April 2024 - Linux-stable-mirror

[PATCH 2/2] gpio: cdev: check for NULL labels when sanitizing them for irqs

by Kent Gibson

From: Bartosz Golaszewski <bartosz.golaszewski(a)linaro.org> We need to take into account that a line's consumer label may be NULL and not try to kstrdup() it in that case but rather pass the NULL pointer up the stack to the interrupt request function. To that end: let make_irq_label() return NULL as a valid return value and use ERR_PTR() instead to signal an allocation failure to callers. Cc: stable(a)vger.kernel.org Fixes: b34490879baa ("gpio: cdev: sanitize the label before requesting the interrupt") Reported-by: Linux Kernel Functional Testing <lkft(a)linaro.org> Closes: https://lore.kernel.org/lkml/20240402093534.212283-1-naresh.kamboju@linaro.… Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski(a)linaro.org> Signed-off-by: Kent Gibson <warthog618(a)gmail.com> (rebased) --- drivers/gpio/gpiolib-cdev.c | 23 ++++++++++++++++------- 1 file changed, 16 insertions(+), 7 deletions(-) diff --git a/drivers/gpio/gpiolib-cdev.c b/drivers/gpio/gpiolib-cdev.c index f4c2da2041e5..8112ec36e55f 100644 --- a/drivers/gpio/gpiolib-cdev.c +++ b/drivers/gpio/gpiolib-cdev.c @@ -730,7 +730,16 @@ static u32 line_event_id(int level) static inline char *make_irq_label(const char *orig) { - return kstrdup_and_replace(orig, '/', ':', GFP_KERNEL); + char *new; + + if (!orig) + return NULL; + + new = kstrdup_and_replace(orig, '/', ':', GFP_KERNEL); + if (!new) + return ERR_PTR(-ENOMEM); + + return new; } static inline void free_irq_label(const char *label) @@ -1049,8 +1058,8 @@ static int debounce_setup(struct line *line, unsigned int debounce_period_us) return -ENXIO; label = make_irq_label(line->req->label); - if (!label) - return -ENOMEM; + if (IS_ERR(label)) + return PTR_ERR(label); irqflags = IRQF_TRIGGER_FALLING | IRQF_TRIGGER_RISING; ret = request_irq(irq, debounce_irq_handler, irqflags, @@ -1165,8 +1174,8 @@ static int edge_detector_setup(struct line *line, irqflags |= IRQF_ONESHOT; label = make_irq_label(line->req->label); - if (!label) - return -ENOMEM; + if (IS_ERR(label)) + return PTR_ERR(label); /* Request a thread to read the events */ ret = request_threaded_irq(irq, edge_irq_handler, edge_irq_thread, @@ -2224,8 +2233,8 @@ static int lineevent_create(struct gpio_device *gdev, void __user *ip) goto out_free_le; label = make_irq_label(le->label); - if (!label) { - ret = -ENOMEM; + if (IS_ERR(label)) { + ret = PTR_ERR(label); goto out_free_le; } -- 2.39.2

