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

[PATCH -stable] pinctrl: sh-pfc: r8a77970: remove SH_PFC_PIN_CFG_DRIVE_STRENGTH flag

by Niklas Söderlund

The datasheet does not document any registers to control drive strength, and no drive strength registers are for this reason described for this SoC. The flags indicating that drive strength can be controlled are however set for some pins in the driver. This leads to a NULL pointer dereference when the sh-pfc core tries to access the struct describing the drive strength registers, for example when reading the sysfs file pinconf-pins. Fix this by removing the SH_PFC_PIN_CFG_DRIVE_STRENGTH from all pins Fixes: b92ac66a1819602b ("pinctrl: sh-pfc: Add R8A77970 PFC support") Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas(a)ragnatech.se> --- drivers/pinctrl/sh-pfc/pfc-r8a77970.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) --- Hi, This is a backport of commit 550b6f7e8cf93fc2753aa01e655ed5471012ab5a from Linusw linux-pinctrl.git tree. It's applicable for v4.16 and v4.17. This is my first backport patch submission for stable, please let me know if I can improve the format of this patch commit message or subject in any way. diff --git a/drivers/pinctrl/sh-pfc/pfc-r8a77970.c b/drivers/pinctrl/sh-pfc/pfc-r8a77970.c index b1bb7263532b3bf9..049b374aa4ae4698 100644 --- a/drivers/pinctrl/sh-pfc/pfc-r8a77970.c +++ b/drivers/pinctrl/sh-pfc/pfc-r8a77970.c @@ -22,12 +22,12 @@ #include "sh_pfc.h" #define CPU_ALL_PORT(fn, sfx) \ - PORT_GP_CFG_22(0, fn, sfx, SH_PFC_PIN_CFG_DRIVE_STRENGTH), \ - PORT_GP_CFG_28(1, fn, sfx, SH_PFC_PIN_CFG_DRIVE_STRENGTH), \ - PORT_GP_CFG_17(2, fn, sfx, SH_PFC_PIN_CFG_DRIVE_STRENGTH), \ - PORT_GP_CFG_17(3, fn, sfx, SH_PFC_PIN_CFG_DRIVE_STRENGTH), \ - PORT_GP_CFG_6(4, fn, sfx, SH_PFC_PIN_CFG_DRIVE_STRENGTH), \ - PORT_GP_CFG_15(5, fn, sfx, SH_PFC_PIN_CFG_DRIVE_STRENGTH) + PORT_GP_22(0, fn, sfx), \ + PORT_GP_28(1, fn, sfx), \ + PORT_GP_17(2, fn, sfx), \ + PORT_GP_17(3, fn, sfx), \ + PORT_GP_6(4, fn, sfx), \ + PORT_GP_15(5, fn, sfx) /* * F_() : just information * FM() : macro for FN_xxx / xxx_MARK -- 2.18.0

7 years, 4 months

4
3
0 0

request for 4.14-stable

by Sudip Mukherjee

Hi Greg, These commits are missing from 4.14-stable tree. I have seen your failed mail for one of them. (resending as I missed adding stable in the previous mail) -- Regards Sudip

7 years, 4 months

2
1
0 0

[PATCH] arm64: fix kernel panic on serror caused by user process

by Hari Vyas

changes in v1: Based on processor mode, action is taken i.e. for user mode, only user process is killed otherwise as usual kernel will panic and system halts. Hari Vyas (1): arm64: fix kernel panic on serror exception caused by user process arch/arm64/kernel/traps.c | 7 +++++++ 1 file changed, 7 insertions(+) -- 1.9.1

7 years, 4 months

5
9
0 0

[PATCH v5] PM / devfreq: Fix devfreq_add_device() when drivers are built as modules.

by Enric Balletbo i Serra

When the devfreq driver and the governor driver are built as modules, the call to devfreq_add_device() or governor_store() fails because the governor driver is not loaded at the time the devfreq driver loads. The devfreq driver has a build dependency on the governor but also should have a runtime dependency. We need to make sure that the governor driver is loaded before the devfreq driver. This patch fixes this bug by adding a try_then_request_governor() function. First tries to find the governor, and then, if it is not found, it requests the module and tries again. Fixes: 1b5c1be2c88e (PM / devfreq: map devfreq drivers to governor using name) Signed-off-by: Enric Balletbo i Serra <enric.balletbo(a)collabora.com> --- Changes in v5: - Requested by MyungJoo and Chanwoo. - Fix returning without the lock acquired after request_module. - Requested by Chanwoo. - In request governor function check if governor name is NULL or not. - Remove some unrelated changes (added/removed some blank lines). Changes in v4: - Kept "locked" devfreq_list from the return of find_devfreq_governor() to the unlock of governor_store(). Requested by MyungJoo Ham. Changes in v3: - Remove unneded change in dev_err message. - Fix err returned value in case to not find the governor. Changes in v2: - Add a new function to request the module and call that function from devfreq_add_device and governor_store. drivers/devfreq/devfreq.c | 53 ++++++++++++++++++++++++++++++++++++--- 1 file changed, 49 insertions(+), 4 deletions(-) diff --git a/drivers/devfreq/devfreq.c b/drivers/devfreq/devfreq.c index 0b5b3abe054e..aa92fbf9f0dd 100644 --- a/drivers/devfreq/devfreq.c +++ b/drivers/devfreq/devfreq.c @@ -11,6 +11,7 @@ */ #include <linux/kernel.h> +#include <linux/kmod.h> #include <linux/sched.h> #include <linux/errno.h> #include <linux/err.h> @@ -221,6 +222,49 @@ static struct devfreq_governor *find_devfreq_governor(const char *name) return ERR_PTR(-ENODEV); } +/** + * try_then_request_governor() - Try to find the governor and request the + * module if is not found. + * @name: name of the governor + * + * Search the list of devfreq governors and request the module and try again + * if is not found. This can happen when both drivers (the governor driver + * and the driver that call devfreq_add_device) are built as modules. + * devfreq_list_lock should be held by the caller. Returns the matched + * governor's pointer. + */ +static struct devfreq_governor *try_then_request_governor(const char *name) +{ + struct devfreq_governor *governor; + int err = 0; + + if (IS_ERR_OR_NULL(name)) { + pr_err("DEVFREQ: %s: Invalid parameters\n", __func__); + return ERR_PTR(-EINVAL); + } + WARN(!mutex_is_locked(&devfreq_list_lock), + "devfreq_list_lock must be locked."); + + governor = find_devfreq_governor(name); + if (IS_ERR(governor)) { + mutex_unlock(&devfreq_list_lock); + + if (!strncmp(name, DEVFREQ_GOV_SIMPLE_ONDEMAND, + DEVFREQ_NAME_LEN)) + err = request_module("governor_%s", "simpleondemand"); + else + err = request_module("governor_%s", name); + /* Restore previous state before return */ + mutex_lock(&devfreq_list_lock); + if (err) + return NULL; + + governor = find_devfreq_governor(name); + } + + return governor; +} + static int devfreq_notify_transition(struct devfreq *devfreq, struct devfreq_freqs *freqs, unsigned int state) { @@ -645,9 +689,8 @@ struct devfreq *devfreq_add_device(struct device *dev, mutex_unlock(&devfreq->lock); mutex_lock(&devfreq_list_lock); - list_add(&devfreq->node, &devfreq_list); - governor = find_devfreq_governor(devfreq->governor_name); + governor = try_then_request_governor(devfreq->governor_name); if (IS_ERR(governor)) { dev_err(dev, "%s: Unable to find governor for the device\n", __func__); @@ -663,12 +706,14 @@ struct devfreq *devfreq_add_device(struct device *dev, __func__); goto err_init; } + + list_add(&devfreq->node, &devfreq_list); + mutex_unlock(&devfreq_list_lock); return devfreq; err_init: - list_del(&devfreq->node); mutex_unlock(&devfreq_list_lock); device_unregister(&devfreq->dev); @@ -989,7 +1034,7 @@ static ssize_t governor_store(struct device *dev, struct device_attribute *attr, return -EINVAL; mutex_lock(&devfreq_list_lock); - governor = find_devfreq_governor(str_governor); + governor = try_then_request_governor(str_governor); if (IS_ERR(governor)) { ret = PTR_ERR(governor); goto out; -- 2.18.0

