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

FAILED: patch "[PATCH] wifi: mt76: Fix DTS power-limits on little endian systems" 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 38b845e1f9e810869b0a0b69f202b877b7b7fb12 # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025122936-impotence-sandworm-9ea8@gregkh' --subject-prefix 'PATCH 6.1.y' HEAD^.. Possible dependencies: thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 38b845e1f9e810869b0a0b69f202b877b7b7fb12 Mon Sep 17 00:00:00 2001 From: "Sven Eckelmann (Plasma Cloud)" <se(a)simonwunderlich.de> Date: Fri, 26 Sep 2025 11:32:54 +0200 Subject: [PATCH] wifi: mt76: Fix DTS power-limits on little endian systems The power-limits for ru and mcs and stored in the devicetree as bytewise array (often with sizes which are not a multiple of 4). These arrays have a prefix which defines for how many modes a line is applied. This prefix is also only a byte - but the code still tried to fix the endianness of this byte with a be32 operation. As result, loading was mostly failing or was sending completely unexpected values to the firmware. Since the other rates are also stored in the devicetree as bytewise arrays, just drop the u32 access + be32_to_cpu conversion and directly access them as bytes arrays. Cc: stable(a)vger.kernel.org Fixes: 22b980badc0f ("mt76: add functions for parsing rate power limits from DT") Fixes: a9627d992b5e ("mt76: extend DT rate power limits to support 11ax devices") Signed-off-by: Sven Eckelmann (Plasma Cloud) <se(a)simonwunderlich.de> Signed-off-by: Felix Fietkau <nbd(a)nbd.name> diff --git a/drivers/net/wireless/mediatek/mt76/eeprom.c b/drivers/net/wireless/mediatek/mt76/eeprom.c index a987c5e4eff6..6ce8e4af18fe 100644 --- a/drivers/net/wireless/mediatek/mt76/eeprom.c +++ b/drivers/net/wireless/mediatek/mt76/eeprom.c @@ -253,6 +253,19 @@ mt76_get_of_array(struct device_node *np, char *name, size_t *len, int min) return prop->value; } +static const s8 * +mt76_get_of_array_s8(struct device_node *np, char *name, size_t *len, int min) +{ + struct property *prop = of_find_property(np, name, NULL); + + if (!prop || !prop->value || prop->length < min) + return NULL; + + *len = prop->length; + + return prop->value; +} + struct device_node * mt76_find_channel_node(struct device_node *np, struct ieee80211_channel *chan) { @@ -294,7 +307,7 @@ mt76_get_txs_delta(struct device_node *np, u8 nss) } static void -mt76_apply_array_limit(s8 *pwr, size_t pwr_len, const __be32 *data, +mt76_apply_array_limit(s8 *pwr, size_t pwr_len, const s8 *data, s8 target_power, s8 nss_delta, s8 *max_power) { int i; @@ -303,15 +316,14 @@ mt76_apply_array_limit(s8 *pwr, size_t pwr_len, const __be32 *data, return; for (i = 0; i < pwr_len; i++) { - pwr[i] = min_t(s8, target_power, - be32_to_cpu(data[i]) + nss_delta); + pwr[i] = min_t(s8, target_power, data[i] + nss_delta); *max_power = max(*max_power, pwr[i]); } } static void mt76_apply_multi_array_limit(s8 *pwr, size_t pwr_len, s8 pwr_num, - const __be32 *data, size_t len, s8 target_power, + const s8 *data, size_t len, s8 target_power, s8 nss_delta, s8 *max_power) { int i, cur; @@ -319,8 +331,7 @@ mt76_apply_multi_array_limit(s8 *pwr, size_t pwr_len, s8 pwr_num, if (!data) return; - len /= 4; - cur = be32_to_cpu(data[0]); + cur = data[0]; for (i = 0; i < pwr_num; i++) { if (len < pwr_len + 1) break; @@ -335,7 +346,7 @@ mt76_apply_multi_array_limit(s8 *pwr, size_t pwr_len, s8 pwr_num, if (!len) break; - cur = be32_to_cpu(data[0]); + cur = data[0]; } } @@ -346,7 +357,7 @@ s8 mt76_get_rate_power_limits(struct mt76_phy *phy, { struct mt76_dev *dev = phy->dev; struct device_node *np; - const __be32 *val; + const s8 *val; char name[16]; u32 mcs_rates = dev->drv->mcs_rates; u32 ru_rates = ARRAY_SIZE(dest->ru[0]); @@ -392,21 +403,21 @@ s8 mt76_get_rate_power_limits(struct mt76_phy *phy, txs_delta = mt76_get_txs_delta(np, hweight16(phy->chainmask)); - val = mt76_get_of_array(np, "rates-cck", &len, ARRAY_SIZE(dest->cck)); + val = mt76_get_of_array_s8(np, "rates-cck", &len, ARRAY_SIZE(dest->cck)); mt76_apply_array_limit(dest->cck, ARRAY_SIZE(dest->cck), val, target_power, txs_delta, &max_power); - val = mt76_get_of_array(np, "rates-ofdm", - &len, ARRAY_SIZE(dest->ofdm)); + val = mt76_get_of_array_s8(np, "rates-ofdm", + &len, ARRAY_SIZE(dest->ofdm)); mt76_apply_array_limit(dest->ofdm, ARRAY_SIZE(dest->ofdm), val, target_power, txs_delta, &max_power); - val = mt76_get_of_array(np, "rates-mcs", &len, mcs_rates + 1); + val = mt76_get_of_array_s8(np, "rates-mcs", &len, mcs_rates + 1); mt76_apply_multi_array_limit(dest->mcs[0], ARRAY_SIZE(dest->mcs[0]), ARRAY_SIZE(dest->mcs), val, len, target_power, txs_delta, &max_power); - val = mt76_get_of_array(np, "rates-ru", &len, ru_rates + 1); + val = mt76_get_of_array_s8(np, "rates-ru", &len, ru_rates + 1); mt76_apply_multi_array_limit(dest->ru[0], ARRAY_SIZE(dest->ru[0]), ARRAY_SIZE(dest->ru), val, len, target_power, txs_delta, &max_power);

5 hours, 27 minutes

1
0
0 0

FAILED: patch "[PATCH] wifi: mt76: Fix DTS power-limits on little endian systems" 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 38b845e1f9e810869b0a0b69f202b877b7b7fb12 # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025122936-python-fidgeting-f6f4@gregkh' --subject-prefix 'PATCH 6.6.y' HEAD^.. Possible dependencies: thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 38b845e1f9e810869b0a0b69f202b877b7b7fb12 Mon Sep 17 00:00:00 2001 From: "Sven Eckelmann (Plasma Cloud)" <se(a)simonwunderlich.de> Date: Fri, 26 Sep 2025 11:32:54 +0200 Subject: [PATCH] wifi: mt76: Fix DTS power-limits on little endian systems The power-limits for ru and mcs and stored in the devicetree as bytewise array (often with sizes which are not a multiple of 4). These arrays have a prefix which defines for how many modes a line is applied. This prefix is also only a byte - but the code still tried to fix the endianness of this byte with a be32 operation. As result, loading was mostly failing or was sending completely unexpected values to the firmware. Since the other rates are also stored in the devicetree as bytewise arrays, just drop the u32 access + be32_to_cpu conversion and directly access them as bytes arrays. Cc: stable(a)vger.kernel.org Fixes: 22b980badc0f ("mt76: add functions for parsing rate power limits from DT") Fixes: a9627d992b5e ("mt76: extend DT rate power limits to support 11ax devices") Signed-off-by: Sven Eckelmann (Plasma Cloud) <se(a)simonwunderlich.de> Signed-off-by: Felix Fietkau <nbd(a)nbd.name> diff --git a/drivers/net/wireless/mediatek/mt76/eeprom.c b/drivers/net/wireless/mediatek/mt76/eeprom.c index a987c5e4eff6..6ce8e4af18fe 100644 --- a/drivers/net/wireless/mediatek/mt76/eeprom.c +++ b/drivers/net/wireless/mediatek/mt76/eeprom.c @@ -253,6 +253,19 @@ mt76_get_of_array(struct device_node *np, char *name, size_t *len, int min) return prop->value; } +static const s8 * +mt76_get_of_array_s8(struct device_node *np, char *name, size_t *len, int min) +{ + struct property *prop = of_find_property(np, name, NULL); + + if (!prop || !prop->value || prop->length < min) + return NULL; + + *len = prop->length; + + return prop->value; +} + struct device_node * mt76_find_channel_node(struct device_node *np, struct ieee80211_channel *chan) { @@ -294,7 +307,7 @@ mt76_get_txs_delta(struct device_node *np, u8 nss) } static void -mt76_apply_array_limit(s8 *pwr, size_t pwr_len, const __be32 *data, +mt76_apply_array_limit(s8 *pwr, size_t pwr_len, const s8 *data, s8 target_power, s8 nss_delta, s8 *max_power) { int i; @@ -303,15 +316,14 @@ mt76_apply_array_limit(s8 *pwr, size_t pwr_len, const __be32 *data, return; for (i = 0; i < pwr_len; i++) { - pwr[i] = min_t(s8, target_power, - be32_to_cpu(data[i]) + nss_delta); + pwr[i] = min_t(s8, target_power, data[i] + nss_delta); *max_power = max(*max_power, pwr[i]); } } static void mt76_apply_multi_array_limit(s8 *pwr, size_t pwr_len, s8 pwr_num, - const __be32 *data, size_t len, s8 target_power, + const s8 *data, size_t len, s8 target_power, s8 nss_delta, s8 *max_power) { int i, cur; @@ -319,8 +331,7 @@ mt76_apply_multi_array_limit(s8 *pwr, size_t pwr_len, s8 pwr_num, if (!data) return; - len /= 4; - cur = be32_to_cpu(data[0]); + cur = data[0]; for (i = 0; i < pwr_num; i++) { if (len < pwr_len + 1) break; @@ -335,7 +346,7 @@ mt76_apply_multi_array_limit(s8 *pwr, size_t pwr_len, s8 pwr_num, if (!len) break; - cur = be32_to_cpu(data[0]); + cur = data[0]; } } @@ -346,7 +357,7 @@ s8 mt76_get_rate_power_limits(struct mt76_phy *phy, { struct mt76_dev *dev = phy->dev; struct device_node *np; - const __be32 *val; + const s8 *val; char name[16]; u32 mcs_rates = dev->drv->mcs_rates; u32 ru_rates = ARRAY_SIZE(dest->ru[0]); @@ -392,21 +403,21 @@ s8 mt76_get_rate_power_limits(struct mt76_phy *phy, txs_delta = mt76_get_txs_delta(np, hweight16(phy->chainmask)); - val = mt76_get_of_array(np, "rates-cck", &len, ARRAY_SIZE(dest->cck)); + val = mt76_get_of_array_s8(np, "rates-cck", &len, ARRAY_SIZE(dest->cck)); mt76_apply_array_limit(dest->cck, ARRAY_SIZE(dest->cck), val, target_power, txs_delta, &max_power); - val = mt76_get_of_array(np, "rates-ofdm", - &len, ARRAY_SIZE(dest->ofdm)); + val = mt76_get_of_array_s8(np, "rates-ofdm", + &len, ARRAY_SIZE(dest->ofdm)); mt76_apply_array_limit(dest->ofdm, ARRAY_SIZE(dest->ofdm), val, target_power, txs_delta, &max_power); - val = mt76_get_of_array(np, "rates-mcs", &len, mcs_rates + 1); + val = mt76_get_of_array_s8(np, "rates-mcs", &len, mcs_rates + 1); mt76_apply_multi_array_limit(dest->mcs[0], ARRAY_SIZE(dest->mcs[0]), ARRAY_SIZE(dest->mcs), val, len, target_power, txs_delta, &max_power); - val = mt76_get_of_array(np, "rates-ru", &len, ru_rates + 1); + val = mt76_get_of_array_s8(np, "rates-ru", &len, ru_rates + 1); mt76_apply_multi_array_limit(dest->ru[0], ARRAY_SIZE(dest->ru[0]), ARRAY_SIZE(dest->ru), val, len, target_power, txs_delta, &max_power);

5 hours, 27 minutes

1
0
0 0

FAILED: patch "[PATCH] clocksource/drivers/nxp-stm: Fix section mismatches" failed to apply to 6.18-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 6.18-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-6.18.y git checkout FETCH_HEAD git cherry-pick -x b452d2c97eeccbf9c7ac5b3d2d9e80bf6d8a23db # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025122943-snowbound-unpack-801e@gregkh' --subject-prefix 'PATCH 6.18.y' HEAD^.. Possible dependencies: thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From b452d2c97eeccbf9c7ac5b3d2d9e80bf6d8a23db Mon Sep 17 00:00:00 2001 From: Johan Hovold <johan(a)kernel.org> Date: Fri, 17 Oct 2025 07:49:43 +0200 Subject: [PATCH] clocksource/drivers/nxp-stm: Fix section mismatches Platform drivers can be probed after their init sections have been discarded (e.g. on probe deferral or manual rebind through sysfs) so the probe function must not live in init. Device managed resource actions similarly cannot be discarded. The "_probe" suffix of the driver structure name prevents modpost from warning about this so replace it to catch any similar future issues. Fixes: cec32ac75827 ("clocksource/drivers/nxp-timer: Add the System Timer Module for the s32gx platforms") Signed-off-by: Johan Hovold <johan(a)kernel.org> Signed-off-by: Daniel Lezcano <daniel.lezcano(a)linaro.org> Cc: stable(a)vger.kernel.org # 6.16 Cc: Daniel Lezcano <daniel.lezcano(a)linaro.org> Link: https://patch.msgid.link/20251017054943.7195-1-johan@kernel.org diff --git a/drivers/clocksource/timer-nxp-stm.c b/drivers/clocksource/timer-nxp-stm.c index 16d52167e949..c320d764b12e 100644 --- a/drivers/clocksource/timer-nxp-stm.c +++ b/drivers/clocksource/timer-nxp-stm.c @@ -177,15 +177,15 @@ static void nxp_stm_clocksource_resume(struct clocksource *cs) nxp_stm_clocksource_enable(cs); } -static void __init devm_clocksource_unregister(void *data) +static void devm_clocksource_unregister(void *data) { struct stm_timer *stm_timer = data; clocksource_unregister(&stm_timer->cs); } -static int __init nxp_stm_clocksource_init(struct device *dev, struct stm_timer *stm_timer, - const char *name, void __iomem *base, struct clk *clk) +static int nxp_stm_clocksource_init(struct device *dev, struct stm_timer *stm_timer, + const char *name, void __iomem *base, struct clk *clk) { int ret; @@ -296,9 +296,9 @@ static void nxp_stm_clockevent_resume(struct clock_event_device *ced) nxp_stm_module_get(stm_timer); } -static int __init nxp_stm_clockevent_per_cpu_init(struct device *dev, struct stm_timer *stm_timer, - const char *name, void __iomem *base, int irq, - struct clk *clk, int cpu) +static int nxp_stm_clockevent_per_cpu_init(struct device *dev, struct stm_timer *stm_timer, + const char *name, void __iomem *base, int irq, + struct clk *clk, int cpu) { stm_timer->base = base; stm_timer->rate = clk_get_rate(clk); @@ -386,7 +386,7 @@ static irqreturn_t nxp_stm_module_interrupt(int irq, void *dev_id) return IRQ_HANDLED; } -static int __init nxp_stm_timer_probe(struct platform_device *pdev) +static int nxp_stm_timer_probe(struct platform_device *pdev) { struct stm_timer *stm_timer; struct device *dev = &pdev->dev; @@ -482,14 +482,14 @@ static const struct of_device_id nxp_stm_of_match[] = { }; MODULE_DEVICE_TABLE(of, nxp_stm_of_match); -static struct platform_driver nxp_stm_probe = { +static struct platform_driver nxp_stm_driver = { .probe = nxp_stm_timer_probe, .driver = { .name = "nxp-stm", .of_match_table = nxp_stm_of_match, }, }; -module_platform_driver(nxp_stm_probe); +module_platform_driver(nxp_stm_driver); MODULE_DESCRIPTION("NXP System Timer Module driver"); MODULE_LICENSE("GPL");

5 hours, 29 minutes

1
0
0 0

FAILED: patch "[PATCH] sched_ext: Fix incorrect sched_class settings for per-cpu" failed to apply to 6.12-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 6.12-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-6.12.y git checkout FETCH_HEAD git cherry-pick -x 1dd6c84f1c544e552848a8968599220bd464e338 # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025122909-remover-casino-d84a@gregkh' --subject-prefix 'PATCH 6.12.y' HEAD^.. Possible dependencies: thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 1dd6c84f1c544e552848a8968599220bd464e338 Mon Sep 17 00:00:00 2001 From: Zqiang <qiang.zhang(a)linux.dev> Date: Mon, 1 Dec 2025 19:25:40 +0800 Subject: [PATCH] sched_ext: Fix incorrect sched_class settings for per-cpu migration tasks When loading the ebpf scheduler, the tasks in the scx_tasks list will be traversed and invoke __setscheduler_class() to get new sched_class. however, this would also incorrectly set the per-cpu migration task's->sched_class to rt_sched_class, even after unload, the per-cpu migration task's->sched_class remains sched_rt_class. The log for this issue is as follows: ./scx_rustland --stats 1 [ 199.245639][ T630] sched_ext: "rustland" does not implement cgroup cpu.weight [ 199.269213][ T630] sched_ext: BPF scheduler "rustland" enabled 04:25:09 [INFO] RustLand scheduler attached bpftrace -e 'iter:task /strcontains(ctx->task->comm, "migration")/ { printf("%s:%d->%pS\n", ctx->task->comm, ctx->task->pid, ctx->task->sched_class); }' Attaching 1 probe... migration/0:24->rt_sched_class+0x0/0xe0 migration/1:27->rt_sched_class+0x0/0xe0 migration/2:33->rt_sched_class+0x0/0xe0 migration/3:39->rt_sched_class+0x0/0xe0 migration/4:45->rt_sched_class+0x0/0xe0 migration/5:52->rt_sched_class+0x0/0xe0 migration/6:58->rt_sched_class+0x0/0xe0 migration/7:64->rt_sched_class+0x0/0xe0 sched_ext: BPF scheduler "rustland" disabled (unregistered from user space) EXIT: unregistered from user space 04:25:21 [INFO] Unregister RustLand scheduler bpftrace -e 'iter:task /strcontains(ctx->task->comm, "migration")/ { printf("%s:%d->%pS\n", ctx->task->comm, ctx->task->pid, ctx->task->sched_class); }' Attaching 1 probe... migration/0:24->rt_sched_class+0x0/0xe0 migration/1:27->rt_sched_class+0x0/0xe0 migration/2:33->rt_sched_class+0x0/0xe0 migration/3:39->rt_sched_class+0x0/0xe0 migration/4:45->rt_sched_class+0x0/0xe0 migration/5:52->rt_sched_class+0x0/0xe0 migration/6:58->rt_sched_class+0x0/0xe0 migration/7:64->rt_sched_class+0x0/0xe0 This commit therefore generate a new scx_setscheduler_class() and add check for stop_sched_class to replace __setscheduler_class(). Fixes: f0e1a0643a59 ("sched_ext: Implement BPF extensible scheduler class") Cc: stable(a)vger.kernel.org # v6.12+ Signed-off-by: Zqiang <qiang.zhang(a)linux.dev> Reviewed-by: Andrea Righi <arighi(a)nvidia.com> Signed-off-by: Tejun Heo <tj(a)kernel.org> diff --git a/kernel/sched/ext.c b/kernel/sched/ext.c index dca9ca0c1854..b563b8c3fd24 100644 --- a/kernel/sched/ext.c +++ b/kernel/sched/ext.c @@ -248,6 +248,14 @@ static struct scx_dispatch_q *find_user_dsq(struct scx_sched *sch, u64 dsq_id) return rhashtable_lookup(&sch->dsq_hash, &dsq_id, dsq_hash_params); } +static const struct sched_class *scx_setscheduler_class(struct task_struct *p) +{ + if (p->sched_class == &stop_sched_class) + return &stop_sched_class; + + return __setscheduler_class(p->policy, p->prio); +} + /* * scx_kf_mask enforcement. Some kfuncs can only be called from specific SCX * ops. When invoking SCX ops, SCX_CALL_OP[_RET]() should be used to indicate @@ -4241,8 +4249,7 @@ static void scx_disable_workfn(struct kthread_work *work) while ((p = scx_task_iter_next_locked(&sti))) { unsigned int queue_flags = DEQUEUE_SAVE | DEQUEUE_MOVE | DEQUEUE_NOCLOCK; const struct sched_class *old_class = p->sched_class; - const struct sched_class *new_class = - __setscheduler_class(p->policy, p->prio); + const struct sched_class *new_class = scx_setscheduler_class(p); update_rq_clock(task_rq(p)); @@ -5042,8 +5049,7 @@ static int scx_enable(struct sched_ext_ops *ops, struct bpf_link *link) while ((p = scx_task_iter_next_locked(&sti))) { unsigned int queue_flags = DEQUEUE_SAVE | DEQUEUE_MOVE; const struct sched_class *old_class = p->sched_class; - const struct sched_class *new_class = - __setscheduler_class(p->policy, p->prio); + const struct sched_class *new_class = scx_setscheduler_class(p); if (scx_get_task_state(p) != SCX_TASK_READY) continue;

