtimer_clock_base *new_base; bool force_local, first; @@ -1206,9 +1242,15 @@ static int __hrtimer_start_range_ns(struct hrtimer *timer, ktime_t tim, * and enforce reprogramming after it is queued no matter whether * it is the new first expiring timer again or not. */ - force_local = base->cpu_base == this_cpu_ptr(&hrtimer_bases); + force_local = base->cpu_base == this_cpu_base; force_local &= base->cpu_base->next_timer == timer; + /* + * Don't force local queuing if this enqueue happens on a unplugged + * CPU after hrtimer_cpu_dying() has been invoked. + */ + force_local &= this_cpu_base->online; + /* * Remove an active timer from the queue. In case it is not queued * on the current CPU, make sure that remove_hrtimer() updates the @@ -1238,8 +1280,27 @@ static int __hrtimer_start_range_ns(struct hrtimer *timer, ktime_t tim, } first = enqueue_hrtimer(timer, new_base, mode); - if (!force_local) - return first; + if (!force_local) { + /* + * If the current CPU base is online, then the timer is + * never queued on a remote CPU if it would be the first + * expiring timer there. + */ + if (hrtimer_base_is_online(this_cpu_base)) + return first; + + /* + * Timer was enqueued remote because the current base is + * already offline. If the timer is the first to expire, + * kick the remote CPU to reprogram the clock event. + */ + if (first) { + struct hrtimer_cpu_base *new_cpu_base = new_base->cpu_base; + + smp_call_function_single_async(new_cpu_base->cpu, &new_cpu_base->csd); + } + return 0; + } /* * Timer was forced to stay on the current CPU to avoid

1 month, 2 weeks

5
6
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror April 2025