7 years, 4 months

2
2
0 0

Linux 4.17.8

by Greg KH

I'm announcing the release of the 4.17.8 kernel. This is to fix the i386 issue that was in the 4.17.7 release. All should be fine now. The updated 4.17.y git tree can be found at: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git linux-4.17.y and can be browsed at the normal kernel.org git web browser: http://git.kernel.org/?p=linux/kernel/git/stable/linux-stable.git;a=summary thanks, greg k-h ------------ Makefile | 2 +- include/linux/mm.h | 2 +- mm/page_alloc.c | 4 ++-- 3 files changed, 4 insertions(+), 4 deletions(-) Greg Kroah-Hartman (1): Linux 4.17.8 Pavel Tatashin (1): mm: don't do zero_resv_unavail if memmap is not allocated

7 years, 4 months

1
1
0 0

[PATCH -v2] random: mix rdrand with entropy sent in from userspace

by Theodore Ts'o

Fedora has integrated the jitter entropy daemon to work around slow boot problems, especially on VM's that don't support virtio-rng: https://bugzilla.redhat.com/show_bug.cgi?id=1572944 It's understandable why they did this, but the Jitter entropy daemon works fundamentally on the principle: "the CPU microarchitecture is **so** complicated and we can't figure it out, so it *must* be random". Yes, it uses statistical tests to "prove" it is secure, but AES_ENCRYPT(NSA_KEY, COUNTER++) will also pass statistical tests with flying colors. So if RDRAND is available, mix it into entropy submitted from userspace. It can't hurt, and if you believe the NSA has backdoored RDRAND, then they probably have enough details about the Intel microarchitecture that they can reverse engineer how the Jitter entropy daemon affects the microarchitecture, and attack its output stream. And if RDRAND is in fact an honest DRNG, it will immeasurably improve on what the Jitter entropy daemon might produce. This also provides some protection against someone who is able to read or set the entropy seed file. Signed-off-by: Theodore Ts'o <tytso(a)mit.edu> Cc: stable(a)vger.kernel.org Cc: Arnd Bergmann <arnd(a)arndb.de> --- Changes in v2: - Fix silly typo that Arnd pointed out in check the return value of arch_get_random_int() - Break out of the loop after the first failure reported by arch_get_random_int() drivers/char/random.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) diff --git a/drivers/char/random.c b/drivers/char/random.c index cd888d4ee605..bd449ad52442 100644 --- a/drivers/char/random.c +++ b/drivers/char/random.c @@ -1895,14 +1895,22 @@ static int write_pool(struct entropy_store *r, const char __user *buffer, size_t count) { size_t bytes; - __u32 buf[16]; + __u32 t, buf[16]; const char __user *p = buffer; while (count > 0) { + int b, i = 0; + bytes = min(count, sizeof(buf)); if (copy_from_user(&buf, p, bytes)) return -EFAULT; + for (b = bytes ; b > 0 ; b -= sizeof(__u32), i++) { + if (!arch_get_random_int(&t)) + break; + buf[i] ^= t; + } + count -= bytes; p += bytes; -- 2.18.0.rc0

7 years, 4 months

1
0
0 0

[PATCH] drm/amdgpu: add another ATPX quirk for TOPAZ

by Alex Deucher

Needs ATPX rather than _PR3. Bug: https://bugzilla.kernel.org/show_bug.cgi?id=200517 Signed-off-by: Alex Deucher <alexander.deucher(a)amd.com> Cc: stable(a)vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c index b33f1680c9a3..a028661d9e20 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c @@ -575,6 +575,7 @@ static const struct amdgpu_px_quirk amdgpu_px_quirk_list[] = { { 0x1002, 0x6900, 0x1002, 0x0124, AMDGPU_PX_QUIRK_FORCE_ATPX }, { 0x1002, 0x6900, 0x1028, 0x0812, AMDGPU_PX_QUIRK_FORCE_ATPX }, { 0x1002, 0x6900, 0x1028, 0x0813, AMDGPU_PX_QUIRK_FORCE_ATPX }, + { 0x1002, 0x6900, 0x1025, 0x125A, AMDGPU_PX_QUIRK_FORCE_ATPX }, { 0, 0, 0, 0, 0 }, }; -- 2.13.6

7 years, 4 months

2
1
0 0

v4.14.56 build: 0 failures 0 warnings (v4.14.56)

by Build bot for Mark Brown

Tree/Branch: v4.14.56 Git describe: v4.14.56 Commit: cff26c95b2 Linux 4.14.56 Build Time: 98 min 59 sec Passed: 11 / 11 (100.00 %) Failed: 0 / 11 ( 0.00 %) Errors: 0 Warnings: 0 Section Mismatches: 0 ------------------------------------------------------------------------------- defconfigs with issues (other than build errors): ------------------------------------------------------------------------------- =============================================================================== Detailed per-defconfig build reports below: ------------------------------------------------------------------------------- Passed with no errors, warnings or mismatches: arm64-allnoconfig arm64-allmodconfig arm-multi_v5_defconfig arm-multi_v7_defconfig x86_64-defconfig arm-allmodconfig arm-allnoconfig x86_64-allnoconfig arm-multi_v4t_defconfig x86_64-allmodconfig arm64-defconfig close failed in file object destructor: sys.excepthook is missing lost sys.stderr

7 years, 4 months

1
0
0 0

[PATCH 1/3] drm/i915: Nuke the LVDS lid notifier

by Ville Syrjala

From: Ville Syrjälä <ville.syrjala(a)linux.intel.com> We broke the LVDS notifier resume thing in (presumably) commit e2c8b8701e2d ("drm/i915: Use atomic helpers for suspend, v2.") as we no longer duplicate the current state in the LVDS notifier and thus we never resume it properly either. Instead of trying to fix it again let's just kill off the lid notifier entirely. None of the machines tested thus far have apparently needed it. Originally the lid notifier was added to work around cases where the VBIOS was clobbering some of the hardware state behind the driver's back, mostly on Thinkpads. We now have a few report of Thinkpads working just fine without the notifier. So maybe it was misdiagnosed originally, or something else has changed (ACPI video stuff perhaps?). If we do end up finding a machine where the VBIOS is still causing problems I would suggest that we first try setting various bits in the VBIOS scratch registers. There are several to choose from that may instruct the VBIOS to steer clear. With the notifier gone we'll also stop looking at the panel status in ->detect(). Cc: stable(a)vger.kernel.org Cc: Wolfgang Draxinger <wdraxinger.maillist(a)draxit.de> Cc: Vito Caputo <vcaputo(a)pengaru.com> Cc: kitsunyan <kitsunyan(a)airmail.cc> Cc: Joonas Saarinen <jza(a)saunalahti.fi> Tested-by: Vito Caputo <vcaputo(a)pengaru.com> # Thinkapd X61s Tested-by: kitsunyan <kitsunyan(a)airmail.cc> # ThinkPad X200 Tested-by: Joonas Saarinen <jza(a)saunalahti.fi> # Fujitsu Siemens U9210 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=105902 References: https://lists.freedesktop.org/archives/intel-gfx/2018-June/169315.html References: https://bugs.freedesktop.org/show_bug.cgi?id=21230 Fixes: e2c8b8701e2d ("drm/i915: Use atomic helpers for suspend, v2.") Signed-off-by: Ville Syrjälä <ville.syrjala(a)linux.intel.com> --- drivers/gpu/drm/i915/i915_drv.c | 10 --- drivers/gpu/drm/i915/i915_drv.h | 2 - drivers/gpu/drm/i915/intel_lvds.c | 136 +------------------------------------- 3 files changed, 2 insertions(+), 146 deletions(-) diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c index 3834bd758a2e..f8cfd16be534 100644 --- a/drivers/gpu/drm/i915/i915_drv.c +++ b/drivers/gpu/drm/i915/i915_drv.c @@ -900,7 +900,6 @@ static int i915_driver_init_early(struct drm_i915_private *dev_priv, spin_lock_init(&dev_priv->uncore.lock); mutex_init(&dev_priv->sb_lock); - mutex_init(&dev_priv->modeset_restore_lock); mutex_init(&dev_priv->av_mutex); mutex_init(&dev_priv->wm.wm_mutex); mutex_init(&dev_priv->pps_mutex); @@ -1570,11 +1569,6 @@ static int i915_drm_suspend(struct drm_device *dev) struct pci_dev *pdev = dev_priv->drm.pdev; pci_power_t opregion_target_state; - /* ignore lid events during suspend */ - mutex_lock(&dev_priv->modeset_restore_lock); - dev_priv->modeset_restore = MODESET_SUSPENDED; - mutex_unlock(&dev_priv->modeset_restore_lock); - disable_rpm_wakeref_asserts(dev_priv); /* We do a lot of poking in a lot of registers, make sure they work @@ -1770,10 +1764,6 @@ static int i915_drm_resume(struct drm_device *dev) intel_fbdev_set_suspend(dev, FBINFO_STATE_RUNNING, false); - mutex_lock(&dev_priv->modeset_restore_lock); - dev_priv->modeset_restore = MODESET_DONE; - mutex_unlock(&dev_priv->modeset_restore_lock); - intel_opregion_notify_adapter(dev_priv, PCI_D0); enable_rpm_wakeref_asserts(dev_priv); diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h index 4fb937399440..1b0af905b74c 100644 --- a/drivers/gpu/drm/i915/i915_drv.h +++ b/drivers/gpu/drm/i915/i915_drv.h @@ -1730,8 +1730,6 @@ struct drm_i915_private { unsigned long quirks; - enum modeset_restore modeset_restore; - struct mutex modeset_restore_lock; struct drm_atomic_state *modeset_restore_state; struct drm_modeset_acquire_ctx reset_ctx; diff --git a/drivers/gpu/drm/i915/intel_lvds.c b/drivers/gpu/drm/i915/intel_lvds.c index ca55b0a82ba6..f9f3b0885ba5 100644 --- a/drivers/gpu/drm/i915/intel_lvds.c +++ b/drivers/gpu/drm/i915/intel_lvds.c @@ -44,8 +44,6 @@ /* Private structure for the integrated LVDS support */ struct intel_lvds_connector { struct intel_connector base; - - struct notifier_block lid_notifier; }; struct intel_lvds_pps { @@ -452,26 +450,9 @@ static bool intel_lvds_compute_config(struct intel_encoder *intel_encoder, return true; } -/* - * Detect the LVDS connection. - * - * Since LVDS doesn't have hotlug, we use the lid as a proxy. Open means - * connected and closed means disconnected. We also send hotplug events as - * needed, using lid status notification from the input layer. - */ static enum drm_connector_status intel_lvds_detect(struct drm_connector *connector, bool force) { - struct drm_i915_private *dev_priv = to_i915(connector->dev); - enum drm_connector_status status; - - DRM_DEBUG_KMS("[CONNECTOR:%d:%s]\n", - connector->base.id, connector->name); - - status = intel_panel_detect(dev_priv); - if (status != connector_status_unknown) - return status; - return connector_status_connected; } @@ -496,117 +477,6 @@ static int intel_lvds_get_modes(struct drm_connector *connector) return 1; } -static int intel_no_modeset_on_lid_dmi_callback(const struct dmi_system_id *id) -{ - DRM_INFO("Skipping forced modeset for %s\n", id->ident); - return 1; -} - -/* The GPU hangs up on these systems if modeset is performed on LID open */ -static const struct dmi_system_id intel_no_modeset_on_lid[] = { - { - .callback = intel_no_modeset_on_lid_dmi_callback, - .ident = "Toshiba Tecra A11", - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "TOSHIBA"), - DMI_MATCH(DMI_PRODUCT_NAME, "TECRA A11"), - }, - }, - - { } /* terminating entry */ -}; - -/* - * Lid events. Note the use of 'modeset': - * - we set it to MODESET_ON_LID_OPEN on lid close, - * and set it to MODESET_DONE on open - * - we use it as a "only once" bit (ie we ignore - * duplicate events where it was already properly set) - * - the suspend/resume paths will set it to - * MODESET_SUSPENDED and ignore the lid open event, - * because they restore the mode ("lid open"). - */ -static int intel_lid_notify(struct notifier_block *nb, unsigned long val, - void *unused) -{ - struct intel_lvds_connector *lvds_connector = - container_of(nb, struct intel_lvds_connector, lid_notifier); - struct drm_connector *connector = &lvds_connector->base.base; - struct drm_device *dev = connector->dev; - struct drm_i915_private *dev_priv = to_i915(dev); - - if (dev->switch_power_state != DRM_SWITCH_POWER_ON) - return NOTIFY_OK; - - mutex_lock(&dev_priv->modeset_restore_lock); - if (dev_priv->modeset_restore == MODESET_SUSPENDED) - goto exit; - /* - * check and update the status of LVDS connector after receiving - * the LID nofication event. - */ - connector->status = connector->funcs->detect(connector, false); - - /* Don't force modeset on machines where it causes a GPU lockup */ - if (dmi_check_system(intel_no_modeset_on_lid)) - goto exit; - if (!acpi_lid_open()) { - /* do modeset on next lid open event */ - dev_priv->modeset_restore = MODESET_ON_LID_OPEN; - goto exit; - } - - if (dev_priv->modeset_restore == MODESET_DONE) - goto exit; - - /* - * Some old platform's BIOS love to wreak havoc while the lid is closed. - * We try to detect this here and undo any damage. The split for PCH - * platforms is rather conservative and a bit arbitrary expect that on - * those platforms VGA disabling requires actual legacy VGA I/O access, - * and as part of the cleanup in the hw state restore we also redisable - * the vga plane. - */ - if (!HAS_PCH_SPLIT(dev_priv)) - intel_display_resume(dev); - - dev_priv->modeset_restore = MODESET_DONE; - -exit: - mutex_unlock(&dev_priv->modeset_restore_lock); - return NOTIFY_OK; -} - -static int -intel_lvds_connector_register(struct drm_connector *connector) -{ - struct intel_lvds_connector *lvds = to_lvds_connector(connector); - int ret; - - ret = intel_connector_register(connector); - if (ret) - return ret; - - lvds->lid_notifier.notifier_call = intel_lid_notify; - if (acpi_lid_notifier_register(&lvds->lid_notifier)) { - DRM_DEBUG_KMS("lid notifier registration failed\n"); - lvds->lid_notifier.notifier_call = NULL; - } - - return 0; -} - -static void -intel_lvds_connector_unregister(struct drm_connector *connector) -{ - struct intel_lvds_connector *lvds = to_lvds_connector(connector); - - if (lvds->lid_notifier.notifier_call) - acpi_lid_notifier_unregister(&lvds->lid_notifier); - - intel_connector_unregister(connector); -} - /** * intel_lvds_destroy - unregister and free LVDS structures * @connector: connector to free @@ -639,8 +509,8 @@ static const struct drm_connector_funcs intel_lvds_connector_funcs = { .fill_modes = drm_helper_probe_single_connector_modes, .atomic_get_property = intel_digital_connector_atomic_get_property, .atomic_set_property = intel_digital_connector_atomic_set_property, - .late_register = intel_lvds_connector_register, - .early_unregister = intel_lvds_connector_unregister, + .late_register = intel_connector_register, + .early_unregister = intel_connector_unregister, .destroy = intel_lvds_destroy, .atomic_destroy_state = drm_atomic_helper_connector_destroy_state, .atomic_duplicate_state = intel_digital_connector_duplicate_state, @@ -1114,8 +984,6 @@ void intel_lvds_init(struct drm_i915_private *dev_priv) * 2) check for VBT data * 3) check to see if LVDS is already on * if none of the above, no panel - * 4) make sure lid is open - * if closed, act like it's not there for now */ /* -- 2.16.4

7 years, 4 months

2
1
0 0

[PATCH v4] stop_machine: Disable preemption after queueing stopper threads

by Isaac J. Manjarres

After cpu_stop_queue_two_works() queues the cpu_stop works for the stopper threads, it releases the locks held for both threads, which enables preemption, which allows the following race condition to occur: On one CPU, call it CPU 3, thread 1 invokes cpu_stop_queue_two_works(2, 3,...), and the execution is such that thread 1 queues the works for migration/2 and migration/3, and is preempted after releasing the locks for migration/2 and migration/3, but before waking the threads. Then, On CPU 2, a kworker, call it thread 2, is running, and it invokes cpu_stop_queue_two_works(1, 2,...), such that thread 2 queues the works for migration/1 and migration/2. Meanwhile, on CPU 3, thread 1 resumes execution, and wakes migration/2 and migration/3. This means that when CPU 2 releases the locks for migration/1 and migration/2, but before it wakes those threads, it can be preempted by migration/2. If thread 2 is preempted by migration/2, then migration/2 will execute the first work item successfully, since migration/3 was woken up by CPU 3, but when it goes to execute the second work item, it disables preemption, calls multi_cpu_stop(), and thus, CPU 2 will wait forever for migration/1, which should have been woken up by thread 2. However migration/1 cannot be woken up by thread 2, since it is a kworker, so it is affine to CPU 2, but CPU 2 is running migration/2 with preemption disabled, so thread 2 will never run. Disable preemption after queueing works for stopper threads to ensure that the operation of queueing the works and waking the stopper threads is atomic. Fixes: 0b26351b910f ("stop_machine, sched: Fix migrate_swap() vs. active_balance() deadlock") Co-Developed-by: Prasad Sodagudi <psodagud(a)codeaurora.org> Co-Developed-by: Pavankumar Kondeti <pkondeti(a)codeaurora.org> Signed-off-by: Isaac J. Manjarres <isaacm(a)codeaurora.org> Signed-off-by: Prasad Sodagudi <psodagud(a)codeaurora.org> Signed-off-by: Pavankumar Kondeti <pkondeti(a)codeaurora.org> Cc: stable(a)vger.kernel.org --- kernel/stop_machine.c | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) diff --git a/kernel/stop_machine.c b/kernel/stop_machine.c index f89014a..e190d1e 100644 --- a/kernel/stop_machine.c +++ b/kernel/stop_machine.c @@ -260,6 +260,15 @@ static int cpu_stop_queue_two_works(int cpu1, struct cpu_stop_work *work1, err = 0; __cpu_stop_queue_work(stopper1, work1, &wakeq); __cpu_stop_queue_work(stopper2, work2, &wakeq); + /* + * The waking up of stopper threads has to happen + * in the same scheduling context as the queueing. + * Otherwise, there is a possibility of one of the + * above stoppers being woken up by another CPU, + * and preempting us. This will cause us to n ot + * wake up the other stopper forever. + */ + preempt_disable(); unlock: raw_spin_unlock(&stopper2->lock); raw_spin_unlock_irq(&stopper1->lock); @@ -270,7 +279,10 @@ static int cpu_stop_queue_two_works(int cpu1, struct cpu_stop_work *work1, goto retry; } - wake_up_q(&wakeq); + if (!err) { + wake_up_q(&wakeq); + preempt_enable(); + } return err; } -- The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project

7 years, 4 months

3
2
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror