From 889a2070dc09e47af5972cd0a3730c9350be4e07 Mon Sep 17 00:00:00 2001 From: Krzysztof Kozlowski Date: Sat, 11 Mar 2023 10:40:21 +0100 Subject: [PATCH 01/43] rtc: omap: include header for omap_rtc_power_off_program prototype MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit [ Upstream commit f69c2b5420497b7a54181ce170d682cbeb1f119f ] Non-static functions should have a prototype: drivers/rtc/rtc-omap.c:410:5: error: no previous prototype for ‘omap_rtc_power_off_program’ [-Werror=missing-prototypes] Fixes: 6256f7f7f217 ("rtc: OMAP: Add support for rtc-only mode") Signed-off-by: Krzysztof Kozlowski Link: https://lore.kernel.org/r/20230311094021.79730-1-krzysztof.kozlowski@linaro.org Signed-off-by: Alexandre Belloni Signed-off-by: Sasha Levin --- drivers/rtc/rtc-omap.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/rtc/rtc-omap.c b/drivers/rtc/rtc-omap.c index c20fc7937dfa..18ae2a4f26ea 100644 --- a/drivers/rtc/rtc-omap.c +++ b/drivers/rtc/rtc-omap.c @@ -25,6 +25,7 @@ #include #include #include +#include /* * The OMAP RTC is a year/month/day/hours/minutes/seconds BCD clock From 9ad3221c86cc9c6305594b742d4a72dfbd4ea579 Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Tue, 7 Mar 2023 12:51:27 +0300 Subject: [PATCH 02/43] RDMA/mlx4: Prevent shift wrapping in set_user_sq_size() [ Upstream commit d50b3c73f1ac20dabc53dc6e9d64ce9c79a331eb ] The ucmd->log_sq_bb_count variable is controlled by the user so this shift can wrap. Fix it by using check_shl_overflow() in the same way that it was done in commit 515f60004ed9 ("RDMA/hns: Prevent undefined behavior in hns_roce_set_user_sq_size()"). Fixes: 839041329fd3 ("IB/mlx4: Sanity check userspace send queue sizes") Signed-off-by: Dan Carpenter Link: https://lore.kernel.org/r/a8dfbd1d-c019-4556-930b-bab1ded73b10@kili.mountain Signed-off-by: Leon Romanovsky Signed-off-by: Sasha Levin --- drivers/infiniband/hw/mlx4/qp.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c index c6a815a705fe..255194029e2d 100644 --- a/drivers/infiniband/hw/mlx4/qp.c +++ b/drivers/infiniband/hw/mlx4/qp.c @@ -412,9 +412,13 @@ static int set_user_sq_size(struct mlx4_ib_dev *dev, struct mlx4_ib_qp *qp, struct mlx4_ib_create_qp *ucmd) { + u32 cnt; + /* Sanity check SQ size before proceeding */ - if ((1 << ucmd->log_sq_bb_count) > dev->dev->caps.max_wqes || - ucmd->log_sq_stride > + if (check_shl_overflow(1, ucmd->log_sq_bb_count, &cnt) || + cnt > dev->dev->caps.max_wqes) + return -EINVAL; + if (ucmd->log_sq_stride > ilog2(roundup_pow_of_two(dev->dev->caps.max_sq_desc_sz)) || ucmd->log_sq_stride < MLX4_IB_MIN_SQ_STRIDE) return -EINVAL; From 9e5a7c110645524accef030e39ffa50ff01dbe86 Mon Sep 17 00:00:00 2001 From: Martin Blumenstingl Date: Mon, 20 Mar 2023 22:21:42 +0100 Subject: [PATCH 03/43] rtc: meson-vrtc: Use ktime_get_real_ts64() to get the current time [ Upstream commit 0e6255fa3f649170da6bd1a544680589cfae1131 ] The VRTC alarm register can be programmed with an amount of seconds after which the SoC will be woken up by the VRTC timer again. We are already converting the alarm time from meson_vrtc_set_alarm() to "seconds since 1970". This means we also need to use "seconds since 1970" for the current time. This fixes a problem where setting the alarm to one minute in the future results in the firmware (which handles wakeup) to output (on the serial console) that the system will be woken up in billions of seconds. ktime_get_raw_ts64() returns the time since boot, not since 1970. Switch to ktime_get_real_ts64() to fix the calculation of the alarm time and to make the SoC wake up at the specified date/time. Also the firmware (which manages suspend) now prints either 59 or 60 seconds until wakeup (depending on how long it takes for the system to enter suspend). Fixes: 6ef35398e827 ("rtc: Add Amlogic Virtual Wake RTC") Signed-off-by: Martin Blumenstingl Reviewed-by: Neil Armstrong Reviewed-by: Kevin Hilman Link: https://lore.kernel.org/r/20230320212142.2355062-1-martin.blumenstingl@googlemail.com Signed-off-by: Alexandre Belloni Signed-off-by: Sasha Levin --- drivers/rtc/rtc-meson-vrtc.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/rtc/rtc-meson-vrtc.c b/drivers/rtc/rtc-meson-vrtc.c index e6bd0808a092..18ff8439b5bb 100644 --- a/drivers/rtc/rtc-meson-vrtc.c +++ b/drivers/rtc/rtc-meson-vrtc.c @@ -23,7 +23,7 @@ static int meson_vrtc_read_time(struct device *dev, struct rtc_time *tm) struct timespec64 time; dev_dbg(dev, "%s\n", __func__); - ktime_get_raw_ts64(&time); + ktime_get_real_ts64(&time); rtc_time64_to_tm(time.tv_sec, tm); return 0; @@ -96,7 +96,7 @@ static int __maybe_unused meson_vrtc_suspend(struct device *dev) long alarm_secs; struct timespec64 time; - ktime_get_raw_ts64(&time); + ktime_get_real_ts64(&time); local_time = time.tv_sec; dev_dbg(dev, "alarm_time = %lus, local_time=%lus\n", From 391fbf0d081d23a9919cb7f3a7aa649455991e3f Mon Sep 17 00:00:00 2001 From: Sebastian Reichel Date: Fri, 17 Mar 2023 23:56:57 +0100 Subject: [PATCH 04/43] power: supply: generic-adc-battery: fix unit scaling MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit [ Upstream commit 44263f50065969f2344808388bd589740f026167 ] power-supply properties are reported in µV, µA and µW. The IIO API provides mV, mA, mW, so the values need to be multiplied by 1000. Fixes: e60fea794e6e ("power: battery: Generic battery driver using IIO") Reviewed-by: Linus Walleij Reviewed-by: Matti Vaittinen Signed-off-by: Sebastian Reichel Signed-off-by: Sasha Levin --- drivers/power/supply/generic-adc-battery.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/drivers/power/supply/generic-adc-battery.c b/drivers/power/supply/generic-adc-battery.c index 58f09314741a..09bb4ff291cb 100644 --- a/drivers/power/supply/generic-adc-battery.c +++ b/drivers/power/supply/generic-adc-battery.c @@ -138,6 +138,9 @@ static int read_channel(struct gab *adc_bat, enum power_supply_property psp, result); if (ret < 0) pr_err("read channel error\n"); + else + *result *= 1000; + return ret; } From ac7f79ee11c3e965d88fe66bfedaa4a52b624262 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Cl=C3=A9ment=20L=C3=A9ger?= Date: Tue, 31 Jan 2023 09:32:27 +0100 Subject: [PATCH 05/43] clk: add missing of_node_put() in "assigned-clocks" property parsing MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit [ Upstream commit 27a6e1b09a782517fddac91259970ac466a3f7b6 ] When returning from of_parse_phandle_with_args(), the np member of the of_phandle_args structure should be put after usage. Add missing of_node_put() calls in both __set_clk_parents() and __set_clk_rates(). Fixes: 86be408bfbd8 ("clk: Support for clock parents and rates assigned from device tree") Signed-off-by: Clément Léger Link: https://lore.kernel.org/r/20230131083227.10990-1-clement.leger@bootlin.com Signed-off-by: Stephen Boyd Signed-off-by: Sasha Levin --- drivers/clk/clk-conf.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) diff --git a/drivers/clk/clk-conf.c b/drivers/clk/clk-conf.c index 2ef819606c41..1a4e6340f95c 100644 --- a/drivers/clk/clk-conf.c +++ b/drivers/clk/clk-conf.c @@ -33,9 +33,12 @@ static int __set_clk_parents(struct device_node *node, bool clk_supplier) else return rc; } - if (clkspec.np == node && !clk_supplier) + if (clkspec.np == node && !clk_supplier) { + of_node_put(clkspec.np); return 0; + } pclk = of_clk_get_from_provider(&clkspec); + of_node_put(clkspec.np); if (IS_ERR(pclk)) { if (PTR_ERR(pclk) != -EPROBE_DEFER) pr_warn("clk: couldn't get parent clock %d for %pOF\n", @@ -48,10 +51,12 @@ static int __set_clk_parents(struct device_node *node, bool clk_supplier) if (rc < 0) goto err; if (clkspec.np == node && !clk_supplier) { + of_node_put(clkspec.np); rc = 0; goto err; } clk = of_clk_get_from_provider(&clkspec); + of_node_put(clkspec.np); if (IS_ERR(clk)) { if (PTR_ERR(clk) != -EPROBE_DEFER) pr_warn("clk: couldn't get assigned clock %d for %pOF\n", @@ -93,10 +98,13 @@ static int __set_clk_rates(struct device_node *node, bool clk_supplier) else return rc; } - if (clkspec.np == node && !clk_supplier) + if (clkspec.np == node && !clk_supplier) { + of_node_put(clkspec.np); return 0; + } clk = of_clk_get_from_provider(&clkspec); + of_node_put(clkspec.np); if (IS_ERR(clk)) { if (PTR_ERR(clk) != -EPROBE_DEFER) pr_warn("clk: couldn't get clock %d for %pOF\n", From 3ef7a4c0d5cd35ce3d9fa62b28720379604eabdd Mon Sep 17 00:00:00 2001 From: Tetsuo Handa Date: Sun, 2 Apr 2023 14:10:13 +0900 Subject: [PATCH 06/43] RDMA/siw: Remove namespace check from siw_netdev_event() [ Upstream commit 266e9b3475ba82212062771fdbc40be0e3c06ec8 ] syzbot is reporting that siw_netdev_event(NETDEV_UNREGISTER) cannot destroy siw_device created after unshare(CLONE_NEWNET) due to net namespace check. It seems that this check was by error there and should be removed. Reported-by: syzbot Link: https://syzkaller.appspot.com/bug?extid=5e70d01ee8985ae62a3b Suggested-by: Jason Gunthorpe Suggested-by: Leon Romanovsky Fixes: bdcf26bf9b3a ("rdma/siw: network and RDMA core interface") Signed-off-by: Tetsuo Handa Link: https://lore.kernel.org/r/a44e9ac5-44e2-d575-9e30-02483cc7ffd1@I-love.SAKURA.ne.jp Reviewed-by: Bernard Metzler Signed-off-by: Leon Romanovsky Signed-off-by: Sasha Levin --- drivers/infiniband/sw/siw/siw_main.c | 3 --- 1 file changed, 3 deletions(-) diff --git a/drivers/infiniband/sw/siw/siw_main.c b/drivers/infiniband/sw/siw/siw_main.c index 32a553a1b905..5ba0893f1f01 100644 --- a/drivers/infiniband/sw/siw/siw_main.c +++ b/drivers/infiniband/sw/siw/siw_main.c @@ -458,9 +458,6 @@ static int siw_netdev_event(struct notifier_block *nb, unsigned long event, dev_dbg(&netdev->dev, "siw: event %lu\n", event); - if (dev_net(netdev) != &init_net) - return NOTIFY_OK; - base_dev = ib_device_get_by_netdev(netdev, RDMA_DRIVER_SIW); if (!base_dev) return NOTIFY_OK; From 16f596cfe95602d03513728061177ba7f866506d Mon Sep 17 00:00:00 2001 From: Mark Zhang Date: Thu, 30 Mar 2023 10:23:51 +0300 Subject: [PATCH 07/43] RDMA/cm: Trace icm_send_rej event before the cm state is reset [ Upstream commit bd9de1badac7e4ff6780365d4aa38983f5e2a436 ] Trace icm_send_rej event before the cm state is reset to idle, so that correct cm state will be logged. For example when an incoming request is rejected, the old trace log was: icm_send_rej: local_id=961102742 remote_id=3829151631 state=IDLE reason=REJ_CONSUMER_DEFINED With this patch: icm_send_rej: local_id=312971016 remote_id=3778819983 state=MRA_REQ_SENT reason=REJ_CONSUMER_DEFINED Fixes: 8dc105befe16 ("RDMA/cm: Add tracepoints to track MAD send operations") Signed-off-by: Mark Zhang Link: https://lore.kernel.org/r/20230330072351.481200-1-markzhang@nvidia.com Signed-off-by: Leon Romanovsky Signed-off-by: Sasha Levin --- drivers/infiniband/core/cm.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c index 3133b6be6cab..db1a25fbe2fa 100644 --- a/drivers/infiniband/core/cm.c +++ b/drivers/infiniband/core/cm.c @@ -2924,6 +2924,8 @@ static int cm_send_rej_locked(struct cm_id_private *cm_id_priv, (ari && ari_length > IB_CM_REJ_ARI_LENGTH)) return -EINVAL; + trace_icm_send_rej(&cm_id_priv->id, reason); + switch (state) { case IB_CM_REQ_SENT: case IB_CM_MRA_REQ_RCVD: @@ -2954,7 +2956,6 @@ static int cm_send_rej_locked(struct cm_id_private *cm_id_priv, return -EINVAL; } - trace_icm_send_rej(&cm_id_priv->id, reason); ret = ib_post_send_mad(msg, NULL); if (ret) { cm_free_msg(msg); From 8ec6acdb9b6a80eeb13e778dfedb5d72a88f14fe Mon Sep 17 00:00:00 2001 From: Saravanan Vajravel Date: Wed, 5 Apr 2023 21:25:49 -0700 Subject: [PATCH 08/43] RDMA/srpt: Add a check for valid 'mad_agent' pointer [ Upstream commit eca5cd9474cd26d62f9756f536e2e656d3f62f3a ] When unregistering MAD agent, srpt module has a non-null check for 'mad_agent' pointer before invoking ib_unregister_mad_agent(). This check can pass if 'mad_agent' variable holds an error value. The 'mad_agent' can have an error value for a short window when srpt_add_one() and srpt_remove_one() is executed simultaneously. In srpt module, added a valid pointer check for 'sport->mad_agent' before unregistering MAD agent. This issue can hit when RoCE driver unregisters ib_device Stack Trace: ------------ BUG: kernel NULL pointer dereference, address: 000000000000004d PGD 145003067 P4D 145003067 PUD 2324fe067 PMD 0 Oops: 0002 [#1] PREEMPT SMP NOPTI CPU: 10 PID: 4459 Comm: kworker/u80:0 Kdump: loaded Tainted: P Hardware name: Dell Inc. PowerEdge R640/06NR82, BIOS 2.5.4 01/13/2020 Workqueue: bnxt_re bnxt_re_task [bnxt_re] RIP: 0010:_raw_spin_lock_irqsave+0x19/0x40 Call Trace: ib_unregister_mad_agent+0x46/0x2f0 [ib_core] IPv6: ADDRCONF(NETDEV_CHANGE): bond0: link becomes ready ? __schedule+0x20b/0x560 srpt_unregister_mad_agent+0x93/0xd0 [ib_srpt] srpt_remove_one+0x20/0x150 [ib_srpt] remove_client_context+0x88/0xd0 [ib_core] bond0: (slave p2p1): link status definitely up, 100000 Mbps full duplex disable_device+0x8a/0x160 [ib_core] bond0: active interface up! ? kernfs_name_hash+0x12/0x80 (NULL device *): Bonding Info Received: rdev: 000000006c0b8247 __ib_unregister_device+0x42/0xb0 [ib_core] (NULL device *): Master: mode: 4 num_slaves:2 ib_unregister_device+0x22/0x30 [ib_core] (NULL device *): Slave: id: 105069936 name:p2p1 link:0 state:0 bnxt_re_stopqps_and_ib_uninit+0x83/0x90 [bnxt_re] bnxt_re_alloc_lag+0x12e/0x4e0 [bnxt_re] Fixes: a42d985bd5b2 ("ib_srpt: Initial SRP Target merge for v3.3-rc1") Reviewed-by: Selvin Xavier Reviewed-by: Kashyap Desai Signed-off-by: Saravanan Vajravel Link: https://lore.kernel.org/r/20230406042549.507328-1-saravanan.vajravel@broadcom.com Reviewed-by: Bart Van Assche Signed-off-by: Leon Romanovsky Signed-off-by: Sasha Levin --- drivers/infiniband/ulp/srpt/ib_srpt.c | 23 +++++++++++++---------- 1 file changed, 13 insertions(+), 10 deletions(-) diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.c b/drivers/infiniband/ulp/srpt/ib_srpt.c index c0ed08fcab48..983f59c87b79 100644 --- a/drivers/infiniband/ulp/srpt/ib_srpt.c +++ b/drivers/infiniband/ulp/srpt/ib_srpt.c @@ -549,6 +549,7 @@ static int srpt_format_guid(char *buf, unsigned int size, const __be64 *guid) */ static int srpt_refresh_port(struct srpt_port *sport) { + struct ib_mad_agent *mad_agent; struct ib_mad_reg_req reg_req; struct ib_port_modify port_modify; struct ib_port_attr port_attr; @@ -593,24 +594,26 @@ static int srpt_refresh_port(struct srpt_port *sport) set_bit(IB_MGMT_METHOD_GET, reg_req.method_mask); set_bit(IB_MGMT_METHOD_SET, reg_req.method_mask); - sport->mad_agent = ib_register_mad_agent(sport->sdev->device, - sport->port, - IB_QPT_GSI, - ®_req, 0, - srpt_mad_send_handler, - srpt_mad_recv_handler, - sport, 0); - if (IS_ERR(sport->mad_agent)) { + mad_agent = ib_register_mad_agent(sport->sdev->device, + sport->port, + IB_QPT_GSI, + ®_req, 0, + srpt_mad_send_handler, + srpt_mad_recv_handler, + sport, 0); + if (IS_ERR(mad_agent)) { pr_err("%s-%d: MAD agent registration failed (%ld). Note: this is expected if SR-IOV is enabled.\n", dev_name(&sport->sdev->device->dev), sport->port, - PTR_ERR(sport->mad_agent)); + PTR_ERR(mad_agent)); sport->mad_agent = NULL; memset(&port_modify, 0, sizeof(port_modify)); port_modify.clr_port_cap_mask = IB_PORT_DEVICE_MGMT_SUP; ib_modify_port(sport->sdev->device, sport->port, 0, &port_modify); - + return 0; } + + sport->mad_agent = mad_agent; } return 0; From 823b59fc56170eac408268511c1d6878844f0496 Mon Sep 17 00:00:00 2001 From: Patrick Kelsey Date: Fri, 7 Apr 2023 12:52:39 -0400 Subject: [PATCH 09/43] IB/hfi1: Fix SDMA mmu_rb_node not being evicted in LRU order [ Upstream commit 9fe8fec5e43d5a80f43cbf61aaada1b047a1eb61 ] hfi1_mmu_rb_remove_unless_exact() did not move mmu_rb_node objects in mmu_rb_handler->lru_list after getting a cache hit on an mmu_rb_node. As a result, hfi1_mmu_rb_evict() was not guaranteed to evict truly least-recently used nodes. This could be a performance issue for an application when that application: - Uses some long-lived buffers frequently. - Uses a large number of buffers once. - Hits the mmu_rb_handler cache size or pinned-page limits, forcing mmu_rb_handler cache entries to be evicted. In this case, the one-time use buffers cause the long-lived buffer entries to eventually filter to the end of the LRU list where hfi1_mmu_rb_evict() will consider evicting a frequently-used long-lived entry instead of evicting one of the one-time use entries. Fix this by inserting new mmu_rb_node at the tail of mmu_rb_handler->lru_list and move mmu_rb_ndoe to the tail of mmu_rb_handler->lru_list when the mmu_rb_node is a hit in hfi1_mmu_rb_remove_unless_exact(). Change hfi1_mmu_rb_evict() to evict from the head of mmu_rb_handler->lru_list instead of the tail. Fixes: 0636e9ab8355 ("IB/hfi1: Add cache evict LRU list") Signed-off-by: Brendan Cunningham Signed-off-by: Patrick Kelsey Signed-off-by: Dennis Dalessandro Link: https://lore.kernel.org/r/168088635931.3027109.10423156330761536044.stgit@252.162.96.66.static.eigbox.net Signed-off-by: Leon Romanovsky Signed-off-by: Sasha Levin --- drivers/infiniband/hw/hfi1/mmu_rb.c | 13 ++++++------- 1 file changed, 6 insertions(+), 7 deletions(-) diff --git a/drivers/infiniband/hw/hfi1/mmu_rb.c b/drivers/infiniband/hw/hfi1/mmu_rb.c index ed8a96ae61ce..d8b6bf271db8 100644 --- a/drivers/infiniband/hw/hfi1/mmu_rb.c +++ b/drivers/infiniband/hw/hfi1/mmu_rb.c @@ -171,7 +171,7 @@ int hfi1_mmu_rb_insert(struct mmu_rb_handler *handler, goto unlock; } __mmu_int_rb_insert(mnode, &handler->root); - list_add(&mnode->list, &handler->lru_list); + list_add_tail(&mnode->list, &handler->lru_list); ret = handler->ops->insert(handler->ops_arg, mnode); if (ret) { @@ -222,8 +222,10 @@ bool hfi1_mmu_rb_remove_unless_exact(struct mmu_rb_handler *handler, spin_lock_irqsave(&handler->lock, flags); node = __mmu_rb_search(handler, addr, len); if (node) { - if (node->addr == addr && node->len == len) + if (node->addr == addr && node->len == len) { + list_move_tail(&node->list, &handler->lru_list); goto unlock; + } __mmu_int_rb_remove(node, &handler->root); list_del(&node->list); /* remove from LRU list */ ret = true; @@ -247,8 +249,7 @@ void hfi1_mmu_rb_evict(struct mmu_rb_handler *handler, void *evict_arg) INIT_LIST_HEAD(&del_list); spin_lock_irqsave(&handler->lock, flags); - list_for_each_entry_safe_reverse(rbnode, ptr, &handler->lru_list, - list) { + list_for_each_entry_safe(rbnode, ptr, &handler->lru_list, list) { if (handler->ops->evict(handler->ops_arg, rbnode, evict_arg, &stop)) { __mmu_int_rb_remove(rbnode, &handler->root); @@ -260,9 +261,7 @@ void hfi1_mmu_rb_evict(struct mmu_rb_handler *handler, void *evict_arg) } spin_unlock_irqrestore(&handler->lock, flags); - while (!list_empty(&del_list)) { - rbnode = list_first_entry(&del_list, struct mmu_rb_node, list); - list_del(&rbnode->list); + list_for_each_entry_safe(rbnode, ptr, &del_list, list) { handler->ops->remove(handler->ops_arg, rbnode); } } From 1af73620a191a7721113ed1f8af2dcefc462b1ad Mon Sep 17 00:00:00 2001 From: Mike Marciniszyn Date: Mon, 29 Mar 2021 09:54:07 -0400 Subject: [PATCH 10/43] IB/hfi1: Add AIP tx traces [ Upstream commit 4bd00b55c978017aad10f0ff3e45525cd62cca07 ] Add traces to allow for debugging issues with AIP tx. Link: https://lore.kernel.org/r/1617026056-50483-2-git-send-email-dennis.dalessandro@cornelisnetworks.com Reviewed-by: Kaike Wan Signed-off-by: Mike Marciniszyn Signed-off-by: Dennis Dalessandro Signed-off-by: Jason Gunthorpe Stable-dep-of: 00cbce5cbf88 ("IB/hfi1: Fix bugs with non-PAGE_SIZE-end multi-iovec user SDMA requests") Signed-off-by: Sasha Levin --- drivers/infiniband/hw/hfi1/ipoib_tx.c | 18 ++++- drivers/infiniband/hw/hfi1/trace_tx.h | 104 ++++++++++++++++++++++++++ 2 files changed, 119 insertions(+), 3 deletions(-) diff --git a/drivers/infiniband/hw/hfi1/ipoib_tx.c b/drivers/infiniband/hw/hfi1/ipoib_tx.c index ab1eefffc14b..ec187ec74674 100644 --- a/drivers/infiniband/hw/hfi1/ipoib_tx.c +++ b/drivers/infiniband/hw/hfi1/ipoib_tx.c @@ -15,6 +15,7 @@ #include "verbs.h" #include "trace_ibhdrs.h" #include "ipoib.h" +#include "trace_tx.h" /* Add a convenience helper */ #define CIRC_ADD(val, add, size) (((val) + (add)) & ((size) - 1)) @@ -63,12 +64,14 @@ static u64 hfi1_ipoib_used(struct hfi1_ipoib_txq *txq) static void hfi1_ipoib_stop_txq(struct hfi1_ipoib_txq *txq) { + trace_hfi1_txq_stop(txq); if (atomic_inc_return(&txq->stops) == 1) netif_stop_subqueue(txq->priv->netdev, txq->q_idx); } static void hfi1_ipoib_wake_txq(struct hfi1_ipoib_txq *txq) { + trace_hfi1_txq_wake(txq); if (atomic_dec_and_test(&txq->stops)) netif_wake_subqueue(txq->priv->netdev, txq->q_idx); } @@ -89,8 +92,10 @@ static void hfi1_ipoib_check_queue_depth(struct hfi1_ipoib_txq *txq) { ++txq->sent_txreqs; if (hfi1_ipoib_used(txq) >= hfi1_ipoib_ring_hwat(txq) && - !atomic_xchg(&txq->ring_full, 1)) + !atomic_xchg(&txq->ring_full, 1)) { + trace_hfi1_txq_full(txq); hfi1_ipoib_stop_txq(txq); + } } static void hfi1_ipoib_check_queue_stopped(struct hfi1_ipoib_txq *txq) @@ -112,8 +117,10 @@ static void hfi1_ipoib_check_queue_stopped(struct hfi1_ipoib_txq *txq) * to protect against ring overflow. */ if (hfi1_ipoib_used(txq) < hfi1_ipoib_ring_lwat(txq) && - atomic_xchg(&txq->ring_full, 0)) + atomic_xchg(&txq->ring_full, 0)) { + trace_hfi1_txq_xmit_unstopped(txq); hfi1_ipoib_wake_txq(txq); + } } static void hfi1_ipoib_free_tx(struct ipoib_txreq *tx, int budget) @@ -405,6 +412,7 @@ static struct ipoib_txreq *hfi1_ipoib_send_dma_common(struct net_device *dev, sdma_select_engine_sc(priv->dd, txp->flow.tx_queue, txp->flow.sc5); + trace_hfi1_flow_switch(txp->txq); } return tx; @@ -525,6 +533,7 @@ static int hfi1_ipoib_send_dma_list(struct net_device *dev, if (txq->flow.as_int != txp->flow.as_int) { int ret; + trace_hfi1_flow_flush(txq); ret = hfi1_ipoib_flush_tx_list(dev, txq); if (unlikely(ret)) { if (ret == -EBUSY) @@ -635,8 +644,10 @@ static int hfi1_ipoib_sdma_sleep(struct sdma_engine *sde, /* came from non-list submit */ list_add_tail(&txreq->list, &txq->tx_list); if (list_empty(&txq->wait.list)) { - if (!atomic_xchg(&txq->no_desc, 1)) + if (!atomic_xchg(&txq->no_desc, 1)) { + trace_hfi1_txq_queued(txq); hfi1_ipoib_stop_txq(txq); + } iowait_queue(pkts_sent, wait->iow, &sde->dmawait); } @@ -659,6 +670,7 @@ static void hfi1_ipoib_sdma_wakeup(struct iowait *wait, int reason) struct hfi1_ipoib_txq *txq = container_of(wait, struct hfi1_ipoib_txq, wait); + trace_hfi1_txq_wakeup(txq); if (likely(txq->priv->netdev->reg_state == NETREG_REGISTERED)) iowait_schedule(wait, system_highpri_wq, WORK_CPU_UNBOUND); } diff --git a/drivers/infiniband/hw/hfi1/trace_tx.h b/drivers/infiniband/hw/hfi1/trace_tx.h index 769e5e4710c6..847654196dd3 100644 --- a/drivers/infiniband/hw/hfi1/trace_tx.h +++ b/drivers/infiniband/hw/hfi1/trace_tx.h @@ -53,6 +53,7 @@ #include "hfi.h" #include "mad.h" #include "sdma.h" +#include "ipoib.h" const char *parse_sdma_flags(struct trace_seq *p, u64 desc0, u64 desc1); @@ -858,6 +859,109 @@ DEFINE_EVENT( TP_ARGS(qp, flag) ); +DECLARE_EVENT_CLASS(/* AIP */ + hfi1_ipoib_txq_template, + TP_PROTO(struct hfi1_ipoib_txq *txq), + TP_ARGS(txq), + TP_STRUCT__entry(/* entry */ + DD_DEV_ENTRY(txq->priv->dd) + __field(struct hfi1_ipoib_txq *, txq) + __field(struct sdma_engine *, sde) + __field(ulong, head) + __field(ulong, tail) + __field(uint, used) + __field(uint, flow) + __field(int, stops) + __field(int, no_desc) + __field(u8, idx) + __field(u8, stopped) + ), + TP_fast_assign(/* assign */ + DD_DEV_ASSIGN(txq->priv->dd) + __entry->txq = txq; + __entry->sde = txq->sde; + __entry->head = txq->tx_ring.head; + __entry->tail = txq->tx_ring.tail; + __entry->idx = txq->q_idx; + __entry->used = + txq->sent_txreqs - + atomic64_read(&txq->complete_txreqs); + __entry->flow = txq->flow.as_int; + __entry->stops = atomic_read(&txq->stops); + __entry->no_desc = atomic_read(&txq->no_desc); + __entry->stopped = + __netif_subqueue_stopped(txq->priv->netdev, txq->q_idx); + ), + TP_printk(/* print */ + "[%s] txq %llx idx %u sde %llx head %lx tail %lx flow %x used %u stops %d no_desc %d stopped %u", + __get_str(dev), + (unsigned long long)__entry->txq, + __entry->idx, + (unsigned long long)__entry->sde, + __entry->head, + __entry->tail, + __entry->flow, + __entry->used, + __entry->stops, + __entry->no_desc, + __entry->stopped + ) +); + +DEFINE_EVENT(/* queue stop */ + hfi1_ipoib_txq_template, hfi1_txq_stop, + TP_PROTO(struct hfi1_ipoib_txq *txq), + TP_ARGS(txq) +); + +DEFINE_EVENT(/* queue wake */ + hfi1_ipoib_txq_template, hfi1_txq_wake, + TP_PROTO(struct hfi1_ipoib_txq *txq), + TP_ARGS(txq) +); + +DEFINE_EVENT(/* flow flush */ + hfi1_ipoib_txq_template, hfi1_flow_flush, + TP_PROTO(struct hfi1_ipoib_txq *txq), + TP_ARGS(txq) +); + +DEFINE_EVENT(/* flow switch */ + hfi1_ipoib_txq_template, hfi1_flow_switch, + TP_PROTO(struct hfi1_ipoib_txq *txq), + TP_ARGS(txq) +); + +DEFINE_EVENT(/* wakeup */ + hfi1_ipoib_txq_template, hfi1_txq_wakeup, + TP_PROTO(struct hfi1_ipoib_txq *txq), + TP_ARGS(txq) +); + +DEFINE_EVENT(/* full */ + hfi1_ipoib_txq_template, hfi1_txq_full, + TP_PROTO(struct hfi1_ipoib_txq *txq), + TP_ARGS(txq) +); + +DEFINE_EVENT(/* queued */ + hfi1_ipoib_txq_template, hfi1_txq_queued, + TP_PROTO(struct hfi1_ipoib_txq *txq), + TP_ARGS(txq) +); + +DEFINE_EVENT(/* xmit_stopped */ + hfi1_ipoib_txq_template, hfi1_txq_xmit_stopped, + TP_PROTO(struct hfi1_ipoib_txq *txq), + TP_ARGS(txq) +); + +DEFINE_EVENT(/* xmit_unstopped */ + hfi1_ipoib_txq_template, hfi1_txq_xmit_unstopped, + TP_PROTO(struct hfi1_ipoib_txq *txq), + TP_ARGS(txq) +); + #endif /* __HFI1_TRACE_TX_H */ #undef TRACE_INCLUDE_PATH From f84c02597857f6b9db24179f9e3ae65942e993fa Mon Sep 17 00:00:00 2001 From: Mike Marciniszyn Date: Mon, 29 Mar 2021 09:54:13 -0400 Subject: [PATCH 11/43] IB/hfi1: Add additional usdma traces [ Upstream commit 6b13215df1d37f5be23fc4a01a915a287b25ce15 ] Add traces that were vital in isolating an issue with pq waitlist in commit fa8dac396863 ("IB/hfi1: Fix another case where pq is left on waitlist") Link: https://lore.kernel.org/r/1617026056-50483-8-git-send-email-dennis.dalessandro@cornelisnetworks.com Reviewed-by: Kaike Wan Signed-off-by: Mike Marciniszyn Signed-off-by: Dennis Dalessandro Signed-off-by: Jason Gunthorpe Stable-dep-of: 00cbce5cbf88 ("IB/hfi1: Fix bugs with non-PAGE_SIZE-end multi-iovec user SDMA requests") Signed-off-by: Sasha Levin --- drivers/infiniband/hw/hfi1/trace_tx.h | 75 ++++++++++++++++++++++++++ drivers/infiniband/hw/hfi1/user_sdma.c | 12 +++-- drivers/infiniband/hw/hfi1/user_sdma.h | 1 + 3 files changed, 85 insertions(+), 3 deletions(-) diff --git a/drivers/infiniband/hw/hfi1/trace_tx.h b/drivers/infiniband/hw/hfi1/trace_tx.h index 847654196dd3..d44fc54858b9 100644 --- a/drivers/infiniband/hw/hfi1/trace_tx.h +++ b/drivers/infiniband/hw/hfi1/trace_tx.h @@ -54,6 +54,7 @@ #include "mad.h" #include "sdma.h" #include "ipoib.h" +#include "user_sdma.h" const char *parse_sdma_flags(struct trace_seq *p, u64 desc0, u64 desc1); @@ -654,6 +655,80 @@ TRACE_EVENT(hfi1_sdma_user_completion, __entry->code) ); +TRACE_EVENT(hfi1_usdma_defer, + TP_PROTO(struct hfi1_user_sdma_pkt_q *pq, + struct sdma_engine *sde, + struct iowait *wait), + TP_ARGS(pq, sde, wait), + TP_STRUCT__entry(DD_DEV_ENTRY(pq->dd) + __field(struct hfi1_user_sdma_pkt_q *, pq) + __field(struct sdma_engine *, sde) + __field(struct iowait *, wait) + __field(int, engine) + __field(int, empty) + ), + TP_fast_assign(DD_DEV_ASSIGN(pq->dd); + __entry->pq = pq; + __entry->sde = sde; + __entry->wait = wait; + __entry->engine = sde->this_idx; + __entry->empty = list_empty(&__entry->wait->list); + ), + TP_printk("[%s] pq %llx sde %llx wait %llx engine %d empty %d", + __get_str(dev), + (unsigned long long)__entry->pq, + (unsigned long long)__entry->sde, + (unsigned long long)__entry->wait, + __entry->engine, + __entry->empty + ) +); + +TRACE_EVENT(hfi1_usdma_activate, + TP_PROTO(struct hfi1_user_sdma_pkt_q *pq, + struct iowait *wait, + int reason), + TP_ARGS(pq, wait, reason), + TP_STRUCT__entry(DD_DEV_ENTRY(pq->dd) + __field(struct hfi1_user_sdma_pkt_q *, pq) + __field(struct iowait *, wait) + __field(int, reason) + ), + TP_fast_assign(DD_DEV_ASSIGN(pq->dd); + __entry->pq = pq; + __entry->wait = wait; + __entry->reason = reason; + ), + TP_printk("[%s] pq %llx wait %llx reason %d", + __get_str(dev), + (unsigned long long)__entry->pq, + (unsigned long long)__entry->wait, + __entry->reason + ) +); + +TRACE_EVENT(hfi1_usdma_we, + TP_PROTO(struct hfi1_user_sdma_pkt_q *pq, + int we_ret), + TP_ARGS(pq, we_ret), + TP_STRUCT__entry(DD_DEV_ENTRY(pq->dd) + __field(struct hfi1_user_sdma_pkt_q *, pq) + __field(int, state) + __field(int, we_ret) + ), + TP_fast_assign(DD_DEV_ASSIGN(pq->dd); + __entry->pq = pq; + __entry->state = pq->state; + __entry->we_ret = we_ret; + ), + TP_printk("[%s] pq %llx state %d we_ret %d", + __get_str(dev), + (unsigned long long)__entry->pq, + __entry->state, + __entry->we_ret + ) +); + const char *print_u32_array(struct trace_seq *, u32 *, int); #define __print_u32_hex(arr, len) print_u32_array(p, arr, len) diff --git a/drivers/infiniband/hw/hfi1/user_sdma.c b/drivers/infiniband/hw/hfi1/user_sdma.c index 4a4956f96a7e..da5b2e37355a 100644 --- a/drivers/infiniband/hw/hfi1/user_sdma.c +++ b/drivers/infiniband/hw/hfi1/user_sdma.c @@ -133,6 +133,7 @@ static int defer_packet_queue( container_of(wait->iow, struct hfi1_user_sdma_pkt_q, busy); write_seqlock(&sde->waitlock); + trace_hfi1_usdma_defer(pq, sde, &pq->busy); if (sdma_progress(sde, seq, txreq)) goto eagain; /* @@ -157,7 +158,8 @@ static void activate_packet_queue(struct iowait *wait, int reason) { struct hfi1_user_sdma_pkt_q *pq = container_of(wait, struct hfi1_user_sdma_pkt_q, busy); - pq->busy.lock = NULL; + + trace_hfi1_usdma_activate(pq, wait, reason); xchg(&pq->state, SDMA_PKT_Q_ACTIVE); wake_up(&wait->wait_dma); }; @@ -599,13 +601,17 @@ int hfi1_user_sdma_process_request(struct hfi1_filedata *fd, while (req->seqsubmitted != req->info.npkts) { ret = user_sdma_send_pkts(req, pcount); if (ret < 0) { + int we_ret; + if (ret != -EBUSY) goto free_req; - if (wait_event_interruptible_timeout( + we_ret = wait_event_interruptible_timeout( pq->busy.wait_dma, pq->state == SDMA_PKT_Q_ACTIVE, msecs_to_jiffies( - SDMA_IOWAIT_TIMEOUT)) <= 0) + SDMA_IOWAIT_TIMEOUT)); + trace_hfi1_usdma_we(pq, we_ret); + if (we_ret <= 0) flush_pq_iowait(pq); } } diff --git a/drivers/infiniband/hw/hfi1/user_sdma.h b/drivers/infiniband/hw/hfi1/user_sdma.h index 1e8c02fe8ad1..fabe58139906 100644 --- a/drivers/infiniband/hw/hfi1/user_sdma.h +++ b/drivers/infiniband/hw/hfi1/user_sdma.h @@ -53,6 +53,7 @@ #include "common.h" #include "iowait.h" #include "user_exp_rcv.h" +#include "mmu_rb.h" /* The maximum number of Data io vectors per message/request */ #define MAX_VECTORS_PER_REQ 8 From 9c4c6512d7330b743c4ffd18bd999a86ca26db0d Mon Sep 17 00:00:00 2001 From: Patrick Kelsey Date: Fri, 7 Apr 2023 12:52:44 -0400 Subject: [PATCH 12/43] IB/hfi1: Fix bugs with non-PAGE_SIZE-end multi-iovec user SDMA requests [ Upstream commit 00cbce5cbf88459cd1aa1d60d0f1df15477df127 ] hfi1 user SDMA request processing has two bugs that can cause data corruption for user SDMA requests that have multiple payload iovecs where an iovec other than the tail iovec does not run up to the page boundary for the buffer pointed to by that iovec.a Here are the specific bugs: 1. user_sdma_txadd() does not use struct user_sdma_iovec->iov.iov_len. Rather, user_sdma_txadd() will add up to PAGE_SIZE bytes from iovec to the packet, even if some of those bytes are past iovec->iov.iov_len and are thus not intended to be in the packet. 2. user_sdma_txadd() and user_sdma_send_pkts() fail to advance to the next iovec in user_sdma_request->iovs when the current iovec is not PAGE_SIZE and does not contain enough data to complete the packet. The transmitted packet will contain the wrong data from the iovec pages. This has not been an issue with SDMA packets from hfi1 Verbs or PSM2 because they only produce iovecs that end short of PAGE_SIZE as the tail iovec of an SDMA request. Fixing these bugs exposes other bugs with the SDMA pin cache (struct mmu_rb_handler) that get in way of supporting user SDMA requests with multiple payload iovecs whose buffers do not end at PAGE_SIZE. So this commit fixes those issues as well. Here are the mmu_rb_handler bugs that non-PAGE_SIZE-end multi-iovec payload user SDMA requests can hit: 1. Overlapping memory ranges in mmu_rb_handler will result in duplicate pinnings. 2. When extending an existing mmu_rb_handler entry (struct mmu_rb_node), the mmu_rb code (1) removes the existing entry under a lock, (2) releases that lock, pins the new pages, (3) then reacquires the lock to insert the extended mmu_rb_node. If someone else comes in and inserts an overlapping entry between (2) and (3), insert in (3) will fail. The failure path code in this case unpins _all_ pages in either the original mmu_rb_node or the new mmu_rb_node that was inserted between (2) and (3). 3. In hfi1_mmu_rb_remove_unless_exact(), mmu_rb_node->refcount is incremented outside of mmu_rb_handler->lock. As a result, mmu_rb_node could be evicted by another thread that gets mmu_rb_handler->lock and checks mmu_rb_node->refcount before mmu_rb_node->refcount is incremented. 4. Related to #2 above, SDMA request submission failure path does not check mmu_rb_node->refcount before freeing mmu_rb_node object. If there are other SDMA requests in progress whose iovecs have pointers to the now-freed mmu_rb_node(s), those pointers to the now-freed mmu_rb nodes will be dereferenced when those SDMA requests complete. Fixes: 7be85676f1d1 ("IB/hfi1: Don't remove RB entry when not needed.") Fixes: 7724105686e7 ("IB/hfi1: add driver files") Signed-off-by: Brendan Cunningham Signed-off-by: Patrick Kelsey Signed-off-by: Dennis Dalessandro Link: https://lore.kernel.org/r/168088636445.3027109.10054635277810177889.stgit@252.162.96.66.static.eigbox.net Signed-off-by: Leon Romanovsky Signed-off-by: Sasha Levin --- drivers/infiniband/hw/hfi1/ipoib_tx.c | 1 + drivers/infiniband/hw/hfi1/mmu_rb.c | 66 +-- drivers/infiniband/hw/hfi1/mmu_rb.h | 8 +- drivers/infiniband/hw/hfi1/sdma.c | 21 +- drivers/infiniband/hw/hfi1/sdma.h | 16 +- drivers/infiniband/hw/hfi1/sdma_txreq.h | 1 + drivers/infiniband/hw/hfi1/trace_mmu.h | 4 - drivers/infiniband/hw/hfi1/user_sdma.c | 600 +++++++++++++++--------- drivers/infiniband/hw/hfi1/user_sdma.h | 5 - drivers/infiniband/hw/hfi1/verbs.c | 4 +- drivers/infiniband/hw/hfi1/vnic_sdma.c | 1 + 11 files changed, 423 insertions(+), 304 deletions(-) diff --git a/drivers/infiniband/hw/hfi1/ipoib_tx.c b/drivers/infiniband/hw/hfi1/ipoib_tx.c index ec187ec74674..956fc3fd88b9 100644 --- a/drivers/infiniband/hw/hfi1/ipoib_tx.c +++ b/drivers/infiniband/hw/hfi1/ipoib_tx.c @@ -251,6 +251,7 @@ static int hfi1_ipoib_build_ulp_payload(struct ipoib_txreq *tx, const skb_frag_t *frag = &skb_shinfo(skb)->frags[i]; ret = sdma_txadd_page(dd, + NULL, txreq, skb_frag_page(frag), frag->bv_offset, diff --git a/drivers/infiniband/hw/hfi1/mmu_rb.c b/drivers/infiniband/hw/hfi1/mmu_rb.c index d8b6bf271db8..d331184ded30 100644 --- a/drivers/infiniband/hw/hfi1/mmu_rb.c +++ b/drivers/infiniband/hw/hfi1/mmu_rb.c @@ -167,7 +167,7 @@ int hfi1_mmu_rb_insert(struct mmu_rb_handler *handler, spin_lock_irqsave(&handler->lock, flags); node = __mmu_rb_search(handler, mnode->addr, mnode->len); if (node) { - ret = -EINVAL; + ret = -EEXIST; goto unlock; } __mmu_int_rb_insert(mnode, &handler->root); @@ -184,6 +184,19 @@ unlock: return ret; } +/* Caller must hold handler lock */ +struct mmu_rb_node *hfi1_mmu_rb_get_first(struct mmu_rb_handler *handler, + unsigned long addr, unsigned long len) +{ + struct mmu_rb_node *node; + + trace_hfi1_mmu_rb_search(addr, len); + node = __mmu_int_rb_iter_first(&handler->root, addr, (addr + len) - 1); + if (node) + list_move_tail(&node->list, &handler->lru_list); + return node; +} + /* Caller must hold handler lock */ static struct mmu_rb_node *__mmu_rb_search(struct mmu_rb_handler *handler, unsigned long addr, @@ -208,34 +221,6 @@ static struct mmu_rb_node *__mmu_rb_search(struct mmu_rb_handler *handler, return node; } -bool hfi1_mmu_rb_remove_unless_exact(struct mmu_rb_handler *handler, - unsigned long addr, unsigned long len, - struct mmu_rb_node **rb_node) -{ - struct mmu_rb_node *node; - unsigned long flags; - bool ret = false; - - if (current->mm != handler->mn.mm) - return ret; - - spin_lock_irqsave(&handler->lock, flags); - node = __mmu_rb_search(handler, addr, len); - if (node) { - if (node->addr == addr && node->len == len) { - list_move_tail(&node->list, &handler->lru_list); - goto unlock; - } - __mmu_int_rb_remove(node, &handler->root); - list_del(&node->list); /* remove from LRU list */ - ret = true; - } -unlock: - spin_unlock_irqrestore(&handler->lock, flags); - *rb_node = node; - return ret; -} - void hfi1_mmu_rb_evict(struct mmu_rb_handler *handler, void *evict_arg) { struct mmu_rb_node *rbnode, *ptr; @@ -266,29 +251,6 @@ void hfi1_mmu_rb_evict(struct mmu_rb_handler *handler, void *evict_arg) } } -/* - * It is up to the caller to ensure that this function does not race with the - * mmu invalidate notifier which may be calling the users remove callback on - * 'node'. - */ -void hfi1_mmu_rb_remove(struct mmu_rb_handler *handler, - struct mmu_rb_node *node) -{ - unsigned long flags; - - if (current->mm != handler->mn.mm) - return; - - /* Validity of handler and node pointers has been checked by caller. */ - trace_hfi1_mmu_rb_remove(node->addr, node->len); - spin_lock_irqsave(&handler->lock, flags); - __mmu_int_rb_remove(node, &handler->root); - list_del(&node->list); /* remove from LRU list */ - spin_unlock_irqrestore(&handler->lock, flags); - - handler->ops->remove(handler->ops_arg, node); -} - static int mmu_notifier_range_start(struct mmu_notifier *mn, const struct mmu_notifier_range *range) { diff --git a/drivers/infiniband/hw/hfi1/mmu_rb.h b/drivers/infiniband/hw/hfi1/mmu_rb.h index 423aacc67e94..0265d81c6206 100644 --- a/drivers/infiniband/hw/hfi1/mmu_rb.h +++ b/drivers/infiniband/hw/hfi1/mmu_rb.h @@ -93,10 +93,8 @@ void hfi1_mmu_rb_unregister(struct mmu_rb_handler *handler); int hfi1_mmu_rb_insert(struct mmu_rb_handler *handler, struct mmu_rb_node *mnode); void hfi1_mmu_rb_evict(struct mmu_rb_handler *handler, void *evict_arg); -void hfi1_mmu_rb_remove(struct mmu_rb_handler *handler, - struct mmu_rb_node *mnode); -bool hfi1_mmu_rb_remove_unless_exact(struct mmu_rb_handler *handler, - unsigned long addr, unsigned long len, - struct mmu_rb_node **rb_node); +struct mmu_rb_node *hfi1_mmu_rb_get_first(struct mmu_rb_handler *handler, + unsigned long addr, + unsigned long len); #endif /* _HFI1_MMU_RB_H */ diff --git a/drivers/infiniband/hw/hfi1/sdma.c b/drivers/infiniband/hw/hfi1/sdma.c index a044bee257f9..061562627dae 100644 --- a/drivers/infiniband/hw/hfi1/sdma.c +++ b/drivers/infiniband/hw/hfi1/sdma.c @@ -1635,22 +1635,7 @@ static inline void sdma_unmap_desc( struct hfi1_devdata *dd, struct sdma_desc *descp) { - switch (sdma_mapping_type(descp)) { - case SDMA_MAP_SINGLE: - dma_unmap_single( - &dd->pcidev->dev, - sdma_mapping_addr(descp), - sdma_mapping_len(descp), - DMA_TO_DEVICE); - break; - case SDMA_MAP_PAGE: - dma_unmap_page( - &dd->pcidev->dev, - sdma_mapping_addr(descp), - sdma_mapping_len(descp), - DMA_TO_DEVICE); - break; - } + system_descriptor_complete(dd, descp); } /* @@ -3170,7 +3155,7 @@ int ext_coal_sdma_tx_descs(struct hfi1_devdata *dd, struct sdma_txreq *tx, /* Add descriptor for coalesce buffer */ tx->desc_limit = MAX_DESC; - return _sdma_txadd_daddr(dd, SDMA_MAP_SINGLE, tx, + return _sdma_txadd_daddr(dd, SDMA_MAP_SINGLE, NULL, tx, addr, tx->tlen); } @@ -3210,10 +3195,12 @@ int _pad_sdma_tx_descs(struct hfi1_devdata *dd, struct sdma_txreq *tx) return rval; } } + /* finish the one just added */ make_tx_sdma_desc( tx, SDMA_MAP_NONE, + NULL, dd->sdma_pad_phys, sizeof(u32) - (tx->packet_len & (sizeof(u32) - 1))); _sdma_close_tx(dd, tx); diff --git a/drivers/infiniband/hw/hfi1/sdma.h b/drivers/infiniband/hw/hfi1/sdma.h index 7a851191f987..7d4f316ac6e4 100644 --- a/drivers/infiniband/hw/hfi1/sdma.h +++ b/drivers/infiniband/hw/hfi1/sdma.h @@ -635,6 +635,7 @@ static inline dma_addr_t sdma_mapping_addr(struct sdma_desc *d) static inline void make_tx_sdma_desc( struct sdma_txreq *tx, int type, + void *pinning_ctx, dma_addr_t addr, size_t len) { @@ -653,6 +654,7 @@ static inline void make_tx_sdma_desc( << SDMA_DESC0_PHY_ADDR_SHIFT) | (((u64)len & SDMA_DESC0_BYTE_COUNT_MASK) << SDMA_DESC0_BYTE_COUNT_SHIFT); + desc->pinning_ctx = pinning_ctx; } /* helper to extend txreq */ @@ -685,6 +687,7 @@ static inline void _sdma_close_tx(struct hfi1_devdata *dd, static inline int _sdma_txadd_daddr( struct hfi1_devdata *dd, int type, + void *pinning_ctx, struct sdma_txreq *tx, dma_addr_t addr, u16 len) @@ -694,6 +697,7 @@ static inline int _sdma_txadd_daddr( make_tx_sdma_desc( tx, type, + pinning_ctx, addr, len); WARN_ON(len > tx->tlen); tx->tlen -= len; @@ -714,6 +718,7 @@ static inline int _sdma_txadd_daddr( /** * sdma_txadd_page() - add a page to the sdma_txreq * @dd: the device to use for mapping + * @pinning_ctx: context to be released at descriptor retirement * @tx: tx request to which the page is added * @page: page to map * @offset: offset within the page @@ -729,6 +734,7 @@ static inline int _sdma_txadd_daddr( */ static inline int sdma_txadd_page( struct hfi1_devdata *dd, + void *pinning_ctx, struct sdma_txreq *tx, struct page *page, unsigned long offset, @@ -756,8 +762,7 @@ static inline int sdma_txadd_page( return -ENOSPC; } - return _sdma_txadd_daddr( - dd, SDMA_MAP_PAGE, tx, addr, len); + return _sdma_txadd_daddr(dd, SDMA_MAP_PAGE, pinning_ctx, tx, addr, len); } /** @@ -791,7 +796,8 @@ static inline int sdma_txadd_daddr( return rval; } - return _sdma_txadd_daddr(dd, SDMA_MAP_NONE, tx, addr, len); + return _sdma_txadd_daddr(dd, SDMA_MAP_NONE, NULL, tx, + addr, len); } /** @@ -837,8 +843,7 @@ static inline int sdma_txadd_kvaddr( return -ENOSPC; } - return _sdma_txadd_daddr( - dd, SDMA_MAP_SINGLE, tx, addr, len); + return _sdma_txadd_daddr(dd, SDMA_MAP_SINGLE, NULL, tx, addr, len); } struct iowait_work; @@ -1090,4 +1095,5 @@ extern uint mod_num_sdma; void sdma_update_lmc(struct hfi1_devdata *dd, u64 mask, u32 lid); +void system_descriptor_complete(struct hfi1_devdata *dd, struct sdma_desc *descp); #endif diff --git a/drivers/infiniband/hw/hfi1/sdma_txreq.h b/drivers/infiniband/hw/hfi1/sdma_txreq.h index 514a4784566b..4204650cebc2 100644 --- a/drivers/infiniband/hw/hfi1/sdma_txreq.h +++ b/drivers/infiniband/hw/hfi1/sdma_txreq.h @@ -61,6 +61,7 @@ struct sdma_desc { /* private: don't use directly */ u64 qw[2]; + void *pinning_ctx; }; /** diff --git a/drivers/infiniband/hw/hfi1/trace_mmu.h b/drivers/infiniband/hw/hfi1/trace_mmu.h index 3b7abbc382c2..c3055cff4d6b 100644 --- a/drivers/infiniband/hw/hfi1/trace_mmu.h +++ b/drivers/infiniband/hw/hfi1/trace_mmu.h @@ -78,10 +78,6 @@ DEFINE_EVENT(hfi1_mmu_rb_template, hfi1_mmu_rb_search, TP_PROTO(unsigned long addr, unsigned long len), TP_ARGS(addr, len)); -DEFINE_EVENT(hfi1_mmu_rb_template, hfi1_mmu_rb_remove, - TP_PROTO(unsigned long addr, unsigned long len), - TP_ARGS(addr, len)); - DEFINE_EVENT(hfi1_mmu_rb_template, hfi1_mmu_mem_invalidate, TP_PROTO(unsigned long addr, unsigned long len), TP_ARGS(addr, len)); diff --git a/drivers/infiniband/hw/hfi1/user_sdma.c b/drivers/infiniband/hw/hfi1/user_sdma.c index da5b2e37355a..1eb5a44a4ae6 100644 --- a/drivers/infiniband/hw/hfi1/user_sdma.c +++ b/drivers/infiniband/hw/hfi1/user_sdma.c @@ -65,7 +65,6 @@ #include "hfi.h" #include "sdma.h" -#include "mmu_rb.h" #include "user_sdma.h" #include "verbs.h" /* for the headers */ #include "common.h" /* for struct hfi1_tid_info */ @@ -80,11 +79,7 @@ static unsigned initial_pkt_count = 8; static int user_sdma_send_pkts(struct user_sdma_request *req, u16 maxpkts); static void user_sdma_txreq_cb(struct sdma_txreq *txreq, int status); static inline void pq_update(struct hfi1_user_sdma_pkt_q *pq); -static void user_sdma_free_request(struct user_sdma_request *req, bool unpin); -static int pin_vector_pages(struct user_sdma_request *req, - struct user_sdma_iovec *iovec); -static void unpin_vector_pages(struct mm_struct *mm, struct page **pages, - unsigned start, unsigned npages); +static void user_sdma_free_request(struct user_sdma_request *req); static int check_header_template(struct user_sdma_request *req, struct hfi1_pkt_header *hdr, u32 lrhlen, u32 datalen); @@ -122,6 +117,11 @@ static struct mmu_rb_ops sdma_rb_ops = { .invalidate = sdma_rb_invalidate }; +static int add_system_pages_to_sdma_packet(struct user_sdma_request *req, + struct user_sdma_txreq *tx, + struct user_sdma_iovec *iovec, + u32 *pkt_remaining); + static int defer_packet_queue( struct sdma_engine *sde, struct iowait_work *wait, @@ -453,6 +453,7 @@ int hfi1_user_sdma_process_request(struct hfi1_filedata *fd, ret = -EINVAL; goto free_req; } + /* Copy the header from the user buffer */ ret = copy_from_user(&req->hdr, iovec[idx].iov_base + sizeof(info), sizeof(req->hdr)); @@ -527,9 +528,8 @@ int hfi1_user_sdma_process_request(struct hfi1_filedata *fd, memcpy(&req->iovs[i].iov, iovec + idx++, sizeof(req->iovs[i].iov)); - ret = pin_vector_pages(req, &req->iovs[i]); - if (ret) { - req->data_iovs = i; + if (req->iovs[i].iov.iov_len == 0) { + ret = -EINVAL; goto free_req; } req->data_len += req->iovs[i].iov.iov_len; @@ -627,7 +627,7 @@ free_req: if (req->seqsubmitted) wait_event(pq->busy.wait_dma, (req->seqcomp == req->seqsubmitted - 1)); - user_sdma_free_request(req, true); + user_sdma_free_request(req); pq_update(pq); set_comp_state(pq, cq, info.comp_idx, ERROR, ret); } @@ -739,48 +739,6 @@ static int user_sdma_txadd_ahg(struct user_sdma_request *req, return ret; } -static int user_sdma_txadd(struct user_sdma_request *req, - struct user_sdma_txreq *tx, - struct user_sdma_iovec *iovec, u32 datalen, - u32 *queued_ptr, u32 *data_sent_ptr, - u64 *iov_offset_ptr) -{ - int ret; - unsigned int pageidx, len; - unsigned long base, offset; - u64 iov_offset = *iov_offset_ptr; - u32 queued = *queued_ptr, data_sent = *data_sent_ptr; - struct hfi1_user_sdma_pkt_q *pq = req->pq; - - base = (unsigned long)iovec->iov.iov_base; - offset = offset_in_page(base + iovec->offset + iov_offset); - pageidx = (((iovec->offset + iov_offset + base) - (base & PAGE_MASK)) >> - PAGE_SHIFT); - len = offset + req->info.fragsize > PAGE_SIZE ? - PAGE_SIZE - offset : req->info.fragsize; - len = min((datalen - queued), len); - ret = sdma_txadd_page(pq->dd, &tx->txreq, iovec->pages[pageidx], - offset, len); - if (ret) { - SDMA_DBG(req, "SDMA txreq add page failed %d\n", ret); - return ret; - } - iov_offset += len; - queued += len; - data_sent += len; - if (unlikely(queued < datalen && pageidx == iovec->npages && - req->iov_idx < req->data_iovs - 1)) { - iovec->offset += iov_offset; - iovec = &req->iovs[++req->iov_idx]; - iov_offset = 0; - } - - *queued_ptr = queued; - *data_sent_ptr = data_sent; - *iov_offset_ptr = iov_offset; - return ret; -} - static int user_sdma_send_pkts(struct user_sdma_request *req, u16 maxpkts) { int ret = 0; @@ -812,8 +770,7 @@ static int user_sdma_send_pkts(struct user_sdma_request *req, u16 maxpkts) maxpkts = req->info.npkts - req->seqnum; while (npkts < maxpkts) { - u32 datalen = 0, queued = 0, data_sent = 0; - u64 iov_offset = 0; + u32 datalen = 0; /* * Check whether any of the completions have come back @@ -906,27 +863,17 @@ static int user_sdma_send_pkts(struct user_sdma_request *req, u16 maxpkts) goto free_txreq; } - /* - * If the request contains any data vectors, add up to - * fragsize bytes to the descriptor. - */ - while (queued < datalen && - (req->sent + data_sent) < req->data_len) { - ret = user_sdma_txadd(req, tx, iovec, datalen, - &queued, &data_sent, &iov_offset); - if (ret) - goto free_txreq; - } - /* - * The txreq was submitted successfully so we can update - * the counters. - */ req->koffset += datalen; if (req_opcode(req->info.ctrl) == EXPECTED) req->tidoffset += datalen; - req->sent += data_sent; - if (req->data_len) - iovec->offset += iov_offset; + req->sent += datalen; + while (datalen) { + ret = add_system_pages_to_sdma_packet(req, tx, iovec, + &datalen); + if (ret) + goto free_txreq; + iovec = &req->iovs[req->iov_idx]; + } list_add_tail(&tx->txreq.list, &req->txps); /* * It is important to increment this here as it is used to @@ -963,133 +910,14 @@ free_tx: static u32 sdma_cache_evict(struct hfi1_user_sdma_pkt_q *pq, u32 npages) { struct evict_data evict_data; + struct mmu_rb_handler *handler = pq->handler; evict_data.cleared = 0; evict_data.target = npages; - hfi1_mmu_rb_evict(pq->handler, &evict_data); + hfi1_mmu_rb_evict(handler, &evict_data); return evict_data.cleared; } -static int pin_sdma_pages(struct user_sdma_request *req, - struct user_sdma_iovec *iovec, - struct sdma_mmu_node *node, - int npages) -{ - int pinned, cleared; - struct page **pages; - struct hfi1_user_sdma_pkt_q *pq = req->pq; - - pages = kcalloc(npages, sizeof(*pages), GFP_KERNEL); - if (!pages) - return -ENOMEM; - memcpy(pages, node->pages, node->npages * sizeof(*pages)); - - npages -= node->npages; -retry: - if (!hfi1_can_pin_pages(pq->dd, current->mm, - atomic_read(&pq->n_locked), npages)) { - cleared = sdma_cache_evict(pq, npages); - if (cleared >= npages) - goto retry; - } - pinned = hfi1_acquire_user_pages(current->mm, - ((unsigned long)iovec->iov.iov_base + - (node->npages * PAGE_SIZE)), npages, 0, - pages + node->npages); - if (pinned < 0) { - kfree(pages); - return pinned; - } - if (pinned != npages) { - unpin_vector_pages(current->mm, pages, node->npages, pinned); - return -EFAULT; - } - kfree(node->pages); - node->rb.len = iovec->iov.iov_len; - node->pages = pages; - atomic_add(pinned, &pq->n_locked); - return pinned; -} - -static void unpin_sdma_pages(struct sdma_mmu_node *node) -{ - if (node->npages) { - unpin_vector_pages(mm_from_sdma_node(node), node->pages, 0, - node->npages); - atomic_sub(node->npages, &node->pq->n_locked); - } -} - -static int pin_vector_pages(struct user_sdma_request *req, - struct user_sdma_iovec *iovec) -{ - int ret = 0, pinned, npages; - struct hfi1_user_sdma_pkt_q *pq = req->pq; - struct sdma_mmu_node *node = NULL; - struct mmu_rb_node *rb_node; - struct iovec *iov; - bool extracted; - - extracted = - hfi1_mmu_rb_remove_unless_exact(pq->handler, - (unsigned long) - iovec->iov.iov_base, - iovec->iov.iov_len, &rb_node); - if (rb_node) { - node = container_of(rb_node, struct sdma_mmu_node, rb); - if (!extracted) { - atomic_inc(&node->refcount); - iovec->pages = node->pages; - iovec->npages = node->npages; - iovec->node = node; - return 0; - } - } - - if (!node) { - node = kzalloc(sizeof(*node), GFP_KERNEL); - if (!node) - return -ENOMEM; - - node->rb.addr = (unsigned long)iovec->iov.iov_base; - node->pq = pq; - atomic_set(&node->refcount, 0); - } - - iov = &iovec->iov; - npages = num_user_pages((unsigned long)iov->iov_base, iov->iov_len); - if (node->npages < npages) { - pinned = pin_sdma_pages(req, iovec, node, npages); - if (pinned < 0) { - ret = pinned; - goto bail; - } - node->npages += pinned; - npages = node->npages; - } - iovec->pages = node->pages; - iovec->npages = npages; - iovec->node = node; - - ret = hfi1_mmu_rb_insert(req->pq->handler, &node->rb); - if (ret) { - iovec->node = NULL; - goto bail; - } - return 0; -bail: - unpin_sdma_pages(node); - kfree(node); - return ret; -} - -static void unpin_vector_pages(struct mm_struct *mm, struct page **pages, - unsigned start, unsigned npages) -{ - hfi1_release_user_pages(mm, pages + start, npages, false); - kfree(pages); -} - static int check_header_template(struct user_sdma_request *req, struct hfi1_pkt_header *hdr, u32 lrhlen, u32 datalen) @@ -1431,7 +1259,7 @@ static void user_sdma_txreq_cb(struct sdma_txreq *txreq, int status) if (req->seqcomp != req->info.npkts - 1) return; - user_sdma_free_request(req, false); + user_sdma_free_request(req); set_comp_state(pq, cq, req->info.comp_idx, state, status); pq_update(pq); } @@ -1442,10 +1270,8 @@ static inline void pq_update(struct hfi1_user_sdma_pkt_q *pq) wake_up(&pq->wait); } -static void user_sdma_free_request(struct user_sdma_request *req, bool unpin) +static void user_sdma_free_request(struct user_sdma_request *req) { - int i; - if (!list_empty(&req->txps)) { struct sdma_txreq *t, *p; @@ -1458,21 +1284,6 @@ static void user_sdma_free_request(struct user_sdma_request *req, bool unpin) } } - for (i = 0; i < req->data_iovs; i++) { - struct sdma_mmu_node *node = req->iovs[i].node; - - if (!node) - continue; - - req->iovs[i].node = NULL; - - if (unpin) - hfi1_mmu_rb_remove(req->pq->handler, - &node->rb); - else - atomic_dec(&node->refcount); - } - kfree(req->tids); clear_bit(req->info.comp_idx, req->pq->req_in_use); } @@ -1490,6 +1301,368 @@ static inline void set_comp_state(struct hfi1_user_sdma_pkt_q *pq, idx, state, ret); } +static void unpin_vector_pages(struct mm_struct *mm, struct page **pages, + unsigned int start, unsigned int npages) +{ + hfi1_release_user_pages(mm, pages + start, npages, false); + kfree(pages); +} + +static void free_system_node(struct sdma_mmu_node *node) +{ + if (node->npages) { + unpin_vector_pages(mm_from_sdma_node(node), node->pages, 0, + node->npages); + atomic_sub(node->npages, &node->pq->n_locked); + } + kfree(node); +} + +static inline void acquire_node(struct sdma_mmu_node *node) +{ + atomic_inc(&node->refcount); + WARN_ON(atomic_read(&node->refcount) < 0); +} + +static inline void release_node(struct mmu_rb_handler *handler, + struct sdma_mmu_node *node) +{ + atomic_dec(&node->refcount); + WARN_ON(atomic_read(&node->refcount) < 0); +} + +static struct sdma_mmu_node *find_system_node(struct mmu_rb_handler *handler, + unsigned long start, + unsigned long end) +{ + struct mmu_rb_node *rb_node; + struct sdma_mmu_node *node; + unsigned long flags; + + spin_lock_irqsave(&handler->lock, flags); + rb_node = hfi1_mmu_rb_get_first(handler, start, (end - start)); + if (!rb_node) { + spin_unlock_irqrestore(&handler->lock, flags); + return NULL; + } + node = container_of(rb_node, struct sdma_mmu_node, rb); + acquire_node(node); + spin_unlock_irqrestore(&handler->lock, flags); + + return node; +} + +static int pin_system_pages(struct user_sdma_request *req, + uintptr_t start_address, size_t length, + struct sdma_mmu_node *node, int npages) +{ + struct hfi1_user_sdma_pkt_q *pq = req->pq; + int pinned, cleared; + struct page **pages; + + pages = kcalloc(npages, sizeof(*pages), GFP_KERNEL); + if (!pages) + return -ENOMEM; + +retry: + if (!hfi1_can_pin_pages(pq->dd, current->mm, atomic_read(&pq->n_locked), + npages)) { + SDMA_DBG(req, "Evicting: nlocked %u npages %u", + atomic_read(&pq->n_locked), npages); + cleared = sdma_cache_evict(pq, npages); + if (cleared >= npages) + goto retry; + } + + SDMA_DBG(req, "Acquire user pages start_address %lx node->npages %u npages %u", + start_address, node->npages, npages); + pinned = hfi1_acquire_user_pages(current->mm, start_address, npages, 0, + pages); + + if (pinned < 0) { + kfree(pages); + SDMA_DBG(req, "pinned %d", pinned); + return pinned; + } + if (pinned != npages) { + unpin_vector_pages(current->mm, pages, node->npages, pinned); + SDMA_DBG(req, "npages %u pinned %d", npages, pinned); + return -EFAULT; + } + node->rb.addr = start_address; + node->rb.len = length; + node->pages = pages; + node->npages = npages; + atomic_add(pinned, &pq->n_locked); + SDMA_DBG(req, "done. pinned %d", pinned); + return 0; +} + +static int add_system_pinning(struct user_sdma_request *req, + struct sdma_mmu_node **node_p, + unsigned long start, unsigned long len) + +{ + struct hfi1_user_sdma_pkt_q *pq = req->pq; + struct sdma_mmu_node *node; + int ret; + + node = kzalloc(sizeof(*node), GFP_KERNEL); + if (!node) + return -ENOMEM; + + node->pq = pq; + ret = pin_system_pages(req, start, len, node, PFN_DOWN(len)); + if (ret == 0) { + ret = hfi1_mmu_rb_insert(pq->handler, &node->rb); + if (ret) + free_system_node(node); + else + *node_p = node; + + return ret; + } + + kfree(node); + return ret; +} + +static int get_system_cache_entry(struct user_sdma_request *req, + struct sdma_mmu_node **node_p, + size_t req_start, size_t req_len) +{ + struct hfi1_user_sdma_pkt_q *pq = req->pq; + u64 start = ALIGN_DOWN(req_start, PAGE_SIZE); + u64 end = PFN_ALIGN(req_start + req_len); + struct mmu_rb_handler *handler = pq->handler; + int ret; + + if ((end - start) == 0) { + SDMA_DBG(req, + "Request for empty cache entry req_start %lx req_len %lx start %llx end %llx", + req_start, req_len, start, end); + return -EINVAL; + } + + SDMA_DBG(req, "req_start %lx req_len %lu", req_start, req_len); + + while (1) { + struct sdma_mmu_node *node = + find_system_node(handler, start, end); + u64 prepend_len = 0; + + SDMA_DBG(req, "node %p start %llx end %llu", node, start, end); + if (!node) { + ret = add_system_pinning(req, node_p, start, + end - start); + if (ret == -EEXIST) { + /* + * Another execution context has inserted a + * conficting entry first. + */ + continue; + } + return ret; + } + + if (node->rb.addr <= start) { + /* + * This entry covers at least part of the region. If it doesn't extend + * to the end, then this will be called again for the next segment. + */ + *node_p = node; + return 0; + } + + SDMA_DBG(req, "prepend: node->rb.addr %lx, node->refcount %d", + node->rb.addr, atomic_read(&node->refcount)); + prepend_len = node->rb.addr - start; + + /* + * This node will not be returned, instead a new node + * will be. So release the reference. + */ + release_node(handler, node); + + /* Prepend a node to cover the beginning of the allocation */ + ret = add_system_pinning(req, node_p, start, prepend_len); + if (ret == -EEXIST) { + /* Another execution context has inserted a conficting entry first. */ + continue; + } + return ret; + } +} + +static int add_mapping_to_sdma_packet(struct user_sdma_request *req, + struct user_sdma_txreq *tx, + struct sdma_mmu_node *cache_entry, + size_t start, + size_t from_this_cache_entry) +{ + struct hfi1_user_sdma_pkt_q *pq = req->pq; + unsigned int page_offset; + unsigned int from_this_page; + size_t page_index; + void *ctx; + int ret; + + /* + * Because the cache may be more fragmented than the memory that is being accessed, + * it's not strictly necessary to have a descriptor per cache entry. + */ + + while (from_this_cache_entry) { + page_index = PFN_DOWN(start - cache_entry->rb.addr); + + if (page_index >= cache_entry->npages) { + SDMA_DBG(req, + "Request for page_index %zu >= cache_entry->npages %u", + page_index, cache_entry->npages); + return -EINVAL; + } + + page_offset = start - ALIGN_DOWN(start, PAGE_SIZE); + from_this_page = PAGE_SIZE - page_offset; + + if (from_this_page < from_this_cache_entry) { + ctx = NULL; + } else { + /* + * In the case they are equal the next line has no practical effect, + * but it's better to do a register to register copy than a conditional + * branch. + */ + from_this_page = from_this_cache_entry; + ctx = cache_entry; + } + + ret = sdma_txadd_page(pq->dd, ctx, &tx->txreq, + cache_entry->pages[page_index], + page_offset, from_this_page); + if (ret) { + /* + * When there's a failure, the entire request is freed by + * user_sdma_send_pkts(). + */ + SDMA_DBG(req, + "sdma_txadd_page failed %d page_index %lu page_offset %u from_this_page %u", + ret, page_index, page_offset, from_this_page); + return ret; + } + start += from_this_page; + from_this_cache_entry -= from_this_page; + } + return 0; +} + +static int add_system_iovec_to_sdma_packet(struct user_sdma_request *req, + struct user_sdma_txreq *tx, + struct user_sdma_iovec *iovec, + size_t from_this_iovec) +{ + struct mmu_rb_handler *handler = req->pq->handler; + + while (from_this_iovec > 0) { + struct sdma_mmu_node *cache_entry; + size_t from_this_cache_entry; + size_t start; + int ret; + + start = (uintptr_t)iovec->iov.iov_base + iovec->offset; + ret = get_system_cache_entry(req, &cache_entry, start, + from_this_iovec); + if (ret) { + SDMA_DBG(req, "pin system segment failed %d", ret); + return ret; + } + + from_this_cache_entry = cache_entry->rb.len - (start - cache_entry->rb.addr); + if (from_this_cache_entry > from_this_iovec) + from_this_cache_entry = from_this_iovec; + + ret = add_mapping_to_sdma_packet(req, tx, cache_entry, start, + from_this_cache_entry); + if (ret) { + /* + * We're guaranteed that there will be no descriptor + * completion callback that releases this node + * because only the last descriptor referencing it + * has a context attached, and a failure means the + * last descriptor was never added. + */ + release_node(handler, cache_entry); + SDMA_DBG(req, "add system segment failed %d", ret); + return ret; + } + + iovec->offset += from_this_cache_entry; + from_this_iovec -= from_this_cache_entry; + } + + return 0; +} + +static int add_system_pages_to_sdma_packet(struct user_sdma_request *req, + struct user_sdma_txreq *tx, + struct user_sdma_iovec *iovec, + u32 *pkt_data_remaining) +{ + size_t remaining_to_add = *pkt_data_remaining; + /* + * Walk through iovec entries, ensure the associated pages + * are pinned and mapped, add data to the packet until no more + * data remains to be added. + */ + while (remaining_to_add > 0) { + struct user_sdma_iovec *cur_iovec; + size_t from_this_iovec; + int ret; + + cur_iovec = iovec; + from_this_iovec = iovec->iov.iov_len - iovec->offset; + + if (from_this_iovec > remaining_to_add) { + from_this_iovec = remaining_to_add; + } else { + /* The current iovec entry will be consumed by this pass. */ + req->iov_idx++; + iovec++; + } + + ret = add_system_iovec_to_sdma_packet(req, tx, cur_iovec, + from_this_iovec); + if (ret) + return ret; + + remaining_to_add -= from_this_iovec; + } + *pkt_data_remaining = remaining_to_add; + + return 0; +} + +void system_descriptor_complete(struct hfi1_devdata *dd, + struct sdma_desc *descp) +{ + switch (sdma_mapping_type(descp)) { + case SDMA_MAP_SINGLE: + dma_unmap_single(&dd->pcidev->dev, sdma_mapping_addr(descp), + sdma_mapping_len(descp), DMA_TO_DEVICE); + break; + case SDMA_MAP_PAGE: + dma_unmap_page(&dd->pcidev->dev, sdma_mapping_addr(descp), + sdma_mapping_len(descp), DMA_TO_DEVICE); + break; + } + + if (descp->pinning_ctx) { + struct sdma_mmu_node *node = descp->pinning_ctx; + + release_node(node->rb.handler, node); + } +} + static bool sdma_rb_filter(struct mmu_rb_node *node, unsigned long addr, unsigned long len) { @@ -1536,8 +1709,7 @@ static void sdma_rb_remove(void *arg, struct mmu_rb_node *mnode) struct sdma_mmu_node *node = container_of(mnode, struct sdma_mmu_node, rb); - unpin_sdma_pages(node); - kfree(node); + free_system_node(node); } static int sdma_rb_invalidate(void *arg, struct mmu_rb_node *mnode) diff --git a/drivers/infiniband/hw/hfi1/user_sdma.h b/drivers/infiniband/hw/hfi1/user_sdma.h index fabe58139906..9d417aacfa8b 100644 --- a/drivers/infiniband/hw/hfi1/user_sdma.h +++ b/drivers/infiniband/hw/hfi1/user_sdma.h @@ -153,16 +153,11 @@ struct sdma_mmu_node { struct user_sdma_iovec { struct list_head list; struct iovec iov; - /* number of pages in this vector */ - unsigned int npages; - /* array of pinned pages for this vector */ - struct page **pages; /* * offset into the virtual address space of the vector at * which we last left off. */ u64 offset; - struct sdma_mmu_node *node; }; /* evict operation argument */ diff --git a/drivers/infiniband/hw/hfi1/verbs.c b/drivers/infiniband/hw/hfi1/verbs.c index 5f3edd255ca3..693922df3543 100644 --- a/drivers/infiniband/hw/hfi1/verbs.c +++ b/drivers/infiniband/hw/hfi1/verbs.c @@ -820,8 +820,8 @@ static int build_verbs_tx_desc( /* add icrc, lt byte, and padding to flit */ if (extra_bytes) - ret = sdma_txadd_daddr(sde->dd, &tx->txreq, - sde->dd->sdma_pad_phys, extra_bytes); + ret = sdma_txadd_daddr(sde->dd, &tx->txreq, sde->dd->sdma_pad_phys, + extra_bytes); bail_txadd: return ret; diff --git a/drivers/infiniband/hw/hfi1/vnic_sdma.c b/drivers/infiniband/hw/hfi1/vnic_sdma.c index 7d90b900131b..7658c620a125 100644 --- a/drivers/infiniband/hw/hfi1/vnic_sdma.c +++ b/drivers/infiniband/hw/hfi1/vnic_sdma.c @@ -106,6 +106,7 @@ static noinline int build_vnic_ulp_payload(struct sdma_engine *sde, /* combine physically continuous fragments later? */ ret = sdma_txadd_page(sde->dd, + NULL, &tx->txreq, skb_frag_page(frag), skb_frag_off(frag), From 73a65744c7fd785b3fd9d9541a6f238d0faae0ac Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 13 Mar 2023 18:45:53 -0400 Subject: [PATCH 13/43] NFSv4.1: Always send a RECLAIM_COMPLETE after establishing lease [ Upstream commit 40882deb83c29d8df4470d4e5e7f137b6acf7ad1 ] The spec requires that we always at least send a RECLAIM_COMPLETE when we're done establishing the lease and recovering any state. Fixes: fce5c838e133 ("nfs41: RECLAIM_COMPLETE functionality") Signed-off-by: Trond Myklebust Signed-off-by: Anna Schumaker Signed-off-by: Sasha Levin --- fs/nfs/nfs4state.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c index 628e030f8e3b..ff6ca05a9d44 100644 --- a/fs/nfs/nfs4state.c +++ b/fs/nfs/nfs4state.c @@ -67,6 +67,8 @@ #define OPENOWNER_POOL_SIZE 8 +static void nfs4_state_start_reclaim_reboot(struct nfs_client *clp); + const nfs4_stateid zero_stateid = { { .data = { 0 } }, .type = NFS4_SPECIAL_STATEID_TYPE, @@ -330,6 +332,8 @@ do_confirm: status = nfs4_proc_create_session(clp, cred); if (status != 0) goto out; + if (!(clp->cl_exchange_flags & EXCHGID4_FLAG_CONFIRMED_R)) + nfs4_state_start_reclaim_reboot(clp); nfs41_finish_session_reset(clp); nfs_mark_client_ready(clp, NFS_CS_READY); out: From 3aa2503717ba8b7cf529527d9bad09947f2f65a0 Mon Sep 17 00:00:00 2001 From: Nicolas Saenz Julienne Date: Mon, 18 Jan 2021 13:32:35 +0100 Subject: [PATCH 14/43] firmware: raspberrypi: Introduce devm_rpi_firmware_get() [ Upstream commit f663204c9a1f8d6fcc590667d9d7a9f44e064644 ] It'll simplify the firmware handling for most consumers. Suggested-by: Bartosz Golaszewski Signed-off-by: Nicolas Saenz Julienne Reviewed-by: Florian Fainelli Reviewed-by: Bartosz Golaszewski Stable-dep-of: 5bca3688bdbc ("Input: raspberrypi-ts - fix refcount leak in rpi_ts_probe") Signed-off-by: Sasha Levin --- drivers/firmware/raspberrypi.c | 29 ++++++++++++++++++++++ include/soc/bcm2835/raspberrypi-firmware.h | 8 ++++++ 2 files changed, 37 insertions(+) diff --git a/drivers/firmware/raspberrypi.c b/drivers/firmware/raspberrypi.c index 9eef49da47e0..45ff03da234a 100644 --- a/drivers/firmware/raspberrypi.c +++ b/drivers/firmware/raspberrypi.c @@ -243,6 +243,13 @@ void rpi_firmware_put(struct rpi_firmware *fw) } EXPORT_SYMBOL_GPL(rpi_firmware_put); +static void devm_rpi_firmware_put(void *data) +{ + struct rpi_firmware *fw = data; + + rpi_firmware_put(fw); +} + static int rpi_firmware_probe(struct platform_device *pdev) { struct device *dev = &pdev->dev; @@ -338,6 +345,28 @@ err_put_device: } EXPORT_SYMBOL_GPL(rpi_firmware_get); +/** + * devm_rpi_firmware_get - Get pointer to rpi_firmware structure. + * @firmware_node: Pointer to the firmware Device Tree node. + * + * Returns NULL is the firmware device is not ready. + */ +struct rpi_firmware *devm_rpi_firmware_get(struct device *dev, + struct device_node *firmware_node) +{ + struct rpi_firmware *fw; + + fw = rpi_firmware_get(firmware_node); + if (!fw) + return NULL; + + if (devm_add_action_or_reset(dev, devm_rpi_firmware_put, fw)) + return NULL; + + return fw; +} +EXPORT_SYMBOL_GPL(devm_rpi_firmware_get); + static const struct of_device_id rpi_firmware_of_match[] = { { .compatible = "raspberrypi,bcm2835-firmware", }, {}, diff --git a/include/soc/bcm2835/raspberrypi-firmware.h b/include/soc/bcm2835/raspberrypi-firmware.h index fdfef7fe40df..73ad784fca96 100644 --- a/include/soc/bcm2835/raspberrypi-firmware.h +++ b/include/soc/bcm2835/raspberrypi-firmware.h @@ -142,6 +142,8 @@ int rpi_firmware_property_list(struct rpi_firmware *fw, void *data, size_t tag_size); void rpi_firmware_put(struct rpi_firmware *fw); struct rpi_firmware *rpi_firmware_get(struct device_node *firmware_node); +struct rpi_firmware *devm_rpi_firmware_get(struct device *dev, + struct device_node *firmware_node); #else static inline int rpi_firmware_property(struct rpi_firmware *fw, u32 tag, void *data, size_t len) @@ -160,6 +162,12 @@ static inline struct rpi_firmware *rpi_firmware_get(struct device_node *firmware { return NULL; } + +static inline struct rpi_firmware *devm_rpi_firmware_get(struct device *dev, + struct device_node *firmware_node) +{ + return NULL; +} #endif #endif /* __SOC_RASPBERRY_FIRMWARE_H__ */ From 09bfd90d040fca9c0374af9b90f83e0f58e0b600 Mon Sep 17 00:00:00 2001 From: Nicolas Saenz Julienne Date: Mon, 18 Jan 2021 13:32:41 +0100 Subject: [PATCH 15/43] input: raspberrypi-ts: Release firmware handle when not needed [ Upstream commit 3b8ddff780b7d12e99ae39177f84b9003097777a ] There is no use for the firmware interface after getting the touch buffer address, so release it. Signed-off-by: Nicolas Saenz Julienne Acked-by: Dmitry Torokhov Reviewed-by: Florian Fainelli Stable-dep-of: 5bca3688bdbc ("Input: raspberrypi-ts - fix refcount leak in rpi_ts_probe") Signed-off-by: Sasha Levin --- drivers/input/touchscreen/raspberrypi-ts.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/input/touchscreen/raspberrypi-ts.c b/drivers/input/touchscreen/raspberrypi-ts.c index ef6aaed217cf..5000f5fd9ec3 100644 --- a/drivers/input/touchscreen/raspberrypi-ts.c +++ b/drivers/input/touchscreen/raspberrypi-ts.c @@ -160,7 +160,7 @@ static int rpi_ts_probe(struct platform_device *pdev) touchbuf = (u32)ts->fw_regs_phys; error = rpi_firmware_property(fw, RPI_FIRMWARE_FRAMEBUFFER_SET_TOUCHBUF, &touchbuf, sizeof(touchbuf)); - + rpi_firmware_put(fw); if (error || touchbuf != 0) { dev_warn(dev, "Failed to set touchbuf, %d\n", error); return error; From 0d6a5c9489c8a3d434e685066119c4333476dccd Mon Sep 17 00:00:00 2001 From: Miaoqian Lin Date: Thu, 13 Apr 2023 23:05:20 -0700 Subject: [PATCH 16/43] Input: raspberrypi-ts - fix refcount leak in rpi_ts_probe [ Upstream commit 5bca3688bdbc3b58a2894b8671a8e2378efe28bd ] rpi_firmware_get() take reference, we need to release it in error paths as well. Use devm_rpi_firmware_get() helper to handling the resources. Also remove the existing rpi_firmware_put(). Fixes: 0b9f28fed3f7 ("Input: add official Raspberry Pi's touchscreen driver") Fixes: 3b8ddff780b7 ("input: raspberrypi-ts: Release firmware handle when not needed") Signed-off-by: Miaoqian Lin Reviewed-by: Mattijs Korpershoek Link: https://lore.kernel.org/r/20221223074657.810346-1-linmq006@gmail.com Signed-off-by: Dmitry Torokhov Signed-off-by: Sasha Levin --- drivers/input/touchscreen/raspberrypi-ts.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/drivers/input/touchscreen/raspberrypi-ts.c b/drivers/input/touchscreen/raspberrypi-ts.c index 5000f5fd9ec3..45c575df994e 100644 --- a/drivers/input/touchscreen/raspberrypi-ts.c +++ b/drivers/input/touchscreen/raspberrypi-ts.c @@ -134,7 +134,7 @@ static int rpi_ts_probe(struct platform_device *pdev) return -ENOENT; } - fw = rpi_firmware_get(fw_node); + fw = devm_rpi_firmware_get(&pdev->dev, fw_node); of_node_put(fw_node); if (!fw) return -EPROBE_DEFER; @@ -160,7 +160,6 @@ static int rpi_ts_probe(struct platform_device *pdev) touchbuf = (u32)ts->fw_regs_phys; error = rpi_firmware_property(fw, RPI_FIRMWARE_FRAMEBUFFER_SET_TOUCHBUF, &touchbuf, sizeof(touchbuf)); - rpi_firmware_put(fw); if (error || touchbuf != 0) { dev_warn(dev, "Failed to set touchbuf, %d\n", error); return error; From 50440cdb0b4799ddb6b56b41cee7e92443544c76 Mon Sep 17 00:00:00 2001 From: Mark Bloch Date: Thu, 13 Apr 2023 12:23:09 +0300 Subject: [PATCH 17/43] RDMA/mlx5: Fix flow counter query via DEVX [ Upstream commit 3e358ea8614ddfbc59ca7a3f5dff5dde2b350b2c ] Commit cited in "fixes" tag added bulk support for flow counters but it didn't account that's also possible to query a counter using a non-base id if the counter was allocated as bulk. When a user performs a query, validate the flow counter id given in the mailbox is inside the valid range taking bulk value into account. Fixes: 208d70f562e5 ("IB/mlx5: Support flow counters offset for bulk counters") Signed-off-by: Mark Bloch Reviewed-by: Maor Gottlieb Link: https://lore.kernel.org/r/79d7fbe291690128e44672418934256254d93115.1681377114.git.leon@kernel.org Signed-off-by: Leon Romanovsky Signed-off-by: Sasha Levin --- drivers/infiniband/hw/mlx5/devx.c | 31 ++++++++++++++++++++++++++----- include/linux/mlx5/mlx5_ifc.h | 3 ++- 2 files changed, 28 insertions(+), 6 deletions(-) diff --git a/drivers/infiniband/hw/mlx5/devx.c b/drivers/infiniband/hw/mlx5/devx.c index 2f053f48f1be..a56ebdc15723 100644 --- a/drivers/infiniband/hw/mlx5/devx.c +++ b/drivers/infiniband/hw/mlx5/devx.c @@ -595,7 +595,21 @@ static bool devx_is_valid_obj_id(struct uverbs_attr_bundle *attrs, obj_id; case MLX5_IB_OBJECT_DEVX_OBJ: - return ((struct devx_obj *)uobj->object)->obj_id == obj_id; + { + u16 opcode = MLX5_GET(general_obj_in_cmd_hdr, in, opcode); + struct devx_obj *devx_uobj = uobj->object; + + if (opcode == MLX5_CMD_OP_QUERY_FLOW_COUNTER && + devx_uobj->flow_counter_bulk_size) { + u64 end; + + end = devx_uobj->obj_id + + devx_uobj->flow_counter_bulk_size; + return devx_uobj->obj_id <= obj_id && end > obj_id; + } + + return devx_uobj->obj_id == obj_id; + } default: return false; @@ -1416,10 +1430,17 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_CREATE)( goto obj_free; if (opcode == MLX5_CMD_OP_ALLOC_FLOW_COUNTER) { - u8 bulk = MLX5_GET(alloc_flow_counter_in, - cmd_in, - flow_counter_bulk); - obj->flow_counter_bulk_size = 128UL * bulk; + u32 bulk = MLX5_GET(alloc_flow_counter_in, + cmd_in, + flow_counter_bulk_log_size); + + if (bulk) + bulk = 1 << bulk; + else + bulk = 128UL * MLX5_GET(alloc_flow_counter_in, + cmd_in, + flow_counter_bulk); + obj->flow_counter_bulk_size = bulk; } uobj->object = obj; diff --git a/include/linux/mlx5/mlx5_ifc.h b/include/linux/mlx5/mlx5_ifc.h index 6ca97729b54a..88dbb2009080 100644 --- a/include/linux/mlx5/mlx5_ifc.h +++ b/include/linux/mlx5/mlx5_ifc.h @@ -8331,7 +8331,8 @@ struct mlx5_ifc_alloc_flow_counter_in_bits { u8 reserved_at_20[0x10]; u8 op_mod[0x10]; - u8 reserved_at_40[0x38]; + u8 reserved_at_40[0x33]; + u8 flow_counter_bulk_log_size[0x5]; u8 flow_counter_bulk[0x8]; }; From 95468f165d9379d7714d0ca099df5a306deff0e5 Mon Sep 17 00:00:00 2001 From: Dai Ngo Date: Tue, 18 Apr 2023 13:19:02 -0700 Subject: [PATCH 18/43] SUNRPC: remove the maximum number of retries in call_bind_status [ Upstream commit 691d0b782066a6eeeecbfceb7910a8f6184e6105 ] Currently call_bind_status places a hard limit of 3 to the number of retries on EACCES error. This limit was done to prevent NLM unlock requests from being hang forever when the server keeps returning garbage. However this change causes problem for cases when NLM service takes longer than 9 seconds to register with the port mapper after a restart. This patch removes this hard coded limit and let the RPC handles the retry based on the standard hard/soft task semantics. Fixes: 0b760113a3a1 ("NLM: Don't hang forever on NLM unlock requests") Reported-by: Helen Chao Tested-by: Helen Chao Signed-off-by: Dai Ngo Reviewed-by: Jeff Layton Signed-off-by: Anna Schumaker Signed-off-by: Sasha Levin --- include/linux/sunrpc/sched.h | 3 +-- net/sunrpc/clnt.c | 3 --- net/sunrpc/sched.c | 1 - 3 files changed, 1 insertion(+), 6 deletions(-) diff --git a/include/linux/sunrpc/sched.h b/include/linux/sunrpc/sched.h index df696efdd675..256dff36cf72 100644 --- a/include/linux/sunrpc/sched.h +++ b/include/linux/sunrpc/sched.h @@ -90,8 +90,7 @@ struct rpc_task { #endif unsigned char tk_priority : 2,/* Task priority */ tk_garb_retry : 2, - tk_cred_retry : 2, - tk_rebind_retry : 2; + tk_cred_retry : 2; }; typedef void (*rpc_action)(struct rpc_task *); diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c index c6e8bd78e35d..e1ce0f261f0b 100644 --- a/net/sunrpc/clnt.c +++ b/net/sunrpc/clnt.c @@ -1967,9 +1967,6 @@ call_bind_status(struct rpc_task *task) status = -EOPNOTSUPP; break; } - if (task->tk_rebind_retry == 0) - break; - task->tk_rebind_retry--; rpc_delay(task, 3*HZ); goto retry_timeout; case -ENOBUFS: diff --git a/net/sunrpc/sched.c b/net/sunrpc/sched.c index f0f55fbd1375..a00890962e11 100644 --- a/net/sunrpc/sched.c +++ b/net/sunrpc/sched.c @@ -796,7 +796,6 @@ rpc_init_task_statistics(struct rpc_task *task) /* Initialize retry counters */ task->tk_garb_retry = 2; task->tk_cred_retry = 2; - task->tk_rebind_retry = 2; /* starting timestamp */ task->tk_start = ktime_get(); From 3bc78eddf0a054a24ea977df7add6f8671e456a9 Mon Sep 17 00:00:00 2001 From: Mark Zhang Date: Thu, 20 Apr 2023 04:39:06 +0300 Subject: [PATCH 19/43] RDMA/mlx5: Use correct device num_ports when modify DC [ Upstream commit 746aa3c8cb1a650ff2583497ac646e505831b9b9 ] Just like other QP types, when modify DC, the port_num should be compared with dev->num_ports, instead of HCA_CAP.num_ports. Otherwise Multi-port vHCA on DC may not work. Fixes: 776a3906b692 ("IB/mlx5: Add support for DC target QP") Link: https://lore.kernel.org/r/20230420013906.1244185-1-markzhang@nvidia.com Signed-off-by: Mark Zhang Reviewed-by: Maor Gottlieb Signed-off-by: Jason Gunthorpe Signed-off-by: Sasha Levin --- drivers/infiniband/hw/mlx5/qp.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c index 0caff276f2c1..0c47e3e24b2a 100644 --- a/drivers/infiniband/hw/mlx5/qp.c +++ b/drivers/infiniband/hw/mlx5/qp.c @@ -4164,7 +4164,7 @@ static int mlx5_ib_modify_dct(struct ib_qp *ibqp, struct ib_qp_attr *attr, return -EINVAL; if (attr->port_num == 0 || - attr->port_num > MLX5_CAP_GEN(dev->mdev, num_ports)) { + attr->port_num > dev->num_ports) { mlx5_ib_dbg(dev, "invalid port number %d. number of ports is %d\n", attr->port_num, dev->num_ports); return -EINVAL; From 1fe1580521dddc86d26401e82c44c793e86a7a5f Mon Sep 17 00:00:00 2001 From: Qinrun Dai Date: Thu, 13 Apr 2023 13:50:37 +0000 Subject: [PATCH 20/43] clocksource/drivers/davinci: Fix memory leak in davinci_timer_register when init fails [ Upstream commit fb73556386e074e9bee9fa2d253aeaefe4e063e0 ] Smatch reports: drivers/clocksource/timer-davinci.c:332 davinci_timer_register() warn: 'base' from ioremap() not released on lines: 274. Fix this and other potential memory leak problems by adding a set of corresponding exit lables. Fixes: 721154f972aa ("clocksource/drivers/davinci: Add support for clockevents") Signed-off-by: Qinrun Dai Link: https://lore.kernel.org/r/20230413135037.1505799-1-flno@hust.edu.cn Signed-off-by: Daniel Lezcano Signed-off-by: Sasha Levin --- drivers/clocksource/timer-davinci.c | 30 +++++++++++++++++++++++------ 1 file changed, 24 insertions(+), 6 deletions(-) diff --git a/drivers/clocksource/timer-davinci.c b/drivers/clocksource/timer-davinci.c index bb4eee31ae08..3dc0c6ceed02 100644 --- a/drivers/clocksource/timer-davinci.c +++ b/drivers/clocksource/timer-davinci.c @@ -258,21 +258,25 @@ int __init davinci_timer_register(struct clk *clk, resource_size(&timer_cfg->reg), "davinci-timer")) { pr_err("Unable to request memory region\n"); - return -EBUSY; + rv = -EBUSY; + goto exit_clk_disable; } base = ioremap(timer_cfg->reg.start, resource_size(&timer_cfg->reg)); if (!base) { pr_err("Unable to map the register range\n"); - return -ENOMEM; + rv = -ENOMEM; + goto exit_mem_region; } davinci_timer_init(base); tick_rate = clk_get_rate(clk); clockevent = kzalloc(sizeof(*clockevent), GFP_KERNEL); - if (!clockevent) - return -ENOMEM; + if (!clockevent) { + rv = -ENOMEM; + goto exit_iounmap_base; + } clockevent->dev.name = "tim12"; clockevent->dev.features = CLOCK_EVT_FEAT_ONESHOT; @@ -297,7 +301,7 @@ int __init davinci_timer_register(struct clk *clk, "clockevent/tim12", clockevent); if (rv) { pr_err("Unable to request the clockevent interrupt\n"); - return rv; + goto exit_free_clockevent; } davinci_clocksource.dev.rating = 300; @@ -324,13 +328,27 @@ int __init davinci_timer_register(struct clk *clk, rv = clocksource_register_hz(&davinci_clocksource.dev, tick_rate); if (rv) { pr_err("Unable to register clocksource\n"); - return rv; + goto exit_free_irq; } sched_clock_register(davinci_timer_read_sched_clock, DAVINCI_TIMER_CLKSRC_BITS, tick_rate); return 0; + +exit_free_irq: + free_irq(timer_cfg->irq[DAVINCI_TIMER_CLOCKEVENT_IRQ].start, + clockevent); +exit_free_clockevent: + kfree(clockevent); +exit_iounmap_base: + iounmap(base); +exit_mem_region: + release_mem_region(timer_cfg->reg.start, + resource_size(&timer_cfg->reg)); +exit_clk_disable: + clk_disable_unprepare(clk); + return rv; } static int __init of_davinci_timer_register(struct device_node *np) From 8f009ae90b2bca5da5dbdf818debdbf917628d49 Mon Sep 17 00:00:00 2001 From: Stafford Horne Date: Sat, 11 Feb 2023 19:14:06 +0900 Subject: [PATCH 21/43] openrisc: Properly store r31 to pt_regs on unhandled exceptions [ Upstream commit 812489ac4dd91144a74ce65ecf232252a2e406fb ] In commit 91993c8c2ed5 ("openrisc: use shadow registers to save regs on exception") the unhandled exception path was changed to do an early store of r30 instead of r31. The entry code was not updated and r31 is not getting stored to pt_regs. This patch updates the entry handler to store r31 instead of r30. We also remove some misleading commented out store r30 and r31 instructrions. I noticed this while working on adding floating point exception handling, This issue probably would never impact anything since we kill the process or Oops right away on unhandled exceptions. Fixes: 91993c8c2ed5 ("openrisc: use shadow registers to save regs on exception") Signed-off-by: Stafford Horne Signed-off-by: Sasha Levin --- arch/openrisc/kernel/entry.S | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) diff --git a/arch/openrisc/kernel/entry.S b/arch/openrisc/kernel/entry.S index b42d32d79b2e..7257e942731d 100644 --- a/arch/openrisc/kernel/entry.S +++ b/arch/openrisc/kernel/entry.S @@ -173,7 +173,6 @@ handler: ;\ l.sw PT_GPR28(r1),r28 ;\ l.sw PT_GPR29(r1),r29 ;\ /* r30 already save */ ;\ -/* l.sw PT_GPR30(r1),r30*/ ;\ l.sw PT_GPR31(r1),r31 ;\ TRACE_IRQS_OFF_ENTRY ;\ /* Store -1 in orig_gpr11 for non-syscall exceptions */ ;\ @@ -211,9 +210,8 @@ handler: ;\ l.sw PT_GPR27(r1),r27 ;\ l.sw PT_GPR28(r1),r28 ;\ l.sw PT_GPR29(r1),r29 ;\ - /* r31 already saved */ ;\ - l.sw PT_GPR30(r1),r30 ;\ -/* l.sw PT_GPR31(r1),r31 */ ;\ + /* r30 already saved */ ;\ + l.sw PT_GPR31(r1),r31 ;\ /* Store -1 in orig_gpr11 for non-syscall exceptions */ ;\ l.addi r30,r0,-1 ;\ l.sw PT_ORIG_GPR11(r1),r30 ;\ From 770b0613637f59f3091dda1ff0c23671a5326b9c Mon Sep 17 00:00:00 2001 From: Ye Bin Date: Thu, 6 Apr 2023 11:16:27 +0000 Subject: [PATCH 22/43] ext4: fix use-after-free read in ext4_find_extent for bigalloc + inline [ Upstream commit 835659598c67907b98cd2aa57bb951dfaf675c69 ] Syzbot found the following issue: loop0: detected capacity change from 0 to 2048 EXT4-fs (loop0): mounted filesystem 00000000-0000-0000-0000-000000000000 without journal. Quota mode: none. ================================================================== BUG: KASAN: use-after-free in ext4_ext_binsearch_idx fs/ext4/extents.c:768 [inline] BUG: KASAN: use-after-free in ext4_find_extent+0x76e/0xd90 fs/ext4/extents.c:931 Read of size 4 at addr ffff888073644750 by task syz-executor420/5067 CPU: 0 PID: 5067 Comm: syz-executor420 Not tainted 6.2.0-rc1-syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 10/26/2022 Call Trace: __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0x1b1/0x290 lib/dump_stack.c:106 print_address_description+0x74/0x340 mm/kasan/report.c:306 print_report+0x107/0x1f0 mm/kasan/report.c:417 kasan_report+0xcd/0x100 mm/kasan/report.c:517 ext4_ext_binsearch_idx fs/ext4/extents.c:768 [inline] ext4_find_extent+0x76e/0xd90 fs/ext4/extents.c:931 ext4_clu_mapped+0x117/0x970 fs/ext4/extents.c:5809 ext4_insert_delayed_block fs/ext4/inode.c:1696 [inline] ext4_da_map_blocks fs/ext4/inode.c:1806 [inline] ext4_da_get_block_prep+0x9e8/0x13c0 fs/ext4/inode.c:1870 ext4_block_write_begin+0x6a8/0x2290 fs/ext4/inode.c:1098 ext4_da_write_begin+0x539/0x760 fs/ext4/inode.c:3082 generic_perform_write+0x2e4/0x5e0 mm/filemap.c:3772 ext4_buffered_write_iter+0x122/0x3a0 fs/ext4/file.c:285 ext4_file_write_iter+0x1d0/0x18f0 call_write_iter include/linux/fs.h:2186 [inline] new_sync_write fs/read_write.c:491 [inline] vfs_write+0x7dc/0xc50 fs/read_write.c:584 ksys_write+0x177/0x2a0 fs/read_write.c:637 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3d/0xb0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd RIP: 0033:0x7f4b7a9737b9 RSP: 002b:00007ffc5cac3668 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f4b7a9737b9 RDX: 00000000175d9003 RSI: 0000000020000200 RDI: 0000000000000004 RBP: 00007f4b7a933050 R08: 0000000000000000 R09: 0000000000000000 R10: 000000000000079f R11: 0000000000000246 R12: 00007f4b7a9330e0 R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000 Above issue is happens when enable bigalloc and inline data feature. As commit 131294c35ed6 fixed delayed allocation bug in ext4_clu_mapped for bigalloc + inline. But it only resolved issue when has inline data, if inline data has been converted to extent(ext4_da_convert_inline_data_to_extent) before writepages, there is no EXT4_STATE_MAY_INLINE_DATA flag. However i_data is still store inline data in this scene. Then will trigger UAF when find extent. To resolve above issue, there is need to add judge "ext4_has_inline_data(inode)" in ext4_clu_mapped(). Fixes: 131294c35ed6 ("ext4: fix delayed allocation bug in ext4_clu_mapped for bigalloc + inline") Reported-by: syzbot+bf4bb7731ef73b83a3b4@syzkaller.appspotmail.com Reviewed-by: Jan Kara Reviewed-by: Ye Bin Reviewed-by: Tudor Ambarus Tested-by: Tudor Ambarus Link: https://lore.kernel.org/r/20230406111627.1916759-1-tudor.ambarus@linaro.org Signed-off-by: Theodore Ts'o Signed-off-by: Sasha Levin --- fs/ext4/extents.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c index 6c06ce9dd6bd..2c2e1cc43e0e 100644 --- a/fs/ext4/extents.c +++ b/fs/ext4/extents.c @@ -5805,7 +5805,8 @@ int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu) * mapped - no physical clusters have been allocated, and the * file has no extents */ - if (ext4_test_inode_state(inode, EXT4_STATE_MAY_INLINE_DATA)) + if (ext4_test_inode_state(inode, EXT4_STATE_MAY_INLINE_DATA) || + ext4_has_inline_data(inode)) return 0; /* search for the extent closest to the first block in the cluster */ From 198fbdd10620e52424691a5251fd0cefca76784a Mon Sep 17 00:00:00 2001 From: Randy Dunlap Date: Sat, 25 Feb 2023 21:39:49 -0800 Subject: [PATCH 23/43] leds: TI_LMU_COMMON: select REGMAP instead of depending on it [ Upstream commit a61079efc87888587e463afaed82417b162fbd69 ] REGMAP is a hidden (not user visible) symbol. Users cannot set it directly thru "make *config", so drivers should select it instead of depending on it if they need it. Consistently using "select" or "depends on" can also help reduce Kconfig circular dependency issues. Therefore, change the use of "depends on REGMAP" to "select REGMAP". Fixes: 3fce8e1eb994 ("leds: TI LMU: Add common code for TI LMU devices") Signed-off-by: Randy Dunlap Acked-by: Pavel Machek Signed-off-by: Lee Jones Link: https://lore.kernel.org/r/20230226053953.4681-5-rdunlap@infradead.org Signed-off-by: Sasha Levin --- drivers/leds/Kconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig index 56e8198e13d1..ad84be4f6817 100644 --- a/drivers/leds/Kconfig +++ b/drivers/leds/Kconfig @@ -871,7 +871,7 @@ config LEDS_SPI_BYTE config LEDS_TI_LMU_COMMON tristate "LED driver for TI LMU" depends on LEDS_CLASS - depends on REGMAP + select REGMAP help Say Y to enable the LED driver for TI LMU devices. This supports common features between the TI LM3532, LM3631, LM3632, From 033f00eb8969e99d9a06ac0c77adcab3501ef1fe Mon Sep 17 00:00:00 2001 From: Christophe JAILLET Date: Sun, 26 Mar 2023 09:06:37 +0200 Subject: [PATCH 24/43] dmaengine: mv_xor_v2: Fix an error code. [ Upstream commit 827026ae2e56ec05ef1155661079badbbfc0b038 ] If the probe is deferred, -EPROBE_DEFER should be returned, not +EPROBE_DEFER. Fixes: 3cd2c313f1d6 ("dmaengine: mv_xor_v2: Fix clock resource by adding a register clock") Signed-off-by: Christophe JAILLET Link: https://lore.kernel.org/r/201170dff832a3c496d125772e10070cd834ebf2.1679814350.git.christophe.jaillet@wanadoo.fr Signed-off-by: Vinod Koul Signed-off-by: Sasha Levin --- drivers/dma/mv_xor_v2.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/dma/mv_xor_v2.c b/drivers/dma/mv_xor_v2.c index 4800c596433a..9f3e011fbd91 100644 --- a/drivers/dma/mv_xor_v2.c +++ b/drivers/dma/mv_xor_v2.c @@ -756,7 +756,7 @@ static int mv_xor_v2_probe(struct platform_device *pdev) xor_dev->clk = devm_clk_get(&pdev->dev, NULL); if (PTR_ERR(xor_dev->clk) == -EPROBE_DEFER) { - ret = EPROBE_DEFER; + ret = -EPROBE_DEFER; goto disable_reg_clk; } if (!IS_ERR(xor_dev->clk)) { From 795cec288e9c3661caff2b9e28012da9e07ea5e5 Mon Sep 17 00:00:00 2001 From: "H. Nikolaus Schaller" Date: Sun, 2 Apr 2023 13:12:59 +0200 Subject: [PATCH 25/43] leds: tca6507: Fix error handling of using fwnode_property_read_string MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit [ Upstream commit c1087c29e96a48e9080377e168d35dcb52fb068b ] Commit 96f524105b9c ("leds: tca6507: use fwnode API instead of OF") changed to fwnode API but did not take into account that a missing property "linux,default-trigger" now seems to return an error and as a side effect sets value to -1. This seems to be different from of_get_property() which always returned NULL in any case of error. Neglecting this side-effect leads to [ 11.201965] Unable to handle kernel paging request at virtual address ffffffff when read in the strcmp() of led_trigger_set_default() if there is no led-trigger defined in the DTS. I don't know if this was recently introduced somewhere in the fwnode lib or if the effect was missed in initial testing. Anyways it seems to be a bug to ignore the error return value of an optional value here in the driver. Fixes: 96f524105b9c ("leds: tca6507: use fwnode API instead of OF") Signed-off-by: H. Nikolaus Schaller Acked-by: Pavel Machek Reviewed-by: Marek Behún Signed-off-by: Lee Jones Link: https://lore.kernel.org/r/cbae7617db83113de726fcc423a805ebaa1bfca6.1680433978.git.hns@goldelico.com Signed-off-by: Sasha Levin --- drivers/leds/leds-tca6507.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/leds/leds-tca6507.c b/drivers/leds/leds-tca6507.c index 225b765830bd..caad9d3e0eac 100644 --- a/drivers/leds/leds-tca6507.c +++ b/drivers/leds/leds-tca6507.c @@ -696,8 +696,9 @@ tca6507_led_dt_init(struct device *dev) if (fwnode_property_read_string(child, "label", &led.name)) led.name = fwnode_get_name(child); - fwnode_property_read_string(child, "linux,default-trigger", - &led.default_trigger); + if (fwnode_property_read_string(child, "linux,default-trigger", + &led.default_trigger)) + led.default_trigger = NULL; led.flags = 0; if (fwnode_property_match_string(child, "compatible", From 060bd30bf70515bd862c6a384de077f0afdcc869 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Uwe=20Kleine-K=C3=B6nig?= Date: Wed, 7 Jul 2021 18:28:27 +0200 Subject: [PATCH 26/43] pwm: mtk-disp: Don't check the return code of pwmchip_remove() MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit [ Upstream commit 9b7b5736ffd5da6f8f6329ebe5f1829cbcf8afae ] pwmchip_remove() returns always 0. Don't use the value to make it possible to eventually change the function to return void. Also the driver core ignores the return value of mtk_disp_pwm_remove(). Signed-off-by: Uwe Kleine-König Signed-off-by: Thierry Reding Stable-dep-of: 36dd7f530ae7 ("pwm: mtk-disp: Disable shadow registers before setting backlight values") Signed-off-by: Sasha Levin --- drivers/pwm/pwm-mtk-disp.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/drivers/pwm/pwm-mtk-disp.c b/drivers/pwm/pwm-mtk-disp.c index 83b8be0209b7..af63aaab8e15 100644 --- a/drivers/pwm/pwm-mtk-disp.c +++ b/drivers/pwm/pwm-mtk-disp.c @@ -240,13 +240,12 @@ disable_clk_main: static int mtk_disp_pwm_remove(struct platform_device *pdev) { struct mtk_disp_pwm *mdp = platform_get_drvdata(pdev); - int ret; - ret = pwmchip_remove(&mdp->chip); + pwmchip_remove(&mdp->chip); clk_unprepare(mdp->clk_mm); clk_unprepare(mdp->clk_main); - return ret; + return 0; } static const struct mtk_pwm_data mt2701_pwm_data = { From 6d13804388d6eb4b74c633a18a3bbd2d609f91bf Mon Sep 17 00:00:00 2001 From: Jitao Shi Date: Sun, 8 Aug 2021 21:24:29 +0800 Subject: [PATCH 27/43] pwm: mtk-disp: Adjust the clocks to avoid them mismatch [ Upstream commit d7a4e582587d97a586b1f7709e3bddcf35928e96 ] The clks "main" and "mm" are prepared in .probe() (and unprepared in .remove()). This results in the clocks being on during suspend which results in unnecessarily increased power consumption. Remove the clock operations from .probe() and .remove(). Add the clk_prepare_enable() in .enable() and the clk_disable_unprepare() in .disable(). Signed-off-by: Jitao Shi [thierry.reding@gmail.com: squashed in fixup patch] Signed-off-by: Thierry Reding Stable-dep-of: 36dd7f530ae7 ("pwm: mtk-disp: Disable shadow registers before setting backlight values") Signed-off-by: Sasha Levin --- drivers/pwm/pwm-mtk-disp.c | 95 +++++++++++++++++--------------------- 1 file changed, 43 insertions(+), 52 deletions(-) diff --git a/drivers/pwm/pwm-mtk-disp.c b/drivers/pwm/pwm-mtk-disp.c index af63aaab8e15..a594a0fdddd7 100644 --- a/drivers/pwm/pwm-mtk-disp.c +++ b/drivers/pwm/pwm-mtk-disp.c @@ -74,6 +74,19 @@ static int mtk_disp_pwm_config(struct pwm_chip *chip, struct pwm_device *pwm, u64 div, rate; int err; + err = clk_prepare_enable(mdp->clk_main); + if (err < 0) { + dev_err(chip->dev, "Can't enable mdp->clk_main: %pe\n", ERR_PTR(err)); + return err; + } + + err = clk_prepare_enable(mdp->clk_mm); + if (err < 0) { + dev_err(chip->dev, "Can't enable mdp->clk_mm: %pe\n", ERR_PTR(err)); + clk_disable_unprepare(mdp->clk_main); + return err; + } + /* * Find period, high_width and clk_div to suit duty_ns and period_ns. * Calculate proper div value to keep period value in the bound. @@ -87,8 +100,11 @@ static int mtk_disp_pwm_config(struct pwm_chip *chip, struct pwm_device *pwm, rate = clk_get_rate(mdp->clk_main); clk_div = div_u64(rate * period_ns, NSEC_PER_SEC) >> PWM_PERIOD_BIT_WIDTH; - if (clk_div > PWM_CLKDIV_MAX) + if (clk_div > PWM_CLKDIV_MAX) { + clk_disable_unprepare(mdp->clk_mm); + clk_disable_unprepare(mdp->clk_main); return -EINVAL; + } div = NSEC_PER_SEC * (clk_div + 1); period = div64_u64(rate * period_ns, div); @@ -98,16 +114,6 @@ static int mtk_disp_pwm_config(struct pwm_chip *chip, struct pwm_device *pwm, high_width = div64_u64(rate * duty_ns, div); value = period | (high_width << PWM_HIGH_WIDTH_SHIFT); - err = clk_enable(mdp->clk_main); - if (err < 0) - return err; - - err = clk_enable(mdp->clk_mm); - if (err < 0) { - clk_disable(mdp->clk_main); - return err; - } - mtk_disp_pwm_update_bits(mdp, mdp->data->con0, PWM_CLKDIV_MASK, clk_div << PWM_CLKDIV_SHIFT); @@ -122,10 +128,21 @@ static int mtk_disp_pwm_config(struct pwm_chip *chip, struct pwm_device *pwm, mtk_disp_pwm_update_bits(mdp, mdp->data->commit, mdp->data->commit_mask, 0x0); + } else { + /* + * For MT2701, disable double buffer before writing register + * and select manual mode and use PWM_PERIOD/PWM_HIGH_WIDTH. + */ + mtk_disp_pwm_update_bits(mdp, mdp->data->bls_debug, + mdp->data->bls_debug_mask, + mdp->data->bls_debug_mask); + mtk_disp_pwm_update_bits(mdp, mdp->data->con0, + mdp->data->con0_sel, + mdp->data->con0_sel); } - clk_disable(mdp->clk_mm); - clk_disable(mdp->clk_main); + clk_disable_unprepare(mdp->clk_mm); + clk_disable_unprepare(mdp->clk_main); return 0; } @@ -135,13 +152,16 @@ static int mtk_disp_pwm_enable(struct pwm_chip *chip, struct pwm_device *pwm) struct mtk_disp_pwm *mdp = to_mtk_disp_pwm(chip); int err; - err = clk_enable(mdp->clk_main); - if (err < 0) - return err; - - err = clk_enable(mdp->clk_mm); + err = clk_prepare_enable(mdp->clk_main); if (err < 0) { - clk_disable(mdp->clk_main); + dev_err(chip->dev, "Can't enable mdp->clk_main: %pe\n", ERR_PTR(err)); + return err; + } + + err = clk_prepare_enable(mdp->clk_mm); + if (err < 0) { + dev_err(chip->dev, "Can't enable mdp->clk_mm: %pe\n", ERR_PTR(err)); + clk_disable_unprepare(mdp->clk_main); return err; } @@ -158,8 +178,8 @@ static void mtk_disp_pwm_disable(struct pwm_chip *chip, struct pwm_device *pwm) mtk_disp_pwm_update_bits(mdp, DISP_PWM_EN, mdp->data->enable_mask, 0x0); - clk_disable(mdp->clk_mm); - clk_disable(mdp->clk_main); + clk_disable_unprepare(mdp->clk_mm); + clk_disable_unprepare(mdp->clk_main); } static const struct pwm_ops mtk_disp_pwm_ops = { @@ -194,14 +214,6 @@ static int mtk_disp_pwm_probe(struct platform_device *pdev) if (IS_ERR(mdp->clk_mm)) return PTR_ERR(mdp->clk_mm); - ret = clk_prepare(mdp->clk_main); - if (ret < 0) - return ret; - - ret = clk_prepare(mdp->clk_mm); - if (ret < 0) - goto disable_clk_main; - mdp->chip.dev = &pdev->dev; mdp->chip.ops = &mtk_disp_pwm_ops; mdp->chip.base = -1; @@ -209,32 +221,13 @@ static int mtk_disp_pwm_probe(struct platform_device *pdev) ret = pwmchip_add(&mdp->chip); if (ret < 0) { - dev_err(&pdev->dev, "pwmchip_add() failed: %d\n", ret); - goto disable_clk_mm; + dev_err(&pdev->dev, "pwmchip_add() failed: %pe\n", ERR_PTR(ret)); + return ret; } platform_set_drvdata(pdev, mdp); - /* - * For MT2701, disable double buffer before writing register - * and select manual mode and use PWM_PERIOD/PWM_HIGH_WIDTH. - */ - if (!mdp->data->has_commit) { - mtk_disp_pwm_update_bits(mdp, mdp->data->bls_debug, - mdp->data->bls_debug_mask, - mdp->data->bls_debug_mask); - mtk_disp_pwm_update_bits(mdp, mdp->data->con0, - mdp->data->con0_sel, - mdp->data->con0_sel); - } - return 0; - -disable_clk_mm: - clk_unprepare(mdp->clk_mm); -disable_clk_main: - clk_unprepare(mdp->clk_main); - return ret; } static int mtk_disp_pwm_remove(struct platform_device *pdev) @@ -242,8 +235,6 @@ static int mtk_disp_pwm_remove(struct platform_device *pdev) struct mtk_disp_pwm *mdp = platform_get_drvdata(pdev); pwmchip_remove(&mdp->chip); - clk_unprepare(mdp->clk_mm); - clk_unprepare(mdp->clk_main); return 0; } From 45e4c00940beecc3a81934d8e4d58ce67d9f88c2 Mon Sep 17 00:00:00 2001 From: AngeloGioacchino Del Regno Date: Mon, 3 Apr 2023 15:30:53 +0200 Subject: [PATCH 28/43] pwm: mtk-disp: Disable shadow registers before setting backlight values MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit [ Upstream commit 36dd7f530ae7d9ce9e853ffb8aa337de65c6600b ] If shadow registers usage is not desired, disable that before performing any write to CON0/1 registers in the .apply() callback, otherwise we may lose clkdiv or period/width updates. Fixes: cd4b45ac449a ("pwm: Add MediaTek MT2701 display PWM driver support") Signed-off-by: AngeloGioacchino Del Regno Reviewed-by: Nícolas F. R. A. Prado Tested-by: Nícolas F. R. A. Prado Reviewed-by: Alexandre Mergnat Tested-by: Alexandre Mergnat Signed-off-by: Thierry Reding Signed-off-by: Sasha Levin --- drivers/pwm/pwm-mtk-disp.c | 24 +++++++++++++----------- 1 file changed, 13 insertions(+), 11 deletions(-) diff --git a/drivers/pwm/pwm-mtk-disp.c b/drivers/pwm/pwm-mtk-disp.c index a594a0fdddd7..327c780d433d 100644 --- a/drivers/pwm/pwm-mtk-disp.c +++ b/drivers/pwm/pwm-mtk-disp.c @@ -114,6 +114,19 @@ static int mtk_disp_pwm_config(struct pwm_chip *chip, struct pwm_device *pwm, high_width = div64_u64(rate * duty_ns, div); value = period | (high_width << PWM_HIGH_WIDTH_SHIFT); + if (mdp->data->bls_debug && !mdp->data->has_commit) { + /* + * For MT2701, disable double buffer before writing register + * and select manual mode and use PWM_PERIOD/PWM_HIGH_WIDTH. + */ + mtk_disp_pwm_update_bits(mdp, mdp->data->bls_debug, + mdp->data->bls_debug_mask, + mdp->data->bls_debug_mask); + mtk_disp_pwm_update_bits(mdp, mdp->data->con0, + mdp->data->con0_sel, + mdp->data->con0_sel); + } + mtk_disp_pwm_update_bits(mdp, mdp->data->con0, PWM_CLKDIV_MASK, clk_div << PWM_CLKDIV_SHIFT); @@ -128,17 +141,6 @@ static int mtk_disp_pwm_config(struct pwm_chip *chip, struct pwm_device *pwm, mtk_disp_pwm_update_bits(mdp, mdp->data->commit, mdp->data->commit_mask, 0x0); - } else { - /* - * For MT2701, disable double buffer before writing register - * and select manual mode and use PWM_PERIOD/PWM_HIGH_WIDTH. - */ - mtk_disp_pwm_update_bits(mdp, mdp->data->bls_debug, - mdp->data->bls_debug_mask, - mdp->data->bls_debug_mask); - mtk_disp_pwm_update_bits(mdp, mdp->data->con0, - mdp->data->con0_sel, - mdp->data->con0_sel); } clk_disable_unprepare(mdp->clk_mm); From da67d60467dd9308a98c6213213a4c39dd5d2a24 Mon Sep 17 00:00:00 2001 From: Gaosheng Cui Date: Tue, 29 Nov 2022 19:16:34 +0800 Subject: [PATCH 29/43] phy: tegra: xusb: Add missing tegra_xusb_port_unregister for usb2_port and ulpi_port [ Upstream commit e024854048e733391b31fe5a398704b31b9af803 ] The tegra_xusb_port_unregister should be called when usb2_port and ulpi_port map fails in tegra_xusb_add_usb2_port() or in tegra_xusb_add_ulpi_port(), fix it. Fixes: 53d2a715c240 ("phy: Add Tegra XUSB pad controller support") Signed-off-by: Gaosheng Cui Acked-by: Thierry Reding Link: https://lore.kernel.org/r/20221129111634.1547747-1-cuigaosheng1@huawei.com Signed-off-by: Vinod Koul Signed-off-by: Sasha Levin --- drivers/phy/tegra/xusb.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/phy/tegra/xusb.c b/drivers/phy/tegra/xusb.c index 181a1be5f491..d07f33ec7939 100644 --- a/drivers/phy/tegra/xusb.c +++ b/drivers/phy/tegra/xusb.c @@ -775,6 +775,7 @@ static int tegra_xusb_add_usb2_port(struct tegra_xusb_padctl *padctl, usb2->base.lane = usb2->base.ops->map(&usb2->base); if (IS_ERR(usb2->base.lane)) { err = PTR_ERR(usb2->base.lane); + tegra_xusb_port_unregister(&usb2->base); goto out; } @@ -841,6 +842,7 @@ static int tegra_xusb_add_ulpi_port(struct tegra_xusb_padctl *padctl, ulpi->base.lane = ulpi->base.ops->map(&ulpi->base); if (IS_ERR(ulpi->base.lane)) { err = PTR_ERR(ulpi->base.lane); + tegra_xusb_port_unregister(&ulpi->base); goto out; } From b29543194594903242907e490e8e50ec05561aee Mon Sep 17 00:00:00 2001 From: Shunsuke Mie Date: Tue, 11 Apr 2023 19:17:57 +0900 Subject: [PATCH 30/43] dmaengine: dw-edma: Fix to change for continuous transfer [ Upstream commit a251994a441ee0a69ba7062c8cd2d08ead3db379 ] The dw-edma driver stops after processing a DMA request even if a request remains in the issued queue, which is not the expected behavior. The DMA engine API requires continuous processing. Add a trigger to start after one processing finished if there are requests remain. Fixes: e63d79d1ffcd ("dmaengine: Add Synopsys eDMA IP core driver") Signed-off-by: Shunsuke Mie Link: https://lore.kernel.org/r/20230411101758.438472-1-mie@igel.co.jp Signed-off-by: Vinod Koul Signed-off-by: Sasha Levin --- drivers/dma/dw-edma/dw-edma-core.c | 20 +++++++++++--------- 1 file changed, 11 insertions(+), 9 deletions(-) diff --git a/drivers/dma/dw-edma/dw-edma-core.c b/drivers/dma/dw-edma/dw-edma-core.c index d7ed50f8b929..a3790f7b1b53 100644 --- a/drivers/dma/dw-edma/dw-edma-core.c +++ b/drivers/dma/dw-edma/dw-edma-core.c @@ -166,7 +166,7 @@ static void vchan_free_desc(struct virt_dma_desc *vdesc) dw_edma_free_desc(vd2dw_edma_desc(vdesc)); } -static void dw_edma_start_transfer(struct dw_edma_chan *chan) +static int dw_edma_start_transfer(struct dw_edma_chan *chan) { struct dw_edma_chunk *child; struct dw_edma_desc *desc; @@ -174,16 +174,16 @@ static void dw_edma_start_transfer(struct dw_edma_chan *chan) vd = vchan_next_desc(&chan->vc); if (!vd) - return; + return 0; desc = vd2dw_edma_desc(vd); if (!desc) - return; + return 0; child = list_first_entry_or_null(&desc->chunk->list, struct dw_edma_chunk, list); if (!child) - return; + return 0; dw_edma_v0_core_start(child, !desc->xfer_sz); desc->xfer_sz += child->ll_region.sz; @@ -191,6 +191,8 @@ static void dw_edma_start_transfer(struct dw_edma_chan *chan) list_del(&child->list); kfree(child); desc->chunks_alloc--; + + return 1; } static int dw_edma_device_config(struct dma_chan *dchan, @@ -497,14 +499,14 @@ static void dw_edma_done_interrupt(struct dw_edma_chan *chan) switch (chan->request) { case EDMA_REQ_NONE: desc = vd2dw_edma_desc(vd); - if (desc->chunks_alloc) { - chan->status = EDMA_ST_BUSY; - dw_edma_start_transfer(chan); - } else { + if (!desc->chunks_alloc) { list_del(&vd->node); vchan_cookie_complete(vd); - chan->status = EDMA_ST_IDLE; } + + /* Continue transferring if there are remaining chunks or issued requests. + */ + chan->status = dw_edma_start_transfer(chan) ? EDMA_ST_BUSY : EDMA_ST_IDLE; break; case EDMA_REQ_STOP: From 4f6303fd8f907adf4785aaeb6f62e470115f1901 Mon Sep 17 00:00:00 2001 From: Shunsuke Mie Date: Tue, 11 Apr 2023 19:17:58 +0900 Subject: [PATCH 31/43] dmaengine: dw-edma: Fix to enable to issue dma request on DMA processing [ Upstream commit 970b17dfe264a9085ba4e593730ecfd496b950ab ] The issue_pending request is ignored while driver is processing a DMA request. Fix to issue the pending requests on any dma channel status. Fixes: e63d79d1ffcd ("dmaengine: Add Synopsys eDMA IP core driver") Signed-off-by: Shunsuke Mie Link: https://lore.kernel.org/r/20230411101758.438472-2-mie@igel.co.jp Signed-off-by: Vinod Koul Signed-off-by: Sasha Levin --- drivers/dma/dw-edma/dw-edma-core.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/drivers/dma/dw-edma/dw-edma-core.c b/drivers/dma/dw-edma/dw-edma-core.c index a3790f7b1b53..f91dbf43a598 100644 --- a/drivers/dma/dw-edma/dw-edma-core.c +++ b/drivers/dma/dw-edma/dw-edma-core.c @@ -276,9 +276,12 @@ static void dw_edma_device_issue_pending(struct dma_chan *dchan) struct dw_edma_chan *chan = dchan2dw_edma_chan(dchan); unsigned long flags; + if (!chan->configured) + return; + spin_lock_irqsave(&chan->vc.lock, flags); - if (chan->configured && chan->request == EDMA_REQ_NONE && - chan->status == EDMA_ST_IDLE && vchan_issue_pending(&chan->vc)) { + if (vchan_issue_pending(&chan->vc) && chan->request == EDMA_REQ_NONE && + chan->status == EDMA_ST_IDLE) { chan->status = EDMA_ST_BUSY; dw_edma_start_transfer(chan); } From 08012d9edc521e8ed66de6ba7cb49d73862d4b98 Mon Sep 17 00:00:00 2001 From: Claudiu Beznea Date: Tue, 14 Feb 2023 17:18:25 +0200 Subject: [PATCH 32/43] dmaengine: at_xdmac: do not enable all cyclic channels [ Upstream commit f8435befd81dd85b7b610598551fadf675849bc1 ] Do not global enable all the cyclic channels in at_xdmac_resume(). Instead save the global status in at_xdmac_suspend() and re-enable the cyclic channel only if it was active before suspend. Fixes: e1f7c9eee707 ("dmaengine: at_xdmac: creation of the atmel eXtended DMA Controller driver") Signed-off-by: Claudiu Beznea Link: https://lore.kernel.org/r/20230214151827.1050280-6-claudiu.beznea@microchip.com Signed-off-by: Vinod Koul Signed-off-by: Sasha Levin --- drivers/dma/at_xdmac.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/drivers/dma/at_xdmac.c b/drivers/dma/at_xdmac.c index b5d691ae45dc..1fe006cc643e 100644 --- a/drivers/dma/at_xdmac.c +++ b/drivers/dma/at_xdmac.c @@ -212,6 +212,7 @@ struct at_xdmac { int irq; struct clk *clk; u32 save_gim; + u32 save_gs; struct dma_pool *at_xdmac_desc_pool; struct at_xdmac_chan chan[]; }; @@ -1910,6 +1911,7 @@ static int atmel_xdmac_suspend(struct device *dev) } } atxdmac->save_gim = at_xdmac_read(atxdmac, AT_XDMAC_GIM); + atxdmac->save_gs = at_xdmac_read(atxdmac, AT_XDMAC_GS); at_xdmac_off(atxdmac); clk_disable_unprepare(atxdmac->clk); @@ -1946,7 +1948,8 @@ static int atmel_xdmac_resume(struct device *dev) at_xdmac_chan_write(atchan, AT_XDMAC_CNDC, atchan->save_cndc); at_xdmac_chan_write(atchan, AT_XDMAC_CIE, atchan->save_cim); wmb(); - at_xdmac_write(atxdmac, AT_XDMAC_GE, atchan->mask); + if (atxdmac->save_gs & atchan->mask) + at_xdmac_write(atxdmac, AT_XDMAC_GE, atchan->mask); } } return 0; From 15da2acad5156937c5b62444b18c6c20dd20f64d Mon Sep 17 00:00:00 2001 From: Kang Chen Date: Wed, 19 Apr 2023 10:07:48 +0800 Subject: [PATCH 33/43] thermal/drivers/mediatek: Use devm_of_iomap to avoid resource leak in mtk_thermal_probe [ Upstream commit f05c7b7d9ea9477fcc388476c6f4ade8c66d2d26 ] Smatch reports: 1. mtk_thermal_probe() warn: 'apmixed_base' from of_iomap() not released. 2. mtk_thermal_probe() warn: 'auxadc_base' from of_iomap() not released. The original code forgets to release iomap resource when handling errors, fix it by switch to devm_of_iomap. Fixes: 89945047b166 ("thermal: mediatek: Add tsensor support for V2 thermal system") Signed-off-by: Kang Chen Reviewed-by: Dongliang Mu Reviewed-by: AngeloGioacchino Del Regno Signed-off-by: Daniel Lezcano Link: https://lore.kernel.org/r/20230419020749.621257-1-void0red@hust.edu.cn Signed-off-by: Sasha Levin --- drivers/thermal/mtk_thermal.c | 14 ++++++++++++-- 1 file changed, 12 insertions(+), 2 deletions(-) diff --git a/drivers/thermal/mtk_thermal.c b/drivers/thermal/mtk_thermal.c index 0bd7aa564bc2..9fe169dbed88 100644 --- a/drivers/thermal/mtk_thermal.c +++ b/drivers/thermal/mtk_thermal.c @@ -1026,7 +1026,12 @@ static int mtk_thermal_probe(struct platform_device *pdev) return -ENODEV; } - auxadc_base = of_iomap(auxadc, 0); + auxadc_base = devm_of_iomap(&pdev->dev, auxadc, 0, NULL); + if (IS_ERR(auxadc_base)) { + of_node_put(auxadc); + return PTR_ERR(auxadc_base); + } + auxadc_phys_base = of_get_phys_base(auxadc); of_node_put(auxadc); @@ -1042,7 +1047,12 @@ static int mtk_thermal_probe(struct platform_device *pdev) return -ENODEV; } - apmixed_base = of_iomap(apmixedsys, 0); + apmixed_base = devm_of_iomap(&pdev->dev, apmixedsys, 0, NULL); + if (IS_ERR(apmixed_base)) { + of_node_put(apmixedsys); + return PTR_ERR(apmixed_base); + } + apmixed_phys_base = of_get_phys_base(apmixedsys); of_node_put(apmixedsys); From 6697a3b0ed8397ed7d837c65b82119f8530c0200 Mon Sep 17 00:00:00 2001 From: Matthias Schiffer Date: Mon, 20 Feb 2023 12:25:44 +0100 Subject: [PATCH 34/43] mfd: tqmx86: Do not access I2C_DETECT register through io_base [ Upstream commit 1be1b23696b3d4b0231c694f5e0767b4471d33a9 ] The I2C_DETECT register is at IO port 0x1a7, which is outside the range passed to devm_ioport_map() for io_base, and was only working because there aren't actually any bounds checks for IO port accesses. Extending the range does not seem like a good solution here, as it would then conflict with the IO resource assigned to the I2C controller. As this is just a one-off access during probe, use a simple inb() instead. While we're at it, drop the unused define TQMX86_REG_I2C_INT_EN. Fixes: 2f17dd34ffed ("mfd: tqmx86: IO controller with I2C, Wachdog and GPIO") Signed-off-by: Matthias Schiffer Reviewed-by: Andrew Lunn Signed-off-by: Lee Jones Link: https://lore.kernel.org/r/e8300a30f0791afb67d79db8089fb6004855f378.1676892223.git.matthias.schiffer@ew.tq-group.com Signed-off-by: Sasha Levin --- drivers/mfd/tqmx86.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) diff --git a/drivers/mfd/tqmx86.c b/drivers/mfd/tqmx86.c index 732013f40e4e..99a59341e449 100644 --- a/drivers/mfd/tqmx86.c +++ b/drivers/mfd/tqmx86.c @@ -45,9 +45,8 @@ #define TQMX86_REG_IO_EXT_INT_MASK 0x3 #define TQMX86_REG_IO_EXT_INT_GPIO_SHIFT 4 -#define TQMX86_REG_I2C_DETECT 0x47 +#define TQMX86_REG_I2C_DETECT 0x1a7 #define TQMX86_REG_I2C_DETECT_SOFT 0xa5 -#define TQMX86_REG_I2C_INT_EN 0x49 static uint gpio_irq; module_param(gpio_irq, uint, 0); @@ -195,7 +194,12 @@ static int tqmx86_probe(struct platform_device *pdev) "Found %s - Board ID %d, PCB Revision %d, PLD Revision %d\n", board_name, board_id, rev >> 4, rev & 0xf); - i2c_det = ioread8(io_base + TQMX86_REG_I2C_DETECT); + /* + * The I2C_DETECT register is in the range assigned to the I2C driver + * later, so we don't extend TQMX86_IOSIZE. Use inb() for this one-off + * access instead of ioport_map + unmap. + */ + i2c_det = inb(TQMX86_REG_I2C_DETECT); if (gpio_irq_cfg) { io_ext_int_val = From 4be49b7c2457d49ea1f590ce961e2a28e61a479f Mon Sep 17 00:00:00 2001 From: Matthias Schiffer Date: Fri, 16 Jul 2021 12:00:49 +0200 Subject: [PATCH 35/43] mfd: tqmx86: Remove incorrect TQMx90UC board ID [ Upstream commit 16b2ad150f74db0eb91f445061f16140b5aaa650 ] No TQMx90UC exists at the moment, and it is undecided whether ID 10 will be used eventually (and if it is, how that SoM will be named). Signed-off-by: Matthias Schiffer Reviewed-by: Andrew Lunn Signed-off-by: Lee Jones Stable-dep-of: 051c69ff4f60 ("mfd: tqmx86: Specify IO port register range more precisely") Signed-off-by: Sasha Levin --- drivers/mfd/tqmx86.c | 4 ---- 1 file changed, 4 deletions(-) diff --git a/drivers/mfd/tqmx86.c b/drivers/mfd/tqmx86.c index 99a59341e449..c5c5846dcf99 100644 --- a/drivers/mfd/tqmx86.c +++ b/drivers/mfd/tqmx86.c @@ -35,7 +35,6 @@ #define TQMX86_REG_BOARD_ID_E39x 7 #define TQMX86_REG_BOARD_ID_70EB 8 #define TQMX86_REG_BOARD_ID_80UC 9 -#define TQMX86_REG_BOARD_ID_90UC 10 #define TQMX86_REG_BOARD_REV 0x21 #define TQMX86_REG_IO_EXT_INT 0x26 #define TQMX86_REG_IO_EXT_INT_NONE 0 @@ -127,8 +126,6 @@ static const char *tqmx86_board_id_to_name(u8 board_id) return "TQMx70EB"; case TQMX86_REG_BOARD_ID_80UC: return "TQMx80UC"; - case TQMX86_REG_BOARD_ID_90UC: - return "TQMx90UC"; default: return "Unknown"; } @@ -141,7 +138,6 @@ static int tqmx86_board_id_to_clk_rate(u8 board_id) case TQMX86_REG_BOARD_ID_60EB: case TQMX86_REG_BOARD_ID_70EB: case TQMX86_REG_BOARD_ID_80UC: - case TQMX86_REG_BOARD_ID_90UC: return 24000; case TQMX86_REG_BOARD_ID_E39M: case TQMX86_REG_BOARD_ID_E39C: From 640a1f7e36e8b60b94a68bd4525a536647597fec Mon Sep 17 00:00:00 2001 From: Matthias Schiffer Date: Fri, 16 Jul 2021 12:00:51 +0200 Subject: [PATCH 36/43] mfd: tqmx86: Add support for TQMx110EB and TQMxE40x [ Upstream commit 3da48ccb1d0f3b53b1e8c9022edbedc2a6e3f50a ] Add the board IDs for the TQMx110EB and the TQMxE40x family. All use a 24MHz LPC clock. Signed-off-by: Matthias Schiffer Reviewed-by: Andrew Lunn Signed-off-by: Lee Jones Stable-dep-of: 051c69ff4f60 ("mfd: tqmx86: Specify IO port register range more precisely") Signed-off-by: Sasha Levin --- drivers/mfd/tqmx86.c | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+) diff --git a/drivers/mfd/tqmx86.c b/drivers/mfd/tqmx86.c index c5c5846dcf99..1db95aed5012 100644 --- a/drivers/mfd/tqmx86.c +++ b/drivers/mfd/tqmx86.c @@ -35,6 +35,11 @@ #define TQMX86_REG_BOARD_ID_E39x 7 #define TQMX86_REG_BOARD_ID_70EB 8 #define TQMX86_REG_BOARD_ID_80UC 9 +#define TQMX86_REG_BOARD_ID_110EB 11 +#define TQMX86_REG_BOARD_ID_E40M 12 +#define TQMX86_REG_BOARD_ID_E40S 13 +#define TQMX86_REG_BOARD_ID_E40C1 14 +#define TQMX86_REG_BOARD_ID_E40C2 15 #define TQMX86_REG_BOARD_REV 0x21 #define TQMX86_REG_IO_EXT_INT 0x26 #define TQMX86_REG_IO_EXT_INT_NONE 0 @@ -126,6 +131,16 @@ static const char *tqmx86_board_id_to_name(u8 board_id) return "TQMx70EB"; case TQMX86_REG_BOARD_ID_80UC: return "TQMx80UC"; + case TQMX86_REG_BOARD_ID_110EB: + return "TQMx110EB"; + case TQMX86_REG_BOARD_ID_E40M: + return "TQMxE40M"; + case TQMX86_REG_BOARD_ID_E40S: + return "TQMxE40S"; + case TQMX86_REG_BOARD_ID_E40C1: + return "TQMxE40C1"; + case TQMX86_REG_BOARD_ID_E40C2: + return "TQMxE40C2"; default: return "Unknown"; } @@ -138,6 +153,11 @@ static int tqmx86_board_id_to_clk_rate(u8 board_id) case TQMX86_REG_BOARD_ID_60EB: case TQMX86_REG_BOARD_ID_70EB: case TQMX86_REG_BOARD_ID_80UC: + case TQMX86_REG_BOARD_ID_110EB: + case TQMX86_REG_BOARD_ID_E40M: + case TQMX86_REG_BOARD_ID_E40S: + case TQMX86_REG_BOARD_ID_E40C1: + case TQMX86_REG_BOARD_ID_E40C2: return 24000; case TQMX86_REG_BOARD_ID_E39M: case TQMX86_REG_BOARD_ID_E39C: From cbe060011b561cf4a0d94aa213bc1c97c789de31 Mon Sep 17 00:00:00 2001 From: Matthias Schiffer Date: Mon, 20 Feb 2023 12:25:45 +0100 Subject: [PATCH 37/43] mfd: tqmx86: Specify IO port register range more precisely [ Upstream commit 051c69ff4f607aa114c7bbdd7c41ed881367aeee ] Registers 0x160..0x17f are unassigned. Use 0x180 as base register and update offets accordingly. Also change the size of the range to include 0x19f. While 0x19f is currently reserved for future extensions, so are several of the previous registers up to 0x19e, and it is weird to leave out just the last one. Fixes: 2f17dd34ffed ("mfd: tqmx86: IO controller with I2C, Wachdog and GPIO") Signed-off-by: Matthias Schiffer Reviewed-by: Andrew Lunn Signed-off-by: Lee Jones Link: https://lore.kernel.org/r/db4677ac318b1283c8956f637f409995a30a31c3.1676892223.git.matthias.schiffer@ew.tq-group.com Signed-off-by: Sasha Levin --- drivers/mfd/tqmx86.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/drivers/mfd/tqmx86.c b/drivers/mfd/tqmx86.c index 1db95aed5012..c8ec0f92bc10 100644 --- a/drivers/mfd/tqmx86.c +++ b/drivers/mfd/tqmx86.c @@ -16,8 +16,8 @@ #include #include -#define TQMX86_IOBASE 0x160 -#define TQMX86_IOSIZE 0x3f +#define TQMX86_IOBASE 0x180 +#define TQMX86_IOSIZE 0x20 #define TQMX86_IOBASE_I2C 0x1a0 #define TQMX86_IOSIZE_I2C 0xa #define TQMX86_IOBASE_WATCHDOG 0x18b @@ -25,7 +25,7 @@ #define TQMX86_IOBASE_GPIO 0x18d #define TQMX86_IOSIZE_GPIO 0x4 -#define TQMX86_REG_BOARD_ID 0x20 +#define TQMX86_REG_BOARD_ID 0x00 #define TQMX86_REG_BOARD_ID_E38M 1 #define TQMX86_REG_BOARD_ID_50UC 2 #define TQMX86_REG_BOARD_ID_E38C 3 @@ -40,8 +40,8 @@ #define TQMX86_REG_BOARD_ID_E40S 13 #define TQMX86_REG_BOARD_ID_E40C1 14 #define TQMX86_REG_BOARD_ID_E40C2 15 -#define TQMX86_REG_BOARD_REV 0x21 -#define TQMX86_REG_IO_EXT_INT 0x26 +#define TQMX86_REG_BOARD_REV 0x01 +#define TQMX86_REG_IO_EXT_INT 0x06 #define TQMX86_REG_IO_EXT_INT_NONE 0 #define TQMX86_REG_IO_EXT_INT_7 1 #define TQMX86_REG_IO_EXT_INT_9 2 From 3530a795bb07597477fb792f625d645e79c4a45c Mon Sep 17 00:00:00 2001 From: Matthias Schiffer Date: Mon, 20 Feb 2023 12:25:46 +0100 Subject: [PATCH 38/43] mfd: tqmx86: Correct board names for TQMxE39x [ Upstream commit f376c479668557bcc2fd9e9fbc0f53e7819a11cd ] It seems that this driver was developed based on preliminary documentation. Report the correct names for all TQMxE39x variants, as they are used by the released hardware revisions: - Fix names for TQMxE39C1/C2 board IDs - Distinguish TQMxE39M and TQMxE39S, which use the same board ID The TQMxE39M/S are distinguished using the SAUC (Sanctioned Alternate Uses Configuration) register of the GPIO controller. This also prepares for the correct handling of the differences between the GPIO controllers of our COMe and SMARC modules. Fixes: 2f17dd34ffed ("mfd: tqmx86: IO controller with I2C, Wachdog and GPIO") Signed-off-by: Matthias Schiffer Reviewed-by: Andrew Lunn Signed-off-by: Lee Jones Link: https://lore.kernel.org/r/aca9a7cb42a85181bcb456c437554d2728e708ec.1676892223.git.matthias.schiffer@ew.tq-group.com Signed-off-by: Sasha Levin --- drivers/mfd/tqmx86.c | 32 +++++++++++++++++--------------- 1 file changed, 17 insertions(+), 15 deletions(-) diff --git a/drivers/mfd/tqmx86.c b/drivers/mfd/tqmx86.c index c8ec0f92bc10..0498f1b7e2e3 100644 --- a/drivers/mfd/tqmx86.c +++ b/drivers/mfd/tqmx86.c @@ -30,9 +30,9 @@ #define TQMX86_REG_BOARD_ID_50UC 2 #define TQMX86_REG_BOARD_ID_E38C 3 #define TQMX86_REG_BOARD_ID_60EB 4 -#define TQMX86_REG_BOARD_ID_E39M 5 -#define TQMX86_REG_BOARD_ID_E39C 6 -#define TQMX86_REG_BOARD_ID_E39x 7 +#define TQMX86_REG_BOARD_ID_E39MS 5 +#define TQMX86_REG_BOARD_ID_E39C1 6 +#define TQMX86_REG_BOARD_ID_E39C2 7 #define TQMX86_REG_BOARD_ID_70EB 8 #define TQMX86_REG_BOARD_ID_80UC 9 #define TQMX86_REG_BOARD_ID_110EB 11 @@ -48,6 +48,7 @@ #define TQMX86_REG_IO_EXT_INT_12 3 #define TQMX86_REG_IO_EXT_INT_MASK 0x3 #define TQMX86_REG_IO_EXT_INT_GPIO_SHIFT 4 +#define TQMX86_REG_SAUC 0x17 #define TQMX86_REG_I2C_DETECT 0x1a7 #define TQMX86_REG_I2C_DETECT_SOFT 0xa5 @@ -110,7 +111,7 @@ static const struct mfd_cell tqmx86_devs[] = { }, }; -static const char *tqmx86_board_id_to_name(u8 board_id) +static const char *tqmx86_board_id_to_name(u8 board_id, u8 sauc) { switch (board_id) { case TQMX86_REG_BOARD_ID_E38M: @@ -121,12 +122,12 @@ static const char *tqmx86_board_id_to_name(u8 board_id) return "TQMxE38C"; case TQMX86_REG_BOARD_ID_60EB: return "TQMx60EB"; - case TQMX86_REG_BOARD_ID_E39M: - return "TQMxE39M"; - case TQMX86_REG_BOARD_ID_E39C: - return "TQMxE39C"; - case TQMX86_REG_BOARD_ID_E39x: - return "TQMxE39x"; + case TQMX86_REG_BOARD_ID_E39MS: + return (sauc == 0xff) ? "TQMxE39M" : "TQMxE39S"; + case TQMX86_REG_BOARD_ID_E39C1: + return "TQMxE39C1"; + case TQMX86_REG_BOARD_ID_E39C2: + return "TQMxE39C2"; case TQMX86_REG_BOARD_ID_70EB: return "TQMx70EB"; case TQMX86_REG_BOARD_ID_80UC: @@ -159,9 +160,9 @@ static int tqmx86_board_id_to_clk_rate(u8 board_id) case TQMX86_REG_BOARD_ID_E40C1: case TQMX86_REG_BOARD_ID_E40C2: return 24000; - case TQMX86_REG_BOARD_ID_E39M: - case TQMX86_REG_BOARD_ID_E39C: - case TQMX86_REG_BOARD_ID_E39x: + case TQMX86_REG_BOARD_ID_E39MS: + case TQMX86_REG_BOARD_ID_E39C1: + case TQMX86_REG_BOARD_ID_E39C2: return 25000; case TQMX86_REG_BOARD_ID_E38M: case TQMX86_REG_BOARD_ID_E38C: @@ -173,7 +174,7 @@ static int tqmx86_board_id_to_clk_rate(u8 board_id) static int tqmx86_probe(struct platform_device *pdev) { - u8 board_id, rev, i2c_det, io_ext_int_val; + u8 board_id, sauc, rev, i2c_det, io_ext_int_val; struct device *dev = &pdev->dev; u8 gpio_irq_cfg, readback; const char *board_name; @@ -203,7 +204,8 @@ static int tqmx86_probe(struct platform_device *pdev) return -ENOMEM; board_id = ioread8(io_base + TQMX86_REG_BOARD_ID); - board_name = tqmx86_board_id_to_name(board_id); + sauc = ioread8(io_base + TQMX86_REG_SAUC); + board_name = tqmx86_board_id_to_name(board_id, sauc); rev = ioread8(io_base + TQMX86_REG_BOARD_REV); dev_info(dev, From 9aabb5f9ef3214fa1e0bc267dbbc00c534ba2b9b Mon Sep 17 00:00:00 2001 From: Marc Dionne Date: Fri, 2 Dec 2022 10:07:01 -0400 Subject: [PATCH 39/43] afs: Fix updating of i_size with dv jump from server [ Upstream commit d7f74e9a917503ee78f2b603a456d7227cf38919 ] If the data version returned from the server is larger than expected, the local data is invalidated, but we may still want to note the remote file size. Since we're setting change_size, we have to also set data_changed for the i_size to get updated. Fixes: 3f4aa9818163 ("afs: Fix EOF corruption") Signed-off-by: Marc Dionne Signed-off-by: David Howells cc: linux-afs@lists.infradead.org Signed-off-by: Sasha Levin --- fs/afs/inode.c | 1 + 1 file changed, 1 insertion(+) diff --git a/fs/afs/inode.c b/fs/afs/inode.c index 826fae22a8cc..fdca4262f806 100644 --- a/fs/afs/inode.c +++ b/fs/afs/inode.c @@ -218,6 +218,7 @@ static void afs_apply_status(struct afs_operation *op, set_bit(AFS_VNODE_ZAP_DATA, &vnode->flags); } change_size = true; + data_changed = true; } else if (vnode->status.type == AFS_FTYPE_DIR) { /* Expected directory change is handled elsewhere so * that we can locally edit the directory and save on a From f7fd1eed3125e36c8087814af89adef00ee1f7be Mon Sep 17 00:00:00 2001 From: Peng Liu Date: Tue, 21 Mar 2023 14:19:29 +0800 Subject: [PATCH 40/43] scripts/gdb: fix lx-timerlist for Python3 commit 7362042f3556528e9e9b1eb5ce8d7a3a6331476b upstream. Below incompatibilities between Python2 and Python3 made lx-timerlist fail to run under Python3. o xrange() is replaced by range() in Python3 o bytes and str are different types in Python3 o the return value of Inferior.read_memory() is memoryview object in Python3 akpm: cc stable so that older kernels are properly debuggable under newer Python. Link: https://lkml.kernel.org/r/TYCP286MB2146EE1180A4D5176CBA8AB2C6819@TYCP286MB2146.JPNP286.PROD.OUTLOOK.COM Signed-off-by: Peng Liu Reviewed-by: Jan Kiszka Cc: Florian Fainelli Cc: Kieran Bingham Cc: Signed-off-by: Andrew Morton Signed-off-by: Greg Kroah-Hartman --- scripts/gdb/linux/timerlist.py | 4 +++- scripts/gdb/linux/utils.py | 5 ++++- 2 files changed, 7 insertions(+), 2 deletions(-) diff --git a/scripts/gdb/linux/timerlist.py b/scripts/gdb/linux/timerlist.py index 071d0dd5a634..51def847f1ef 100644 --- a/scripts/gdb/linux/timerlist.py +++ b/scripts/gdb/linux/timerlist.py @@ -73,7 +73,7 @@ def print_cpu(hrtimer_bases, cpu, max_clock_bases): ts = cpus.per_cpu(tick_sched_ptr, cpu) text = "cpu: {}\n".format(cpu) - for i in xrange(max_clock_bases): + for i in range(max_clock_bases): text += " clock {}:\n".format(i) text += print_base(cpu_base['clock_base'][i]) @@ -158,6 +158,8 @@ def pr_cpumask(mask): num_bytes = (nr_cpu_ids + 7) / 8 buf = utils.read_memoryview(inf, bits, num_bytes).tobytes() buf = binascii.b2a_hex(buf) + if type(buf) is not str: + buf=buf.decode() chunks = [] i = num_bytes diff --git a/scripts/gdb/linux/utils.py b/scripts/gdb/linux/utils.py index ff7c1799d588..db59f986c7fd 100644 --- a/scripts/gdb/linux/utils.py +++ b/scripts/gdb/linux/utils.py @@ -89,7 +89,10 @@ def get_target_endianness(): def read_memoryview(inf, start, length): - return memoryview(inf.read_memory(start, length)) + m = inf.read_memory(start, length) + if type(m) is memoryview: + return m + return memoryview(m) def read_u16(buffer, offset): From c96807a4adbe2906167e8c7020a15a0f38ef060b Mon Sep 17 00:00:00 2001 From: Qu Wenruo Date: Thu, 6 Apr 2023 13:00:34 +0800 Subject: [PATCH 41/43] btrfs: scrub: reject unsupported scrub flags commit 604e6681e114d05a2e384c4d1e8ef81918037ef5 upstream. Since the introduction of scrub interface, the only flag that we support is BTRFS_SCRUB_READONLY. Thus there is no sanity checks, if there are some undefined flags passed in, we just ignore them. This is problematic if we want to introduce new scrub flags, as we have no way to determine if such flags are supported. Address the problem by introducing a check for the flags, and if unsupported flags are set, return -EOPNOTSUPP to inform the user space. This check should be backported for all supported kernels before any new scrub flags are introduced. CC: stable@vger.kernel.org # 4.14+ Reviewed-by: Anand Jain Signed-off-by: Qu Wenruo Reviewed-by: David Sterba Signed-off-by: David Sterba Signed-off-by: Greg Kroah-Hartman --- fs/btrfs/ioctl.c | 5 +++++ include/uapi/linux/btrfs.h | 1 + 2 files changed, 6 insertions(+) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 10686b494f0a..63bf68e0b006 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -3702,6 +3702,11 @@ static long btrfs_ioctl_scrub(struct file *file, void __user *arg) if (IS_ERR(sa)) return PTR_ERR(sa); + if (sa->flags & ~BTRFS_SCRUB_SUPPORTED_FLAGS) { + ret = -EOPNOTSUPP; + goto out; + } + if (!(sa->flags & BTRFS_SCRUB_READONLY)) { ret = mnt_want_write_file(file); if (ret) diff --git a/include/uapi/linux/btrfs.h b/include/uapi/linux/btrfs.h index 2c39d15a2beb..e65300d63d7c 100644 --- a/include/uapi/linux/btrfs.h +++ b/include/uapi/linux/btrfs.h @@ -181,6 +181,7 @@ struct btrfs_scrub_progress { }; #define BTRFS_SCRUB_READONLY 1 +#define BTRFS_SCRUB_SUPPORTED_FLAGS (BTRFS_SCRUB_READONLY) struct btrfs_ioctl_scrub_args { __u64 devid; /* in */ __u64 start; /* in */ From abb427cb7730a0515ced6b10194fcf843d8787d1 Mon Sep 17 00:00:00 2001 From: Stefan Haberland Date: Wed, 5 Apr 2023 16:20:17 +0200 Subject: [PATCH 42/43] s390/dasd: fix hanging blockdevice after request requeue commit d8898ee50edecacdf0141f26fd90acf43d7e9cd7 upstream. The DASD driver does not kick the requeue list when requeuing IO requests to the blocklayer. This might lead to hanging blockdevice when there is no other trigger for this. Fix by automatically kick the requeue list when requeuing DASD requests to the blocklayer. Fixes: e443343e509a ("s390/dasd: blk-mq conversion") CC: stable@vger.kernel.org # 4.14+ Signed-off-by: Stefan Haberland Reviewed-by: Jan Hoeppner Reviewed-by: Halil Pasic Link: https://lore.kernel.org/r/20230405142017.2446986-8-sth@linux.ibm.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman --- drivers/s390/block/dasd.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/s390/block/dasd.c b/drivers/s390/block/dasd.c index 9f26f55e4988..792f8f568808 100644 --- a/drivers/s390/block/dasd.c +++ b/drivers/s390/block/dasd.c @@ -3000,7 +3000,7 @@ static int _dasd_requeue_request(struct dasd_ccw_req *cqr) return 0; spin_lock_irq(&cqr->dq->lock); req = (struct request *) cqr->callback_data; - blk_mq_requeue_request(req, false); + blk_mq_requeue_request(req, true); spin_unlock_irq(&cqr->dq->lock); return 0; From eded3ad80a048deb6668bc44494e830122dcfc3a Mon Sep 17 00:00:00 2001 From: Hugh Dickins Date: Sun, 16 Apr 2023 22:17:05 -0700 Subject: [PATCH 43/43] ia64: fix an addr to taddr in huge_pte_offset() commit 3647ebcfbfca384840231fe13fae665453238a61 upstream. I know nothing of ia64 htlbpage_to_page(), but guess that the p4d line should be using taddr rather than addr, like everywhere else. Link: https://lkml.kernel.org/r/732eae88-3beb-246-2c72-281de786740@google.com Fixes: c03ab9e32a2c ("ia64: add support for folded p4d page tables") Signed-off-by: Hugh Dickins Acked-by: Mike Rapoport (IBM) Cc: Ard Biesheuvel Cc: Signed-off-by: Andrew Morton Signed-off-by: Greg Kroah-Hartman --- arch/ia64/mm/hugetlbpage.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/ia64/mm/hugetlbpage.c b/arch/ia64/mm/hugetlbpage.c index b331f94d20ac..292a28a2008f 100644 --- a/arch/ia64/mm/hugetlbpage.c +++ b/arch/ia64/mm/hugetlbpage.c @@ -57,7 +57,7 @@ huge_pte_offset (struct mm_struct *mm, unsigned long addr, unsigned long sz) pgd = pgd_offset(mm, taddr); if (pgd_present(*pgd)) { - p4d = p4d_offset(pgd, addr); + p4d = p4d_offset(pgd, taddr); if (p4d_present(*p4d)) { pud = pud_offset(p4d, taddr); if (pud_present(*pud)) {