1 year

1
0
0 0

[PATCH net v3] virtio_net: Do not send RSS key if it is not supported

by Breno Leitao

There is a bug when setting the RSS options in virtio_net that can break the whole machine, getting the kernel into an infinite loop. Running the following command in any QEMU virtual machine with virtionet will reproduce this problem: # ethtool -X eth0 hfunc toeplitz This is how the problem happens: 1) ethtool_set_rxfh() calls virtnet_set_rxfh() 2) virtnet_set_rxfh() calls virtnet_commit_rss_command() 3) virtnet_commit_rss_command() populates 4 entries for the rss scatter-gather 4) Since the command above does not have a key, then the last scatter-gatter entry will be zeroed, since rss_key_size == 0. sg_buf_size = vi->rss_key_size; 5) This buffer is passed to qemu, but qemu is not happy with a buffer with zero length, and do the following in virtqueue_map_desc() (QEMU function): if (!sz) { virtio_error(vdev, "virtio: zero sized buffers are not allowed"); 6) virtio_error() (also QEMU function) set the device as broken vdev->broken = true; 7) Qemu bails out, and do not repond this crazy kernel. 8) The kernel is waiting for the response to come back (function virtnet_send_command()) 9) The kernel is waiting doing the following : while (!virtqueue_get_buf(vi->cvq, &tmp) && !virtqueue_is_broken(vi->cvq)) cpu_relax(); 10) None of the following functions above is true, thus, the kernel loops here forever. Keeping in mind that virtqueue_is_broken() does not look at the qemu `vdev->broken`, so, it never realizes that the vitio is broken at QEMU side. Fix it by not sending RSS commands if the feature is not available in the device. Fixes: c7114b1249fa ("drivers/net/virtio_net: Added basic RSS support.") Cc: stable(a)vger.kernel.org Cc: qemu-devel(a)nongnu.org Signed-off-by: Breno Leitao <leitao(a)debian.org> --- Changelog: V2: * Moved from creating a valid packet, by rejecting the request completely V3: * Got some good feedback from and Xuan Zhuo and Heng Qi, and reworked the rejection path. --- drivers/net/virtio_net.c | 22 ++++++++++++++++++---- 1 file changed, 18 insertions(+), 4 deletions(-) diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c index c22d1118a133..c4a21ec51adf 100644 --- a/drivers/net/virtio_net.c +++ b/drivers/net/virtio_net.c @@ -3807,6 +3807,7 @@ static int virtnet_set_rxfh(struct net_device *dev, struct netlink_ext_ack *extack) { struct virtnet_info *vi = netdev_priv(dev); + bool update = false; int i; if (rxfh->hfunc != ETH_RSS_HASH_NO_CHANGE && @@ -3814,13 +3815,24 @@ static int virtnet_set_rxfh(struct net_device *dev, return -EOPNOTSUPP; if (rxfh->indir) { + if (!vi->has_rss) + return -EOPNOTSUPP; + for (i = 0; i < vi->rss_indir_table_size; ++i) vi->ctrl->rss.indirection_table[i] = rxfh->indir[i]; + update = true; } - if (rxfh->key) + + if (rxfh->key) { + if (!vi->has_rss && !vi->has_rss_hash_report) + return -EOPNOTSUPP; + memcpy(vi->ctrl->rss.key, rxfh->key, vi->rss_key_size); + update = true; + } - virtnet_commit_rss_command(vi); + if (update) + virtnet_commit_rss_command(vi); return 0; } @@ -4729,13 +4741,15 @@ static int virtnet_probe(struct virtio_device *vdev) if (virtio_has_feature(vdev, VIRTIO_NET_F_HASH_REPORT)) vi->has_rss_hash_report = true; - if (virtio_has_feature(vdev, VIRTIO_NET_F_RSS)) + if (virtio_has_feature(vdev, VIRTIO_NET_F_RSS)) { vi->has_rss = true; - if (vi->has_rss || vi->has_rss_hash_report) { vi->rss_indir_table_size = virtio_cread16(vdev, offsetof(struct virtio_net_config, rss_max_indirection_table_length)); + } + + if (vi->has_rss || vi->has_rss_hash_report) { vi->rss_key_size = virtio_cread8(vdev, offsetof(struct virtio_net_config, rss_max_key_size)); -- 2.43.0

1 year

4
5
0 0

[PATCH v3] HID: i2c-hid: Revert to await reset ACK before reading report descriptor

by Hans de Goede

From: Kenny Levinsen <kl(a)kl.wtf> In af93a167eda9, i2c_hid_parse was changed to continue with reading the report descriptor before waiting for reset to be acknowledged. This has lead to two regressions: 1. We fail to handle reset acknowledgment if it happens while reading the report descriptor. The transfer sets I2C_HID_READ_PENDING, which causes the IRQ handler to return without doing anything. This affects both a Wacom touchscreen and a Sensel touchpad. 2. On a Sensel touchpad, reading the report descriptor this quickly after reset results in all zeroes or partial zeroes. The issues were observed on the Lenovo Thinkpad Z16 Gen 2. The change in question was made based on a Microsoft article[0] stating that Windows 8 *may* read the report descriptor in parallel with awaiting reset acknowledgment, intended as a slight reset performance optimization. Perhaps they only do this if reset is not completing quickly enough for their tastes? As the code is not currently ready to read registers in parallel with a pending reset acknowledgment, and as reading quickly breaks the report descriptor on the Sensel touchpad, revert to waiting for reset acknowledgment before proceeding to read the report descriptor. [0]: https://learn.microsoft.com/en-us/windows-hardware/drivers/hid/plug-and-pla… Fixes: af93a167eda9 ("HID: i2c-hid: Move i2c_hid_finish_hwreset() to after reading the report-descriptor") Closes: https://bugzilla.redhat.com/show_bug.cgi?id=2271136 Cc: stable(a)vger.kernel.org Signed-off-by: Kenny Levinsen <kl(a)kl.wtf> Link: https://lore.kernel.org/r/20240331182440.14477-1-kl@kl.wtf [hdegoede(a)redhat.com Drop no longer necessary abort_reset error exit path] Signed-off-by: Hans de Goede <hdegoede(a)redhat.com> --- drivers/hid/i2c-hid/i2c-hid-core.c | 29 ++++++++--------------------- 1 file changed, 8 insertions(+), 21 deletions(-) diff --git a/drivers/hid/i2c-hid/i2c-hid-core.c b/drivers/hid/i2c-hid/i2c-hid-core.c index 2df1ab3c31cc..13d67d7c67b4 100644 --- a/drivers/hid/i2c-hid/i2c-hid-core.c +++ b/drivers/hid/i2c-hid/i2c-hid-core.c @@ -735,12 +735,15 @@ static int i2c_hid_parse(struct hid_device *hid) mutex_lock(&ihid->reset_lock); do { ret = i2c_hid_start_hwreset(ihid); - if (ret) + if (ret == 0) + ret = i2c_hid_finish_hwreset(ihid); + else msleep(1000); } while (tries-- > 0 && ret); + mutex_unlock(&ihid->reset_lock); if (ret) - goto abort_reset; + return ret; use_override = i2c_hid_get_dmi_hid_report_desc_override(client->name, &rsize); @@ -750,11 +753,8 @@ static int i2c_hid_parse(struct hid_device *hid) i2c_hid_dbg(ihid, "Using a HID report descriptor override\n"); } else { rdesc = kzalloc(rsize, GFP_KERNEL); - - if (!rdesc) { - ret = -ENOMEM; - goto abort_reset; - } + if (!rdesc) + return -ENOMEM; i2c_hid_dbg(ihid, "asking HID report descriptor\n"); @@ -763,23 +763,10 @@ static int i2c_hid_parse(struct hid_device *hid) rdesc, rsize); if (ret) { hid_err(hid, "reading report descriptor failed\n"); - goto abort_reset; + goto out; } } - /* - * Windows directly reads the report-descriptor after sending reset - * and then waits for resets completion afterwards. Some touchpads - * actually wait for the report-descriptor to be read before signalling - * reset completion. - */ - ret = i2c_hid_finish_hwreset(ihid); -abort_reset: - clear_bit(I2C_HID_RESET_PENDING, &ihid->flags); - mutex_unlock(&ihid->reset_lock); - if (ret) - goto out; - i2c_hid_dbg(ihid, "Report Descriptor: %*ph\n", rsize, rdesc); ret = hid_parse_report(hid, rdesc, rsize); -- 2.44.0

1 year

3
2
0 0

[PATCH v2 1/2] Revert "mei: vsc: Call wake_up() in the threaded IRQ handler"

by Wentong Wu

From: Sakari Ailus <sakari.ailus(a)linux.intel.com> This reverts commit 058a38acba15fd8e7b262ec6e17c4204cb15f984. It's not necessary to avoid a spinlock, a sleeping lock on PREEMPT_RT, in an interrupt handler as the interrupt handler itself would be called in a process context if PREEMPT_RT is enabled. So revert the patch. Cc: stable(a)vger.kernel.org # for 6.8 Signed-off-by: Sakari Ailus <sakari.ailus(a)linux.intel.com> --- drivers/misc/mei/vsc-tp.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/misc/mei/vsc-tp.c b/drivers/misc/mei/vsc-tp.c index ecfb70cd057c..968a92a7425d 100644 --- a/drivers/misc/mei/vsc-tp.c +++ b/drivers/misc/mei/vsc-tp.c @@ -419,6 +419,8 @@ static irqreturn_t vsc_tp_isr(int irq, void *data) atomic_inc(&tp->assert_cnt); + wake_up(&tp->xfer_wait); + return IRQ_WAKE_THREAD; } @@ -426,8 +428,6 @@ static irqreturn_t vsc_tp_thread_isr(int irq, void *data) { struct vsc_tp *tp = data; - wake_up(&tp->xfer_wait); - if (tp->event_notify) tp->event_notify(tp->event_notify_context); -- 2.34.1

1 year

2
3
0 0

[PATCH] gpio: cdev: check for NULL labels when sanitizing them for irqs

by Bartosz Golaszewski

From: Bartosz Golaszewski <bartosz.golaszewski(a)linaro.org> We need to take into account that a line's consumer label may be NULL and not try to kstrdup() it in that case but rather pass the NULL pointer up the stack to the interrupt request function. To that end: let make_irq_label() return NULL as a valid return value and use ERR_PTR() instead to signal an allocation failure to callers. Cc: stable(a)vger.kernel.org Fixes: b34490879baa ("gpio: cdev: sanitize the label before requesting the interrupt") Reported-by: Linux Kernel Functional Testing <lkft(a)linaro.org> Closes: https://lore.kernel.org/lkml/20240402093534.212283-1-naresh.kamboju@linaro.… Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski(a)linaro.org> --- drivers/gpio/gpiolib-cdev.c | 19 ++++++++++++++----- 1 file changed, 14 insertions(+), 5 deletions(-) diff --git a/drivers/gpio/gpiolib-cdev.c b/drivers/gpio/gpiolib-cdev.c index fa9635610251..1426cc1c4a28 100644 --- a/drivers/gpio/gpiolib-cdev.c +++ b/drivers/gpio/gpiolib-cdev.c @@ -1085,7 +1085,16 @@ static u32 gpio_v2_line_config_debounce_period(struct gpio_v2_line_config *lc, static inline char *make_irq_label(const char *orig) { - return kstrdup_and_replace(orig, '/', ':', GFP_KERNEL); + char *new; + + if (!orig) + return NULL; + + new = kstrdup_and_replace(orig, '/', ':', GFP_KERNEL); + if (!new) + return ERR_PTR(-ENOMEM); + + return new; } static inline void free_irq_label(const char *label) @@ -1158,8 +1167,8 @@ static int edge_detector_setup(struct line *line, irqflags |= IRQF_ONESHOT; label = make_irq_label(line->req->label); - if (!label) - return -ENOMEM; + if (IS_ERR(label)) + return PTR_ERR(label); /* Request a thread to read the events */ ret = request_threaded_irq(irq, edge_irq_handler, edge_irq_thread, @@ -2217,8 +2226,8 @@ static int lineevent_create(struct gpio_device *gdev, void __user *ip) goto out_free_le; label = make_irq_label(le->label); - if (!label) { - ret = -ENOMEM; + if (IS_ERR(label)) { + ret = PTR_ERR(label); goto out_free_le; } -- 2.40.1

1 year

3
4
0 0

[PATCH] drm/vkms: call drm_atomic_helper_shutdown before drm_dev_put()

by Guo Mengqi

commit 73a82b22963d ("drm/atomic: Fix potential use-after-free in nonblocking commits") introduced drm_dev_get/put() to drm_atomic_helper_shutdown(). And this cause problem in vkms driver exit process. vkms_exit() drm_dev_put() vkms_release() drm_atomic_helper_shutdown() drm_dev_get() drm_dev_put() vkms_release() ------ null pointer access Using 4.19 stable x86 image on qemu, below stacktrace can be triggered by load and unload vkms.ko. root:~ # insmod vkms.ko [ 142.135449] [drm] Supports vblank timestamp caching Rev 2 (21.10.2013). [ 142.138713] [drm] Driver supports precise vblank timestamp query. [ 142.142390] [drm] Initialized vkms 1.0.0 20180514 for virtual device on minor 0 root:~ # rmmod vkms.ko [ 144.093710] BUG: unable to handle kernel NULL pointer dereference at 00000000000000a0 [ 144.097491] PGD 800000023624e067 P4D 800000023624e067 PUD 22ab59067 PMD 0 [ 144.100802] Oops: 0000 [#1] SMP PTI [ 144.102502] CPU: 0 PID: 3615 Comm: rmmod Not tainted 4.19.310 #1 [ 144.104452] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014 [ 144.107238] RIP: 0010:device_del+0x34/0x3a0 ... [ 144.131323] Call Trace: [ 144.131962] ? __die+0x7d/0xc0 [ 144.132711] ? no_context+0x152/0x3b0 [ 144.133605] ? wake_up_q+0x70/0x70 [ 144.134436] ? __do_page_fault+0x342/0x4b0 [ 144.135445] ? __switch_to_asm+0x41/0x70 [ 144.136416] ? __switch_to_asm+0x35/0x70 [ 144.137366] ? page_fault+0x1e/0x30 [ 144.138214] ? __drm_atomic_state_free+0x51/0x60 [ 144.139331] ? device_del+0x34/0x3a0 [ 144.140197] platform_device_del.part.14+0x19/0x70 [ 144.141348] platform_device_unregister+0xe/0x20 [ 144.142458] vkms_release+0x10/0x30 [vkms] [ 144.143449] __drm_atomic_helper_disable_all.constprop.31+0x13b/0x150 [ 144.144980] drm_atomic_helper_shutdown+0x4b/0x90 [ 144.146102] vkms_release+0x18/0x30 [vkms] [ 144.147107] vkms_exit+0x29/0x8ec [vkms] [ 144.148053] __x64_sys_delete_module+0x155/0x220 [ 144.149168] do_syscall_64+0x43/0x100 [ 144.150056] entry_SYSCALL_64_after_hwframe+0x5c/0xc1 It seems that the proper unload sequence is: drm_atomic_helper_shutdown(); drm_dev_put(); Just put drm_atomic_helper_shutdown() before drm_dev_put() should solve the problem. Fixes: 73a82b22963d ("drm/atomic: Fix potential use-after-free in nonblocking commits") Signed-off-by: Guo Mengqi <guomengqi3(a)huawei.com> --- drivers/gpu/drm/vkms/vkms_drv.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/vkms/vkms_drv.c b/drivers/gpu/drm/vkms/vkms_drv.c index b1201c18d3eb..d32e08f17427 100644 --- a/drivers/gpu/drm/vkms/vkms_drv.c +++ b/drivers/gpu/drm/vkms/vkms_drv.c @@ -39,7 +39,6 @@ static void vkms_release(struct drm_device *dev) struct vkms_device *vkms = container_of(dev, struct vkms_device, drm); platform_device_unregister(vkms->platform); - drm_atomic_helper_shutdown(&vkms->drm); drm_mode_config_cleanup(&vkms->drm); drm_dev_fini(&vkms->drm); } @@ -137,6 +136,7 @@ static void __exit vkms_exit(void) } drm_dev_unregister(&vkms_device->drm); + drm_atomic_helper_shutdown(&vkms_device->drm); drm_dev_put(&vkms_device->drm); kfree(vkms_device); -- 2.17.1

1 year

4
5
0 0

[PATCH 02/11] drm/mgag200: Bind I2C lifetime to DRM device

by Thomas Zimmermann

Managed cleanup with devm_add_action_or_reset() will release the I2C adapter when the underlying Linux device goes away. But the connector still refers to it, so this cleanup leaves behind a stale pointer in struct drm_connector.ddc. Bind the lifetime of the I2C adapter to the connector's lifetime by using DRM's managed release. When the DRM device goes away (after the Linux device) DRM will first clean up the connector and then clean up the I2C adapter. Signed-off-by: Thomas Zimmermann <tzimmermann(a)suse.de> Fixes: b279df242972 ("drm/mgag200: Switch I2C code to managed cleanup") Cc: Thomas Zimmermann <tzimmermann(a)suse.de> Cc: Jocelyn Falempe <jfalempe(a)redhat.com> Cc: Dave Airlie <airlied(a)redhat.com> Cc: dri-devel(a)lists.freedesktop.org Cc: <stable(a)vger.kernel.org> # v6.0+ --- drivers/gpu/drm/mgag200/mgag200_i2c.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/mgag200/mgag200_i2c.c b/drivers/gpu/drm/mgag200/mgag200_i2c.c index 1029fef590f9b..4caeb68f3010c 100644 --- a/drivers/gpu/drm/mgag200/mgag200_i2c.c +++ b/drivers/gpu/drm/mgag200/mgag200_i2c.c @@ -31,6 +31,8 @@ #include <linux/i2c.h> #include <linux/pci.h> +#include <drm/drm_managed.h> + #include "mgag200_drv.h" static int mga_i2c_read_gpio(struct mga_device *mdev) @@ -86,7 +88,7 @@ static int mga_gpio_getscl(void *data) return (mga_i2c_read_gpio(mdev) & i2c->clock) ? 1 : 0; } -static void mgag200_i2c_release(void *res) +static void mgag200_i2c_release(struct drm_device *dev, void *res) { struct mga_i2c_chan *i2c = res; @@ -125,5 +127,5 @@ int mgag200_i2c_init(struct mga_device *mdev, struct mga_i2c_chan *i2c) if (ret) return ret; - return devm_add_action_or_reset(dev->dev, mgag200_i2c_release, i2c); + return drmm_add_action_or_reset(dev, mgag200_i2c_release, i2c); } -- 2.44.0

1 year

1
0
0 0

[PATCH 01/11] drm/mgag200: Set DDC timeout in milliseconds

by Thomas Zimmermann

Compute the i2c timeout in jiffies from a value in milliseconds. The original values of 2 jiffies equals 2 milliseconds if HZ has been configured to a value of 1000. This corresponds to 2.2 milliseconds used by most other DRM drivers. Update mgag200 accordingly. Signed-off-by: Thomas Zimmermann <tzimmermann(a)suse.de> Fixes: 414c45310625 ("mgag200: initial g200se driver (v2)") Cc: Dave Airlie <airlied(a)redhat.com> Cc: Maarten Lankhorst <maarten.lankhorst(a)linux.intel.com> Cc: Maxime Ripard <mripard(a)kernel.org> Cc: Thomas Zimmermann <tzimmermann(a)suse.de> Cc: Jocelyn Falempe <jfalempe(a)redhat.com> Cc: dri-devel(a)lists.freedesktop.org Cc: <stable(a)vger.kernel.org> # v3.5+ --- drivers/gpu/drm/mgag200/mgag200_i2c.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/mgag200/mgag200_i2c.c b/drivers/gpu/drm/mgag200/mgag200_i2c.c index 423eb302be7eb..1029fef590f9b 100644 --- a/drivers/gpu/drm/mgag200/mgag200_i2c.c +++ b/drivers/gpu/drm/mgag200/mgag200_i2c.c @@ -114,7 +114,7 @@ int mgag200_i2c_init(struct mga_device *mdev, struct mga_i2c_chan *i2c) i2c->adapter.algo_data = &i2c->bit; i2c->bit.udelay = 10; - i2c->bit.timeout = 2; + i2c->bit.timeout = usecs_to_jiffies(2200); i2c->bit.data = i2c; i2c->bit.setsda = mga_gpio_setsda; i2c->bit.setscl = mga_gpio_setscl; -- 2.44.0

1 year

1
0
0 0

[PATCH] perf/x86/intel/ds: Don't clear the pebs_data_cfg for the last PEBS event

by kan.liang＠linux.intel.com

From: Kan Liang <kan.liang(a)linux.intel.com> The MSR_PEBS_DATA_CFG is to configure which data groups should be generated into a PEBS record. It's shared among counters. If there are different configurations among counters, perf combines all the configurations. The first perf command as below requires a complete PEBS record (including memory info, GPRs, XMMs, and LBRs). The second perf command only requires a basic group. However, after the second perf command is running, the MSR_PEBS_DATA_CFG is cleared. Only a basic group is generated in a PEBS record, which is wrong. The required information for the first perf command is missed. $perf record --intr-regs=AX,SP,XMM0 -a -C 8 -b -W -d -c 100000003 -o /dev/null -e cpu/event=0xd0,umask=0x81/upp & $sleep 5 $perf record --per-thread -c 1 -e cycles:pp --no-timestamp --no-tid taskset -c 8 ./noploop 1000 The first PEBS event is a system-wide PEBS event. The second PEBS event is a per-thread event. When the thread is scheduled out, the intel_pmu_pebs_del() is invoked to update the PEBS state. Since the system-wide event is still available, the cpuc->n_pebs is 1. The cpuc->pebs_data_cfg is cleared. The data configuration for the system-wide PEBS event is lost. The (cpuc->n_pebs == 1) was introduced in the commit b6a32f023fcc ("perf/x86: Fix PEBS threshold initialization"). At that time, it indeed didn't hurt whether the state was updated during the removal. Because only the threshold is updated. The calculation of the threshold takes the last PEBS event into account. However, the commit b752ea0c28e3 ("perf/x86/intel/ds: Flush PEBS DS when changing PEBS_DATA_CFG") delay the threshold update, and clears the PEBS data config, which brings the issue. The PEBS data config update should not be shrink in the removal. Fixes: b752ea0c28e3 ("perf/x86/intel/ds: Flush PEBS DS when changing PEBS_DATA_CFG") Reported-by: Stephane Eranian <eranian(a)google.com> Signed-off-by: Kan Liang <kan.liang(a)linux.intel.com> Cc: stable(a)vger.kernel.org --- arch/x86/events/intel/ds.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/arch/x86/events/intel/ds.c b/arch/x86/events/intel/ds.c index 2641ba620f12..20ddfed3e721 100644 --- a/arch/x86/events/intel/ds.c +++ b/arch/x86/events/intel/ds.c @@ -1237,11 +1237,11 @@ pebs_update_state(bool needed_cb, struct cpu_hw_events *cpuc, struct pmu *pmu = event->pmu; /* - * Make sure we get updated with the first PEBS - * event. It will trigger also during removal, but - * that does not hurt: + * Make sure we get updated with the first PEBS event. + * During removal, the pebs_data_cfg is still valid for + * the last PEBS event. Don't clear it. */ - if (cpuc->n_pebs == 1) + if ((cpuc->n_pebs == 1) && add) cpuc->pebs_data_cfg = PEBS_UPDATE_DS_SW; if (needed_cb != pebs_needs_sched_cb(cpuc)) { -- 2.35.1

1 year

2
1
0 0

[PATCH v3] mtd: rawnand: qcom: Fix broken misc_cmd_type in exec_op

by Christian Marangi

misc_cmd_type in exec_op have multiple problems. With commit a82990c8a409 ("mtd: rawnand: qcom: Add read/read_start ops in exec_op path") it was reworked and generalized but actually dropped the handling of the OP_RESET_DEVICE command. The rework itself was correct with supporting case where a single misc command is handled, but became problematic by the addition of exiting early if we didn't had an OP_BLOCK_ERASE or an OP_PROGRAM_PAGE operation. Also additional logic was added without clear explaination causing the OP_RESET_DEVICE command to be broken on testing it on a ipq806x nandc. Add some additional logic to restore OP_RESET_DEVICE command handling restoring original functionality. Fixes: a82990c8a409 ("mtd: rawnand: qcom: Add read/read_start ops in exec_op path") Cc: stable(a)vger.kernel.org Signed-off-by: Christian Marangi <ansuelsmth(a)gmail.com> --- Changes v3: - Merge patches - Rework commit description Changes v2: - Split patches drivers/mtd/nand/raw/qcom_nandc.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) diff --git a/drivers/mtd/nand/raw/qcom_nandc.c b/drivers/mtd/nand/raw/qcom_nandc.c index b079605c84d3..b8cff9240b28 100644 --- a/drivers/mtd/nand/raw/qcom_nandc.c +++ b/drivers/mtd/nand/raw/qcom_nandc.c @@ -2815,7 +2815,7 @@ static int qcom_misc_cmd_type_exec(struct nand_chip *chip, const struct nand_sub host->cfg0_raw & ~(7 << CW_PER_PAGE)); nandc_set_reg(chip, NAND_DEV0_CFG1, host->cfg1_raw); instrs = 3; - } else { + } else if (q_op.cmd_reg != OP_RESET_DEVICE) { return 0; } @@ -2830,9 +2830,8 @@ static int qcom_misc_cmd_type_exec(struct nand_chip *chip, const struct nand_sub nandc_set_reg(chip, NAND_EXEC_CMD, 1); write_reg_dma(nandc, NAND_FLASH_CMD, instrs, NAND_BAM_NEXT_SGL); - (q_op.cmd_reg == OP_BLOCK_ERASE) ? write_reg_dma(nandc, NAND_DEV0_CFG0, - 2, NAND_BAM_NEXT_SGL) : read_reg_dma(nandc, - NAND_FLASH_STATUS, 1, NAND_BAM_NEXT_SGL); + if (q_op.cmd_reg == OP_BLOCK_ERASE) + write_reg_dma(nandc, NAND_DEV0_CFG0, 2, NAND_BAM_NEXT_SGL); write_reg_dma(nandc, NAND_EXEC_CMD, 1, NAND_BAM_NEXT_SGL); read_reg_dma(nandc, NAND_FLASH_STATUS, 1, NAND_BAM_NEXT_SGL); -- 2.43.0

1 year

3
2
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror April 2024