5 hours, 29 minutes

1
0
0 0

FAILED: patch "[PATCH] sched_ext: Fix incorrect sched_class settings for per-cpu" failed to apply to 6.18-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 6.18-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-6.18.y git checkout FETCH_HEAD git cherry-pick -x 1dd6c84f1c544e552848a8968599220bd464e338 # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025122908-magician-unlaced-bef4@gregkh' --subject-prefix 'PATCH 6.18.y' HEAD^.. Possible dependencies: thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 1dd6c84f1c544e552848a8968599220bd464e338 Mon Sep 17 00:00:00 2001 From: Zqiang <qiang.zhang(a)linux.dev> Date: Mon, 1 Dec 2025 19:25:40 +0800 Subject: [PATCH] sched_ext: Fix incorrect sched_class settings for per-cpu migration tasks When loading the ebpf scheduler, the tasks in the scx_tasks list will be traversed and invoke __setscheduler_class() to get new sched_class. however, this would also incorrectly set the per-cpu migration task's->sched_class to rt_sched_class, even after unload, the per-cpu migration task's->sched_class remains sched_rt_class. The log for this issue is as follows: ./scx_rustland --stats 1 [ 199.245639][ T630] sched_ext: "rustland" does not implement cgroup cpu.weight [ 199.269213][ T630] sched_ext: BPF scheduler "rustland" enabled 04:25:09 [INFO] RustLand scheduler attached bpftrace -e 'iter:task /strcontains(ctx->task->comm, "migration")/ { printf("%s:%d->%pS\n", ctx->task->comm, ctx->task->pid, ctx->task->sched_class); }' Attaching 1 probe... migration/0:24->rt_sched_class+0x0/0xe0 migration/1:27->rt_sched_class+0x0/0xe0 migration/2:33->rt_sched_class+0x0/0xe0 migration/3:39->rt_sched_class+0x0/0xe0 migration/4:45->rt_sched_class+0x0/0xe0 migration/5:52->rt_sched_class+0x0/0xe0 migration/6:58->rt_sched_class+0x0/0xe0 migration/7:64->rt_sched_class+0x0/0xe0 sched_ext: BPF scheduler "rustland" disabled (unregistered from user space) EXIT: unregistered from user space 04:25:21 [INFO] Unregister RustLand scheduler bpftrace -e 'iter:task /strcontains(ctx->task->comm, "migration")/ { printf("%s:%d->%pS\n", ctx->task->comm, ctx->task->pid, ctx->task->sched_class); }' Attaching 1 probe... migration/0:24->rt_sched_class+0x0/0xe0 migration/1:27->rt_sched_class+0x0/0xe0 migration/2:33->rt_sched_class+0x0/0xe0 migration/3:39->rt_sched_class+0x0/0xe0 migration/4:45->rt_sched_class+0x0/0xe0 migration/5:52->rt_sched_class+0x0/0xe0 migration/6:58->rt_sched_class+0x0/0xe0 migration/7:64->rt_sched_class+0x0/0xe0 This commit therefore generate a new scx_setscheduler_class() and add check for stop_sched_class to replace __setscheduler_class(). Fixes: f0e1a0643a59 ("sched_ext: Implement BPF extensible scheduler class") Cc: stable(a)vger.kernel.org # v6.12+ Signed-off-by: Zqiang <qiang.zhang(a)linux.dev> Reviewed-by: Andrea Righi <arighi(a)nvidia.com> Signed-off-by: Tejun Heo <tj(a)kernel.org> diff --git a/kernel/sched/ext.c b/kernel/sched/ext.c index dca9ca0c1854..b563b8c3fd24 100644 --- a/kernel/sched/ext.c +++ b/kernel/sched/ext.c @@ -248,6 +248,14 @@ static struct scx_dispatch_q *find_user_dsq(struct scx_sched *sch, u64 dsq_id) return rhashtable_lookup(&sch->dsq_hash, &dsq_id, dsq_hash_params); } +static const struct sched_class *scx_setscheduler_class(struct task_struct *p) +{ + if (p->sched_class == &stop_sched_class) + return &stop_sched_class; + + return __setscheduler_class(p->policy, p->prio); +} + /* * scx_kf_mask enforcement. Some kfuncs can only be called from specific SCX * ops. When invoking SCX ops, SCX_CALL_OP[_RET]() should be used to indicate @@ -4241,8 +4249,7 @@ static void scx_disable_workfn(struct kthread_work *work) while ((p = scx_task_iter_next_locked(&sti))) { unsigned int queue_flags = DEQUEUE_SAVE | DEQUEUE_MOVE | DEQUEUE_NOCLOCK; const struct sched_class *old_class = p->sched_class; - const struct sched_class *new_class = - __setscheduler_class(p->policy, p->prio); + const struct sched_class *new_class = scx_setscheduler_class(p); update_rq_clock(task_rq(p)); @@ -5042,8 +5049,7 @@ static int scx_enable(struct sched_ext_ops *ops, struct bpf_link *link) while ((p = scx_task_iter_next_locked(&sti))) { unsigned int queue_flags = DEQUEUE_SAVE | DEQUEUE_MOVE; const struct sched_class *old_class = p->sched_class; - const struct sched_class *new_class = - __setscheduler_class(p->policy, p->prio); + const struct sched_class *new_class = scx_setscheduler_class(p); if (scx_get_task_state(p) != SCX_TASK_READY) continue;

5 hours, 29 minutes

1
0
0 0

FAILED: patch "[PATCH] sched/eevdf: Fix min_vruntime vs avg_vruntime" 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 79f3f9bedd149ea438aaeb0fb6a083637affe205 # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025122903-sterile-from-4520@gregkh' --subject-prefix 'PATCH 6.6.y' HEAD^.. Possible dependencies: thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 79f3f9bedd149ea438aaeb0fb6a083637affe205 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra <peterz(a)infradead.org> Date: Wed, 2 Apr 2025 20:07:34 +0200 Subject: [PATCH] sched/eevdf: Fix min_vruntime vs avg_vruntime Basically, from the constraint that the sum of lag is zero, you can infer that the 0-lag point is the weighted average of the individual vruntime, which is what we're trying to compute: \Sum w_i * v_i avg = -------------- \Sum w_i Now, since vruntime takes the whole u64 (worse, it wraps), this multiplication term in the numerator is not something we can compute; instead we do the min_vruntime (v0 henceforth) thing like: v_i = (v_i - v0) + v0 This does two things: - it keeps the key: (v_i - v0) 'small'; - it creates a relative 0-point in the modular space. If you do that subtitution and work it all out, you end up with: \Sum w_i * (v_i - v0) avg = --------------------- + v0 \Sum w_i Since you cannot very well track a ratio like that (and not suffer terrible numerical problems) we simpy track the numerator and denominator individually and only perform the division when strictly needed. Notably, the numerator lives in cfs_rq->avg_vruntime and the denominator lives in cfs_rq->avg_load. The one extra 'funny' is that these numbers track the entities in the tree, and current is typically outside of the tree, so avg_vruntime() adds current when needed before doing the division. (vruntime_eligible() elides the division by cross-wise multiplication) Anyway, as mentioned above, we currently use the CFS era min_vruntime for this purpose. However, this thing can only move forward, while the above avg can in fact move backward (when a non-eligible task leaves, the average becomes smaller), this can cause trouble when through happenstance (or construction) these values drift far enough apart to wreck the game. Replace cfs_rq::min_vruntime with cfs_rq::zero_vruntime which is kept near/at avg_vruntime, following its motion. The down-side is that this requires computing the avg more often. Fixes: 147f3efaa241 ("sched/fair: Implement an EEVDF-like scheduling policy") Reported-by: Zicheng Qu <quzicheng(a)huawei.com> Signed-off-by: Peter Zijlstra (Intel) <peterz(a)infradead.org> Link: https://patch.msgid.link/20251106111741.GC4068168@noisy.programming.kicks-a… Cc: stable(a)vger.kernel.org diff --git a/kernel/sched/debug.c b/kernel/sched/debug.c index 02e16b70a790..41caa22e0680 100644 --- a/kernel/sched/debug.c +++ b/kernel/sched/debug.c @@ -796,7 +796,7 @@ static void print_rq(struct seq_file *m, struct rq *rq, int rq_cpu) void print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) { - s64 left_vruntime = -1, min_vruntime, right_vruntime = -1, left_deadline = -1, spread; + s64 left_vruntime = -1, zero_vruntime, right_vruntime = -1, left_deadline = -1, spread; struct sched_entity *last, *first, *root; struct rq *rq = cpu_rq(cpu); unsigned long flags; @@ -819,15 +819,15 @@ void print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) last = __pick_last_entity(cfs_rq); if (last) right_vruntime = last->vruntime; - min_vruntime = cfs_rq->min_vruntime; + zero_vruntime = cfs_rq->zero_vruntime; raw_spin_rq_unlock_irqrestore(rq, flags); SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "left_deadline", SPLIT_NS(left_deadline)); SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "left_vruntime", SPLIT_NS(left_vruntime)); - SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "min_vruntime", - SPLIT_NS(min_vruntime)); + SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "zero_vruntime", + SPLIT_NS(zero_vruntime)); SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "avg_vruntime", SPLIT_NS(avg_vruntime(cfs_rq))); SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "right_vruntime", diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 4a11a832d63e..8d971d48669f 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -554,7 +554,7 @@ static inline bool entity_before(const struct sched_entity *a, static inline s64 entity_key(struct cfs_rq *cfs_rq, struct sched_entity *se) { - return (s64)(se->vruntime - cfs_rq->min_vruntime); + return (s64)(se->vruntime - cfs_rq->zero_vruntime); } #define __node_2_se(node) \ @@ -606,13 +606,13 @@ static inline s64 entity_key(struct cfs_rq *cfs_rq, struct sched_entity *se) * * Which we track using: * - * v0 := cfs_rq->min_vruntime + * v0 := cfs_rq->zero_vruntime * \Sum (v_i - v0) * w_i := cfs_rq->avg_vruntime * \Sum w_i := cfs_rq->avg_load * - * Since min_vruntime is a monotonic increasing variable that closely tracks - * the per-task service, these deltas: (v_i - v), will be in the order of the - * maximal (virtual) lag induced in the system due to quantisation. + * Since zero_vruntime closely tracks the per-task service, these + * deltas: (v_i - v), will be in the order of the maximal (virtual) lag + * induced in the system due to quantisation. * * Also, we use scale_load_down() to reduce the size. * @@ -671,7 +671,7 @@ u64 avg_vruntime(struct cfs_rq *cfs_rq) avg = div_s64(avg, load); } - return cfs_rq->min_vruntime + avg; + return cfs_rq->zero_vruntime + avg; } /* @@ -732,7 +732,7 @@ static int vruntime_eligible(struct cfs_rq *cfs_rq, u64 vruntime) load += weight; } - return avg >= (s64)(vruntime - cfs_rq->min_vruntime) * load; + return avg >= (s64)(vruntime - cfs_rq->zero_vruntime) * load; } int entity_eligible(struct cfs_rq *cfs_rq, struct sched_entity *se) @@ -740,42 +740,14 @@ int entity_eligible(struct cfs_rq *cfs_rq, struct sched_entity *se) return vruntime_eligible(cfs_rq, se->vruntime); } -static u64 __update_min_vruntime(struct cfs_rq *cfs_rq, u64 vruntime) +static void update_zero_vruntime(struct cfs_rq *cfs_rq) { - u64 min_vruntime = cfs_rq->min_vruntime; - /* - * open coded max_vruntime() to allow updating avg_vruntime - */ - s64 delta = (s64)(vruntime - min_vruntime); - if (delta > 0) { - avg_vruntime_update(cfs_rq, delta); - min_vruntime = vruntime; - } - return min_vruntime; -} + u64 vruntime = avg_vruntime(cfs_rq); + s64 delta = (s64)(vruntime - cfs_rq->zero_vruntime); -static void update_min_vruntime(struct cfs_rq *cfs_rq) -{ - struct sched_entity *se = __pick_root_entity(cfs_rq); - struct sched_entity *curr = cfs_rq->curr; - u64 vruntime = cfs_rq->min_vruntime; + avg_vruntime_update(cfs_rq, delta); - if (curr) { - if (curr->on_rq) - vruntime = curr->vruntime; - else - curr = NULL; - } - - if (se) { - if (!curr) - vruntime = se->min_vruntime; - else - vruntime = min_vruntime(vruntime, se->min_vruntime); - } - - /* ensure we never gain time by being placed backwards. */ - cfs_rq->min_vruntime = __update_min_vruntime(cfs_rq, vruntime); + cfs_rq->zero_vruntime = vruntime; } static inline u64 cfs_rq_min_slice(struct cfs_rq *cfs_rq) @@ -848,6 +820,7 @@ RB_DECLARE_CALLBACKS(static, min_vruntime_cb, struct sched_entity, static void __enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se) { avg_vruntime_add(cfs_rq, se); + update_zero_vruntime(cfs_rq); se->min_vruntime = se->vruntime; se->min_slice = se->slice; rb_add_augmented_cached(&se->run_node, &cfs_rq->tasks_timeline, @@ -859,6 +832,7 @@ static void __dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se) rb_erase_augmented_cached(&se->run_node, &cfs_rq->tasks_timeline, &min_vruntime_cb); avg_vruntime_sub(cfs_rq, se); + update_zero_vruntime(cfs_rq); } struct sched_entity *__pick_root_entity(struct cfs_rq *cfs_rq) @@ -1226,7 +1200,6 @@ static void update_curr(struct cfs_rq *cfs_rq) curr->vruntime += calc_delta_fair(delta_exec, curr); resched = update_deadline(cfs_rq, curr); - update_min_vruntime(cfs_rq); if (entity_is_task(curr)) { /* @@ -3808,15 +3781,6 @@ static void reweight_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, if (!curr) __enqueue_entity(cfs_rq, se); cfs_rq->nr_queued++; - - /* - * The entity's vruntime has been adjusted, so let's check - * whether the rq-wide min_vruntime needs updated too. Since - * the calculations above require stable min_vruntime rather - * than up-to-date one, we do the update at the end of the - * reweight process. - */ - update_min_vruntime(cfs_rq); } } @@ -5429,15 +5393,6 @@ dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags) update_cfs_group(se); - /* - * Now advance min_vruntime if @se was the entity holding it back, - * except when: DEQUEUE_SAVE && !DEQUEUE_MOVE, in this case we'll be - * put back on, and if we advance min_vruntime, we'll be placed back - * further than we started -- i.e. we'll be penalized. - */ - if ((flags & (DEQUEUE_SAVE | DEQUEUE_MOVE)) != DEQUEUE_SAVE) - update_min_vruntime(cfs_rq); - if (flags & DEQUEUE_DELAYED) finish_delayed_dequeue_entity(se); @@ -9015,7 +8970,6 @@ static void yield_task_fair(struct rq *rq) if (entity_eligible(cfs_rq, se)) { se->vruntime = se->deadline; se->deadline += calc_delta_fair(se->slice, se); - update_min_vruntime(cfs_rq); } } @@ -13078,23 +13032,6 @@ static inline void task_tick_core(struct rq *rq, struct task_struct *curr) * Which shows that S and s_i transform alike (which makes perfect sense * given that S is basically the (weighted) average of s_i). * - * Then: - * - * x -> s_min := min{s_i} (8) - * - * to obtain: - * - * \Sum_i w_i (s_i - s_min) - * S = s_min + ------------------------ (9) - * \Sum_i w_i - * - * Which already looks familiar, and is the basis for our current - * approximation: - * - * S ~= s_min (10) - * - * Now, obviously, (10) is absolute crap :-), but it sorta works. - * * So the thing to remember is that the above is strictly UP. It is * possible to generalize to multiple runqueues -- however it gets really * yuck when you have to add affinity support, as illustrated by our very @@ -13116,23 +13053,23 @@ static inline void task_tick_core(struct rq *rq, struct task_struct *curr) * Let, for our runqueue 'k': * * T_k = \Sum_i w_i s_i - * W_k = \Sum_i w_i ; for all i of k (11) + * W_k = \Sum_i w_i ; for all i of k (8) * * Then we can write (6) like: * * T_k - * S_k = --- (12) + * S_k = --- (9) * W_k * * From which immediately follows that: * * T_k + T_l - * S_k+l = --------- (13) + * S_k+l = --------- (10) * W_k + W_l * * On which we can define a combined lag: * - * lag_k+l(i) := S_k+l - s_i (14) + * lag_k+l(i) := S_k+l - s_i (11) * * And that gives us the tools to compare tasks across a combined runqueue. * @@ -13143,7 +13080,7 @@ static inline void task_tick_core(struct rq *rq, struct task_struct *curr) * using (7); this only requires storing single 'time'-stamps. * * b) when comparing tasks between 2 runqueues of which one is forced-idle, - * compare the combined lag, per (14). + * compare the combined lag, per (11). * * Now, of course cgroups (I so hate them) make this more interesting in * that a) seems to suggest we need to iterate all cgroup on a CPU at such @@ -13191,12 +13128,11 @@ static inline void task_tick_core(struct rq *rq, struct task_struct *curr) * every tick. This limits the observed divergence due to the work * conservancy. * - * On top of that, we can improve upon things by moving away from our - * horrible (10) hack and moving to (9) and employing (13) here. + * On top of that, we can improve upon things by employing (10) here. */ /* - * se_fi_update - Update the cfs_rq->min_vruntime_fi in a CFS hierarchy if needed. + * se_fi_update - Update the cfs_rq->zero_vruntime_fi in a CFS hierarchy if needed. */ static void se_fi_update(const struct sched_entity *se, unsigned int fi_seq, bool forceidle) @@ -13210,7 +13146,7 @@ static void se_fi_update(const struct sched_entity *se, unsigned int fi_seq, cfs_rq->forceidle_seq = fi_seq; } - cfs_rq->min_vruntime_fi = cfs_rq->min_vruntime; + cfs_rq->zero_vruntime_fi = cfs_rq->zero_vruntime; } } @@ -13263,11 +13199,11 @@ bool cfs_prio_less(const struct task_struct *a, const struct task_struct *b, /* * Find delta after normalizing se's vruntime with its cfs_rq's - * min_vruntime_fi, which would have been updated in prior calls + * zero_vruntime_fi, which would have been updated in prior calls * to se_fi_update(). */ delta = (s64)(sea->vruntime - seb->vruntime) + - (s64)(cfs_rqb->min_vruntime_fi - cfs_rqa->min_vruntime_fi); + (s64)(cfs_rqb->zero_vruntime_fi - cfs_rqa->zero_vruntime_fi); return delta > 0; } @@ -13513,7 +13449,7 @@ static void set_next_task_fair(struct rq *rq, struct task_struct *p, bool first) void init_cfs_rq(struct cfs_rq *cfs_rq) { cfs_rq->tasks_timeline = RB_ROOT_CACHED; - cfs_rq->min_vruntime = (u64)(-(1LL << 20)); + cfs_rq->zero_vruntime = (u64)(-(1LL << 20)); raw_spin_lock_init(&cfs_rq->removed.lock); } diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index 82e74e8ca2ea..5a3cf81c27be 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -681,10 +681,10 @@ struct cfs_rq { s64 avg_vruntime; u64 avg_load; - u64 min_vruntime; + u64 zero_vruntime; #ifdef CONFIG_SCHED_CORE unsigned int forceidle_seq; - u64 min_vruntime_fi; + u64 zero_vruntime_fi; #endif struct rb_root_cached tasks_timeline;

5 hours, 29 minutes

1
0
0 0

FAILED: patch "[PATCH] sched/eevdf: Fix min_vruntime vs avg_vruntime" failed to apply to 6.12-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 6.12-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-6.12.y git checkout FETCH_HEAD git cherry-pick -x 79f3f9bedd149ea438aaeb0fb6a083637affe205 # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025122902-untreated-unbiased-5922@gregkh' --subject-prefix 'PATCH 6.12.y' HEAD^.. Possible dependencies: thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 79f3f9bedd149ea438aaeb0fb6a083637affe205 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra <peterz(a)infradead.org> Date: Wed, 2 Apr 2025 20:07:34 +0200 Subject: [PATCH] sched/eevdf: Fix min_vruntime vs avg_vruntime Basically, from the constraint that the sum of lag is zero, you can infer that the 0-lag point is the weighted average of the individual vruntime, which is what we're trying to compute: \Sum w_i * v_i avg = -------------- \Sum w_i Now, since vruntime takes the whole u64 (worse, it wraps), this multiplication term in the numerator is not something we can compute; instead we do the min_vruntime (v0 henceforth) thing like: v_i = (v_i - v0) + v0 This does two things: - it keeps the key: (v_i - v0) 'small'; - it creates a relative 0-point in the modular space. If you do that subtitution and work it all out, you end up with: \Sum w_i * (v_i - v0) avg = --------------------- + v0 \Sum w_i Since you cannot very well track a ratio like that (and not suffer terrible numerical problems) we simpy track the numerator and denominator individually and only perform the division when strictly needed. Notably, the numerator lives in cfs_rq->avg_vruntime and the denominator lives in cfs_rq->avg_load. The one extra 'funny' is that these numbers track the entities in the tree, and current is typically outside of the tree, so avg_vruntime() adds current when needed before doing the division. (vruntime_eligible() elides the division by cross-wise multiplication) Anyway, as mentioned above, we currently use the CFS era min_vruntime for this purpose. However, this thing can only move forward, while the above avg can in fact move backward (when a non-eligible task leaves, the average becomes smaller), this can cause trouble when through happenstance (or construction) these values drift far enough apart to wreck the game. Replace cfs_rq::min_vruntime with cfs_rq::zero_vruntime which is kept near/at avg_vruntime, following its motion. The down-side is that this requires computing the avg more often. Fixes: 147f3efaa241 ("sched/fair: Implement an EEVDF-like scheduling policy") Reported-by: Zicheng Qu <quzicheng(a)huawei.com> Signed-off-by: Peter Zijlstra (Intel) <peterz(a)infradead.org> Link: https://patch.msgid.link/20251106111741.GC4068168@noisy.programming.kicks-a… Cc: stable(a)vger.kernel.org diff --git a/kernel/sched/debug.c b/kernel/sched/debug.c index 02e16b70a790..41caa22e0680 100644 --- a/kernel/sched/debug.c +++ b/kernel/sched/debug.c @@ -796,7 +796,7 @@ static void print_rq(struct seq_file *m, struct rq *rq, int rq_cpu) void print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) { - s64 left_vruntime = -1, min_vruntime, right_vruntime = -1, left_deadline = -1, spread; + s64 left_vruntime = -1, zero_vruntime, right_vruntime = -1, left_deadline = -1, spread; struct sched_entity *last, *first, *root; struct rq *rq = cpu_rq(cpu); unsigned long flags; @@ -819,15 +819,15 @@ void print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) last = __pick_last_entity(cfs_rq); if (last) right_vruntime = last->vruntime; - min_vruntime = cfs_rq->min_vruntime; + zero_vruntime = cfs_rq->zero_vruntime; raw_spin_rq_unlock_irqrestore(rq, flags); SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "left_deadline", SPLIT_NS(left_deadline)); SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "left_vruntime", SPLIT_NS(left_vruntime)); - SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "min_vruntime", - SPLIT_NS(min_vruntime)); + SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "zero_vruntime", + SPLIT_NS(zero_vruntime)); SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "avg_vruntime", SPLIT_NS(avg_vruntime(cfs_rq))); SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "right_vruntime", diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 4a11a832d63e..8d971d48669f 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -554,7 +554,7 @@ static inline bool entity_before(const struct sched_entity *a, static inline s64 entity_key(struct cfs_rq *cfs_rq, struct sched_entity *se) { - return (s64)(se->vruntime - cfs_rq->min_vruntime); + return (s64)(se->vruntime - cfs_rq->zero_vruntime); } #define __node_2_se(node) \ @@ -606,13 +606,13 @@ static inline s64 entity_key(struct cfs_rq *cfs_rq, struct sched_entity *se) * * Which we track using: * - * v0 := cfs_rq->min_vruntime + * v0 := cfs_rq->zero_vruntime * \Sum (v_i - v0) * w_i := cfs_rq->avg_vruntime * \Sum w_i := cfs_rq->avg_load * - * Since min_vruntime is a monotonic increasing variable that closely tracks - * the per-task service, these deltas: (v_i - v), will be in the order of the - * maximal (virtual) lag induced in the system due to quantisation. + * Since zero_vruntime closely tracks the per-task service, these + * deltas: (v_i - v), will be in the order of the maximal (virtual) lag + * induced in the system due to quantisation. * * Also, we use scale_load_down() to reduce the size. * @@ -671,7 +671,7 @@ u64 avg_vruntime(struct cfs_rq *cfs_rq) avg = div_s64(avg, load); } - return cfs_rq->min_vruntime + avg; + return cfs_rq->zero_vruntime + avg; } /* @@ -732,7 +732,7 @@ static int vruntime_eligible(struct cfs_rq *cfs_rq, u64 vruntime) load += weight; } - return avg >= (s64)(vruntime - cfs_rq->min_vruntime) * load; + return avg >= (s64)(vruntime - cfs_rq->zero_vruntime) * load; } int entity_eligible(struct cfs_rq *cfs_rq, struct sched_entity *se) @@ -740,42 +740,14 @@ int entity_eligible(struct cfs_rq *cfs_rq, struct sched_entity *se) return vruntime_eligible(cfs_rq, se->vruntime); } -static u64 __update_min_vruntime(struct cfs_rq *cfs_rq, u64 vruntime) +static void update_zero_vruntime(struct cfs_rq *cfs_rq) { - u64 min_vruntime = cfs_rq->min_vruntime; - /* - * open coded max_vruntime() to allow updating avg_vruntime - */ - s64 delta = (s64)(vruntime - min_vruntime); - if (delta > 0) { - avg_vruntime_update(cfs_rq, delta); - min_vruntime = vruntime; - } - return min_vruntime; -} + u64 vruntime = avg_vruntime(cfs_rq); + s64 delta = (s64)(vruntime - cfs_rq->zero_vruntime); -static void update_min_vruntime(struct cfs_rq *cfs_rq) -{ - struct sched_entity *se = __pick_root_entity(cfs_rq); - struct sched_entity *curr = cfs_rq->curr; - u64 vruntime = cfs_rq->min_vruntime; + avg_vruntime_update(cfs_rq, delta); - if (curr) { - if (curr->on_rq) - vruntime = curr->vruntime; - else - curr = NULL; - } - - if (se) { - if (!curr) - vruntime = se->min_vruntime; - else - vruntime = min_vruntime(vruntime, se->min_vruntime); - } - - /* ensure we never gain time by being placed backwards. */ - cfs_rq->min_vruntime = __update_min_vruntime(cfs_rq, vruntime); + cfs_rq->zero_vruntime = vruntime; } static inline u64 cfs_rq_min_slice(struct cfs_rq *cfs_rq) @@ -848,6 +820,7 @@ RB_DECLARE_CALLBACKS(static, min_vruntime_cb, struct sched_entity, static void __enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se) { avg_vruntime_add(cfs_rq, se); + update_zero_vruntime(cfs_rq); se->min_vruntime = se->vruntime; se->min_slice = se->slice; rb_add_augmented_cached(&se->run_node, &cfs_rq->tasks_timeline, @@ -859,6 +832,7 @@ static void __dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se) rb_erase_augmented_cached(&se->run_node, &cfs_rq->tasks_timeline, &min_vruntime_cb); avg_vruntime_sub(cfs_rq, se); + update_zero_vruntime(cfs_rq); } struct sched_entity *__pick_root_entity(struct cfs_rq *cfs_rq) @@ -1226,7 +1200,6 @@ static void update_curr(struct cfs_rq *cfs_rq) curr->vruntime += calc_delta_fair(delta_exec, curr); resched = update_deadline(cfs_rq, curr); - update_min_vruntime(cfs_rq); if (entity_is_task(curr)) { /* @@ -3808,15 +3781,6 @@ static void reweight_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, if (!curr) __enqueue_entity(cfs_rq, se); cfs_rq->nr_queued++; - - /* - * The entity's vruntime has been adjusted, so let's check - * whether the rq-wide min_vruntime needs updated too. Since - * the calculations above require stable min_vruntime rather - * than up-to-date one, we do the update at the end of the - * reweight process. - */ - update_min_vruntime(cfs_rq); } } @@ -5429,15 +5393,6 @@ dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags) update_cfs_group(se); - /* - * Now advance min_vruntime if @se was the entity holding it back, - * except when: DEQUEUE_SAVE && !DEQUEUE_MOVE, in this case we'll be - * put back on, and if we advance min_vruntime, we'll be placed back - * further than we started -- i.e. we'll be penalized. - */ - if ((flags & (DEQUEUE_SAVE | DEQUEUE_MOVE)) != DEQUEUE_SAVE) - update_min_vruntime(cfs_rq); - if (flags & DEQUEUE_DELAYED) finish_delayed_dequeue_entity(se); @@ -9015,7 +8970,6 @@ static void yield_task_fair(struct rq *rq) if (entity_eligible(cfs_rq, se)) { se->vruntime = se->deadline; se->deadline += calc_delta_fair(se->slice, se); - update_min_vruntime(cfs_rq); } } @@ -13078,23 +13032,6 @@ static inline void task_tick_core(struct rq *rq, struct task_struct *curr) * Which shows that S and s_i transform alike (which makes perfect sense * given that S is basically the (weighted) average of s_i). * - * Then: - * - * x -> s_min := min{s_i} (8) - * - * to obtain: - * - * \Sum_i w_i (s_i - s_min) - * S = s_min + ------------------------ (9) - * \Sum_i w_i - * - * Which already looks familiar, and is the basis for our current - * approximation: - * - * S ~= s_min (10) - * - * Now, obviously, (10) is absolute crap :-), but it sorta works. - * * So the thing to remember is that the above is strictly UP. It is * possible to generalize to multiple runqueues -- however it gets really * yuck when you have to add affinity support, as illustrated by our very @@ -13116,23 +13053,23 @@ static inline void task_tick_core(struct rq *rq, struct task_struct *curr) * Let, for our runqueue 'k': * * T_k = \Sum_i w_i s_i - * W_k = \Sum_i w_i ; for all i of k (11) + * W_k = \Sum_i w_i ; for all i of k (8) * * Then we can write (6) like: * * T_k - * S_k = --- (12) + * S_k = --- (9) * W_k * * From which immediately follows that: * * T_k + T_l - * S_k+l = --------- (13) + * S_k+l = --------- (10) * W_k + W_l * * On which we can define a combined lag: * - * lag_k+l(i) := S_k+l - s_i (14) + * lag_k+l(i) := S_k+l - s_i (11) * * And that gives us the tools to compare tasks across a combined runqueue. * @@ -13143,7 +13080,7 @@ static inline void task_tick_core(struct rq *rq, struct task_struct *curr) * using (7); this only requires storing single 'time'-stamps. * * b) when comparing tasks between 2 runqueues of which one is forced-idle, - * compare the combined lag, per (14). + * compare the combined lag, per (11). * * Now, of course cgroups (I so hate them) make this more interesting in * that a) seems to suggest we need to iterate all cgroup on a CPU at such @@ -13191,12 +13128,11 @@ static inline void task_tick_core(struct rq *rq, struct task_struct *curr) * every tick. This limits the observed divergence due to the work * conservancy. * - * On top of that, we can improve upon things by moving away from our - * horrible (10) hack and moving to (9) and employing (13) here. + * On top of that, we can improve upon things by employing (10) here. */ /* - * se_fi_update - Update the cfs_rq->min_vruntime_fi in a CFS hierarchy if needed. + * se_fi_update - Update the cfs_rq->zero_vruntime_fi in a CFS hierarchy if needed. */ static void se_fi_update(const struct sched_entity *se, unsigned int fi_seq, bool forceidle) @@ -13210,7 +13146,7 @@ static void se_fi_update(const struct sched_entity *se, unsigned int fi_seq, cfs_rq->forceidle_seq = fi_seq; } - cfs_rq->min_vruntime_fi = cfs_rq->min_vruntime; + cfs_rq->zero_vruntime_fi = cfs_rq->zero_vruntime; } } @@ -13263,11 +13199,11 @@ bool cfs_prio_less(const struct task_struct *a, const struct task_struct *b, /* * Find delta after normalizing se's vruntime with its cfs_rq's - * min_vruntime_fi, which would have been updated in prior calls + * zero_vruntime_fi, which would have been updated in prior calls * to se_fi_update(). */ delta = (s64)(sea->vruntime - seb->vruntime) + - (s64)(cfs_rqb->min_vruntime_fi - cfs_rqa->min_vruntime_fi); + (s64)(cfs_rqb->zero_vruntime_fi - cfs_rqa->zero_vruntime_fi); return delta > 0; } @@ -13513,7 +13449,7 @@ static void set_next_task_fair(struct rq *rq, struct task_struct *p, bool first) void init_cfs_rq(struct cfs_rq *cfs_rq) { cfs_rq->tasks_timeline = RB_ROOT_CACHED; - cfs_rq->min_vruntime = (u64)(-(1LL << 20)); + cfs_rq->zero_vruntime = (u64)(-(1LL << 20)); raw_spin_lock_init(&cfs_rq->removed.lock); } diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index 82e74e8ca2ea..5a3cf81c27be 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -681,10 +681,10 @@ struct cfs_rq { s64 avg_vruntime; u64 avg_load; - u64 min_vruntime; + u64 zero_vruntime; #ifdef CONFIG_SCHED_CORE unsigned int forceidle_seq; - u64 min_vruntime_fi; + u64 zero_vruntime_fi; #endif struct rb_root_cached tasks_timeline;

5 hours, 29 minutes

1
0
0 0

FAILED: patch "[PATCH] sched/eevdf: Fix min_vruntime vs avg_vruntime" failed to apply to 6.18-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 6.18-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-6.18.y git checkout FETCH_HEAD git cherry-pick -x 79f3f9bedd149ea438aaeb0fb6a083637affe205 # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025122900-pug-tartly-0d84@gregkh' --subject-prefix 'PATCH 6.18.y' HEAD^.. Possible dependencies: thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 79f3f9bedd149ea438aaeb0fb6a083637affe205 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra <peterz(a)infradead.org> Date: Wed, 2 Apr 2025 20:07:34 +0200 Subject: [PATCH] sched/eevdf: Fix min_vruntime vs avg_vruntime Basically, from the constraint that the sum of lag is zero, you can infer that the 0-lag point is the weighted average of the individual vruntime, which is what we're trying to compute: \Sum w_i * v_i avg = -------------- \Sum w_i Now, since vruntime takes the whole u64 (worse, it wraps), this multiplication term in the numerator is not something we can compute; instead we do the min_vruntime (v0 henceforth) thing like: v_i = (v_i - v0) + v0 This does two things: - it keeps the key: (v_i - v0) 'small'; - it creates a relative 0-point in the modular space. If you do that subtitution and work it all out, you end up with: \Sum w_i * (v_i - v0) avg = --------------------- + v0 \Sum w_i Since you cannot very well track a ratio like that (and not suffer terrible numerical problems) we simpy track the numerator and denominator individually and only perform the division when strictly needed. Notably, the numerator lives in cfs_rq->avg_vruntime and the denominator lives in cfs_rq->avg_load. The one extra 'funny' is that these numbers track the entities in the tree, and current is typically outside of the tree, so avg_vruntime() adds current when needed before doing the division. (vruntime_eligible() elides the division by cross-wise multiplication) Anyway, as mentioned above, we currently use the CFS era min_vruntime for this purpose. However, this thing can only move forward, while the above avg can in fact move backward (when a non-eligible task leaves, the average becomes smaller), this can cause trouble when through happenstance (or construction) these values drift far enough apart to wreck the game. Replace cfs_rq::min_vruntime with cfs_rq::zero_vruntime which is kept near/at avg_vruntime, following its motion. The down-side is that this requires computing the avg more often. Fixes: 147f3efaa241 ("sched/fair: Implement an EEVDF-like scheduling policy") Reported-by: Zicheng Qu <quzicheng(a)huawei.com> Signed-off-by: Peter Zijlstra (Intel) <peterz(a)infradead.org> Link: https://patch.msgid.link/20251106111741.GC4068168@noisy.programming.kicks-a… Cc: stable(a)vger.kernel.org diff --git a/kernel/sched/debug.c b/kernel/sched/debug.c index 02e16b70a790..41caa22e0680 100644 --- a/kernel/sched/debug.c +++ b/kernel/sched/debug.c @@ -796,7 +796,7 @@ static void print_rq(struct seq_file *m, struct rq *rq, int rq_cpu) void print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) { - s64 left_vruntime = -1, min_vruntime, right_vruntime = -1, left_deadline = -1, spread; + s64 left_vruntime = -1, zero_vruntime, right_vruntime = -1, left_deadline = -1, spread; struct sched_entity *last, *first, *root; struct rq *rq = cpu_rq(cpu); unsigned long flags; @@ -819,15 +819,15 @@ void print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) last = __pick_last_entity(cfs_rq); if (last) right_vruntime = last->vruntime; - min_vruntime = cfs_rq->min_vruntime; + zero_vruntime = cfs_rq->zero_vruntime; raw_spin_rq_unlock_irqrestore(rq, flags); SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "left_deadline", SPLIT_NS(left_deadline)); SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "left_vruntime", SPLIT_NS(left_vruntime)); - SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "min_vruntime", - SPLIT_NS(min_vruntime)); + SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "zero_vruntime", + SPLIT_NS(zero_vruntime)); SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "avg_vruntime", SPLIT_NS(avg_vruntime(cfs_rq))); SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "right_vruntime", diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 4a11a832d63e..8d971d48669f 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -554,7 +554,7 @@ static inline bool entity_before(const struct sched_entity *a, static inline s64 entity_key(struct cfs_rq *cfs_rq, struct sched_entity *se) { - return (s64)(se->vruntime - cfs_rq->min_vruntime); + return (s64)(se->vruntime - cfs_rq->zero_vruntime); } #define __node_2_se(node) \ @@ -606,13 +606,13 @@ static inline s64 entity_key(struct cfs_rq *cfs_rq, struct sched_entity *se) * * Which we track using: * - * v0 := cfs_rq->min_vruntime + * v0 := cfs_rq->zero_vruntime * \Sum (v_i - v0) * w_i := cfs_rq->avg_vruntime * \Sum w_i := cfs_rq->avg_load * - * Since min_vruntime is a monotonic increasing variable that closely tracks - * the per-task service, these deltas: (v_i - v), will be in the order of the - * maximal (virtual) lag induced in the system due to quantisation. + * Since zero_vruntime closely tracks the per-task service, these + * deltas: (v_i - v), will be in the order of the maximal (virtual) lag + * induced in the system due to quantisation. * * Also, we use scale_load_down() to reduce the size. * @@ -671,7 +671,7 @@ u64 avg_vruntime(struct cfs_rq *cfs_rq) avg = div_s64(avg, load); } - return cfs_rq->min_vruntime + avg; + return cfs_rq->zero_vruntime + avg; } /* @@ -732,7 +732,7 @@ static int vruntime_eligible(struct cfs_rq *cfs_rq, u64 vruntime) load += weight; } - return avg >= (s64)(vruntime - cfs_rq->min_vruntime) * load; + return avg >= (s64)(vruntime - cfs_rq->zero_vruntime) * load; } int entity_eligible(struct cfs_rq *cfs_rq, struct sched_entity *se) @@ -740,42 +740,14 @@ int entity_eligible(struct cfs_rq *cfs_rq, struct sched_entity *se) return vruntime_eligible(cfs_rq, se->vruntime); } -static u64 __update_min_vruntime(struct cfs_rq *cfs_rq, u64 vruntime) +static void update_zero_vruntime(struct cfs_rq *cfs_rq) { - u64 min_vruntime = cfs_rq->min_vruntime; - /* - * open coded max_vruntime() to allow updating avg_vruntime - */ - s64 delta = (s64)(vruntime - min_vruntime); - if (delta > 0) { - avg_vruntime_update(cfs_rq, delta); - min_vruntime = vruntime; - } - return min_vruntime; -} + u64 vruntime = avg_vruntime(cfs_rq); + s64 delta = (s64)(vruntime - cfs_rq->zero_vruntime); -static void update_min_vruntime(struct cfs_rq *cfs_rq) -{ - struct sched_entity *se = __pick_root_entity(cfs_rq); - struct sched_entity *curr = cfs_rq->curr; - u64 vruntime = cfs_rq->min_vruntime; + avg_vruntime_update(cfs_rq, delta); - if (curr) { - if (curr->on_rq) - vruntime = curr->vruntime; - else - curr = NULL; - } - - if (se) { - if (!curr) - vruntime = se->min_vruntime; - else - vruntime = min_vruntime(vruntime, se->min_vruntime); - } - - /* ensure we never gain time by being placed backwards. */ - cfs_rq->min_vruntime = __update_min_vruntime(cfs_rq, vruntime); + cfs_rq->zero_vruntime = vruntime; } static inline u64 cfs_rq_min_slice(struct cfs_rq *cfs_rq) @@ -848,6 +820,7 @@ RB_DECLARE_CALLBACKS(static, min_vruntime_cb, struct sched_entity, static void __enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se) { avg_vruntime_add(cfs_rq, se); + update_zero_vruntime(cfs_rq); se->min_vruntime = se->vruntime; se->min_slice = se->slice; rb_add_augmented_cached(&se->run_node, &cfs_rq->tasks_timeline, @@ -859,6 +832,7 @@ static void __dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se) rb_erase_augmented_cached(&se->run_node, &cfs_rq->tasks_timeline, &min_vruntime_cb); avg_vruntime_sub(cfs_rq, se); + update_zero_vruntime(cfs_rq); } struct sched_entity *__pick_root_entity(struct cfs_rq *cfs_rq) @@ -1226,7 +1200,6 @@ static void update_curr(struct cfs_rq *cfs_rq) curr->vruntime += calc_delta_fair(delta_exec, curr); resched = update_deadline(cfs_rq, curr); - update_min_vruntime(cfs_rq); if (entity_is_task(curr)) { /* @@ -3808,15 +3781,6 @@ static void reweight_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, if (!curr) __enqueue_entity(cfs_rq, se); cfs_rq->nr_queued++; - - /* - * The entity's vruntime has been adjusted, so let's check - * whether the rq-wide min_vruntime needs updated too. Since - * the calculations above require stable min_vruntime rather - * than up-to-date one, we do the update at the end of the - * reweight process. - */ - update_min_vruntime(cfs_rq); } } @@ -5429,15 +5393,6 @@ dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags) update_cfs_group(se); - /* - * Now advance min_vruntime if @se was the entity holding it back, - * except when: DEQUEUE_SAVE && !DEQUEUE_MOVE, in this case we'll be - * put back on, and if we advance min_vruntime, we'll be placed back - * further than we started -- i.e. we'll be penalized. - */ - if ((flags & (DEQUEUE_SAVE | DEQUEUE_MOVE)) != DEQUEUE_SAVE) - update_min_vruntime(cfs_rq); - if (flags & DEQUEUE_DELAYED) finish_delayed_dequeue_entity(se); @@ -9015,7 +8970,6 @@ static void yield_task_fair(struct rq *rq) if (entity_eligible(cfs_rq, se)) { se->vruntime = se->deadline; se->deadline += calc_delta_fair(se->slice, se); - update_min_vruntime(cfs_rq); } } @@ -13078,23 +13032,6 @@ static inline void task_tick_core(struct rq *rq, struct task_struct *curr) * Which shows that S and s_i transform alike (which makes perfect sense * given that S is basically the (weighted) average of s_i). * - * Then: - * - * x -> s_min := min{s_i} (8) - * - * to obtain: - * - * \Sum_i w_i (s_i - s_min) - * S = s_min + ------------------------ (9) - * \Sum_i w_i - * - * Which already looks familiar, and is the basis for our current - * approximation: - * - * S ~= s_min (10) - * - * Now, obviously, (10) is absolute crap :-), but it sorta works. - * * So the thing to remember is that the above is strictly UP. It is * possible to generalize to multiple runqueues -- however it gets really * yuck when you have to add affinity support, as illustrated by our very @@ -13116,23 +13053,23 @@ static inline void task_tick_core(struct rq *rq, struct task_struct *curr) * Let, for our runqueue 'k': * * T_k = \Sum_i w_i s_i - * W_k = \Sum_i w_i ; for all i of k (11) + * W_k = \Sum_i w_i ; for all i of k (8) * * Then we can write (6) like: * * T_k - * S_k = --- (12) + * S_k = --- (9) * W_k * * From which immediately follows that: * * T_k + T_l - * S_k+l = --------- (13) + * S_k+l = --------- (10) * W_k + W_l * * On which we can define a combined lag: * - * lag_k+l(i) := S_k+l - s_i (14) + * lag_k+l(i) := S_k+l - s_i (11) * * And that gives us the tools to compare tasks across a combined runqueue. * @@ -13143,7 +13080,7 @@ static inline void task_tick_core(struct rq *rq, struct task_struct *curr) * using (7); this only requires storing single 'time'-stamps. * * b) when comparing tasks between 2 runqueues of which one is forced-idle, - * compare the combined lag, per (14). + * compare the combined lag, per (11). * * Now, of course cgroups (I so hate them) make this more interesting in * that a) seems to suggest we need to iterate all cgroup on a CPU at such @@ -13191,12 +13128,11 @@ static inline void task_tick_core(struct rq *rq, struct task_struct *curr) * every tick. This limits the observed divergence due to the work * conservancy. * - * On top of that, we can improve upon things by moving away from our - * horrible (10) hack and moving to (9) and employing (13) here. + * On top of that, we can improve upon things by employing (10) here. */ /* - * se_fi_update - Update the cfs_rq->min_vruntime_fi in a CFS hierarchy if needed. + * se_fi_update - Update the cfs_rq->zero_vruntime_fi in a CFS hierarchy if needed. */ static void se_fi_update(const struct sched_entity *se, unsigned int fi_seq, bool forceidle) @@ -13210,7 +13146,7 @@ static void se_fi_update(const struct sched_entity *se, unsigned int fi_seq, cfs_rq->forceidle_seq = fi_seq; } - cfs_rq->min_vruntime_fi = cfs_rq->min_vruntime; + cfs_rq->zero_vruntime_fi = cfs_rq->zero_vruntime; } } @@ -13263,11 +13199,11 @@ bool cfs_prio_less(const struct task_struct *a, const struct task_struct *b, /* * Find delta after normalizing se's vruntime with its cfs_rq's - * min_vruntime_fi, which would have been updated in prior calls + * zero_vruntime_fi, which would have been updated in prior calls * to se_fi_update(). */ delta = (s64)(sea->vruntime - seb->vruntime) + - (s64)(cfs_rqb->min_vruntime_fi - cfs_rqa->min_vruntime_fi); + (s64)(cfs_rqb->zero_vruntime_fi - cfs_rqa->zero_vruntime_fi); return delta > 0; } @@ -13513,7 +13449,7 @@ static void set_next_task_fair(struct rq *rq, struct task_struct *p, bool first) void init_cfs_rq(struct cfs_rq *cfs_rq) { cfs_rq->tasks_timeline = RB_ROOT_CACHED; - cfs_rq->min_vruntime = (u64)(-(1LL << 20)); + cfs_rq->zero_vruntime = (u64)(-(1LL << 20)); raw_spin_lock_init(&cfs_rq->removed.lock); } diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index 82e74e8ca2ea..5a3cf81c27be 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -681,10 +681,10 @@ struct cfs_rq { s64 avg_vruntime; u64 avg_load; - u64 min_vruntime; + u64 zero_vruntime; #ifdef CONFIG_SCHED_CORE unsigned int forceidle_seq; - u64 min_vruntime_fi; + u64 zero_vruntime_fi; #endif struct rb_root_cached tasks_timeline;

5 hours, 29 minutes

1
0
0 0

FAILED: patch "[PATCH] sched_ext: Fix missing post-enqueue handling in" failed to apply to 6.12-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 6.12-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-6.12.y git checkout FETCH_HEAD git cherry-pick -x f5e1e5ec204da11fa87fdf006d451d80ce06e118 # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025122951-giggle-reveler-2e6b@gregkh' --subject-prefix 'PATCH 6.12.y' HEAD^.. Possible dependencies: thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From f5e1e5ec204da11fa87fdf006d451d80ce06e118 Mon Sep 17 00:00:00 2001 From: Tejun Heo <tj(a)kernel.org> Date: Thu, 11 Dec 2025 15:45:04 -1000 Subject: [PATCH] sched_ext: Fix missing post-enqueue handling in move_local_task_to_local_dsq() move_local_task_to_local_dsq() is used when moving a task from a non-local DSQ to a local DSQ on the same CPU. It directly manipulates the local DSQ without going through dispatch_enqueue() and was missing the post-enqueue handling that triggers preemption when SCX_ENQ_PREEMPT is set or the idle task is running. The function is used by move_task_between_dsqs() which backs scx_bpf_dsq_move() and may be called while the CPU is busy. Add local_dsq_post_enq() call to move_local_task_to_local_dsq(). As the dispatch path doesn't need post-enqueue handling, add SCX_RQ_IN_BALANCE early exit to keep consume_dispatch_q() behavior unchanged and avoid triggering unnecessary resched when scx_bpf_dsq_move() is used from the dispatch path. Fixes: 4c30f5ce4f7a ("sched_ext: Implement scx_bpf_dispatch[_vtime]_from_dsq()") Cc: stable(a)vger.kernel.org # v6.12+ Reviewed-by: Andrea Righi <arighi(a)nvidia.com> Reviewed-by: Emil Tsalapatis <emil(a)etsalapatis.com> Signed-off-by: Tejun Heo <tj(a)kernel.org> diff --git a/kernel/sched/ext.c b/kernel/sched/ext.c index c78efa99406f..695503a2f7d1 100644 --- a/kernel/sched/ext.c +++ b/kernel/sched/ext.c @@ -988,6 +988,14 @@ static void local_dsq_post_enq(struct scx_dispatch_q *dsq, struct task_struct *p struct rq *rq = container_of(dsq, struct rq, scx.local_dsq); bool preempt = false; + /* + * If @rq is in balance, the CPU is already vacant and looking for the + * next task to run. No need to preempt or trigger resched after moving + * @p into its local DSQ. + */ + if (rq->scx.flags & SCX_RQ_IN_BALANCE) + return; + if ((enq_flags & SCX_ENQ_PREEMPT) && p != rq->curr && rq->curr->sched_class == &ext_sched_class) { rq->curr->scx.slice = 0; @@ -1636,6 +1644,8 @@ static void move_local_task_to_local_dsq(struct task_struct *p, u64 enq_flags, dsq_mod_nr(dst_dsq, 1); p->scx.dsq = dst_dsq; + + local_dsq_post_enq(dst_dsq, p, enq_flags); } /**

5 hours, 30 minutes

1
0
0 0

FAILED: patch "[PATCH] sched_ext: Fix bypass depth leak on scx_enable() failure" failed to apply to 6.12-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 6.12-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-6.12.y git checkout FETCH_HEAD git cherry-pick -x 9f769637a93fac81689b80df6855f545839cf999 # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025122940-sneak-unvocal-9de2@gregkh' --subject-prefix 'PATCH 6.12.y' HEAD^.. Possible dependencies: thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 9f769637a93fac81689b80df6855f545839cf999 Mon Sep 17 00:00:00 2001 From: Tejun Heo <tj(a)kernel.org> Date: Tue, 9 Dec 2025 11:04:33 -1000 Subject: [PATCH] sched_ext: Fix bypass depth leak on scx_enable() failure scx_enable() calls scx_bypass(true) to initialize in bypass mode and then scx_bypass(false) on success to exit. If scx_enable() fails during task initialization - e.g. scx_cgroup_init() or scx_init_task() returns an error - it jumps to err_disable while bypass is still active. scx_disable_workfn() then calls scx_bypass(true/false) for its own bypass, leaving the bypass depth at 1 instead of 0. This causes the system to remain permanently in bypass mode after a failed scx_enable(). Failures after task initialization is complete - e.g. scx_tryset_enable_state() at the end - already call scx_bypass(false) before reaching the error path and are not affected. This only affects a subset of failure modes. Fix it by tracking whether scx_enable() called scx_bypass(true) in a bool and having scx_disable_workfn() call an extra scx_bypass(false) to clear it. This is a temporary measure as the bypass depth will be moved into the sched instance, which will make this tracking unnecessary. Fixes: 8c2090c504e9 ("sched_ext: Initialize in bypass mode") Cc: stable(a)vger.kernel.org # v6.12+ Reported-by: Chris Mason <clm(a)meta.com> Reviewed-by: Emil Tsalapatis <emil(a)etsalapatis.com> Link: https://lore.kernel.org/stable/286e6f7787a81239e1ce2989b52391ce%40kernel.org Signed-off-by: Tejun Heo <tj(a)kernel.org> diff --git a/kernel/sched/ext.c b/kernel/sched/ext.c index bd74b371f52d..c4465ccefea4 100644 --- a/kernel/sched/ext.c +++ b/kernel/sched/ext.c @@ -41,6 +41,13 @@ static bool scx_init_task_enabled; static bool scx_switching_all; DEFINE_STATIC_KEY_FALSE(__scx_switched_all); +/* + * Tracks whether scx_enable() called scx_bypass(true). Used to balance bypass + * depth on enable failure. Will be removed when bypass depth is moved into the + * sched instance. + */ +static bool scx_bypassed_for_enable; + static atomic_long_t scx_nr_rejected = ATOMIC_LONG_INIT(0); static atomic_long_t scx_hotplug_seq = ATOMIC_LONG_INIT(0); @@ -4318,6 +4325,11 @@ static void scx_disable_workfn(struct kthread_work *work) scx_dsp_max_batch = 0; free_kick_syncs(); + if (scx_bypassed_for_enable) { + scx_bypassed_for_enable = false; + scx_bypass(false); + } + mutex_unlock(&scx_enable_mutex); WARN_ON_ONCE(scx_set_enable_state(SCX_DISABLED) != SCX_DISABLING); @@ -4970,6 +4982,7 @@ static int scx_enable(struct sched_ext_ops *ops, struct bpf_link *link) * Init in bypass mode to guarantee forward progress. */ scx_bypass(true); + scx_bypassed_for_enable = true; for (i = SCX_OPI_NORMAL_BEGIN; i < SCX_OPI_NORMAL_END; i++) if (((void (**)(void))ops)[i]) @@ -5067,6 +5080,7 @@ static int scx_enable(struct sched_ext_ops *ops, struct bpf_link *link) scx_task_iter_stop(&sti); percpu_up_write(&scx_fork_rwsem); + scx_bypassed_for_enable = false; scx_bypass(false); if (!scx_tryset_enable_state(SCX_ENABLED, SCX_ENABLING)) {

5 hours, 30 minutes

1
0
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror