September 2024 - Linux-stable-mirror

[PATCH 5.10.y] mptcp: fix sometimes-uninitialized warning

by Matthieu Baerts (NGI0)

Nathan reported this issue: $ make -skj"$(nproc)" ARCH=x86_64 LLVM=1 LLVM_IAS=1 mrproper allmodconfig net/mptcp/subflow.o net/mptcp/subflow.c:877:6: warning: variable 'incr' is used uninitialized whenever 'if' condition is true [-Wsometimes-uninitialized] 877 | if (WARN_ON_ONCE(offset > skb->len)) | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ include/asm-generic/bug.h:101:33: note: expanded from macro 'WARN_ON_ONCE' 101 | #define WARN_ON_ONCE(condition) ({ \ | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 102 | int __ret_warn_on = !!(condition); \ | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 103 | if (unlikely(__ret_warn_on)) \ | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 104 | __WARN_FLAGS(BUGFLAG_ONCE | \ | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 105 | BUGFLAG_TAINT(TAINT_WARN)); \ | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 106 | unlikely(__ret_warn_on); \ | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 107 | }) | ~~ net/mptcp/subflow.c:893:6: note: uninitialized use occurs here 893 | if (incr) | ^~~~ net/mptcp/subflow.c:877:2: note: remove the 'if' if its condition is always false 877 | if (WARN_ON_ONCE(offset > skb->len)) | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 878 | goto out; | ~~~~~~~~ net/mptcp/subflow.c:874:18: note: initialize the variable 'incr' to silence this warning 874 | u32 offset, incr, avail_len; | ^ | = 0 1 warning generated. As mentioned by Nathan, this issue is present because 5.10 does not include commit ea4ca586b16f ("mptcp: refine MPTCP-level ack scheduling"), which removed the use of 'incr' in the error path added by this change. This other commit does not really look suitable for stable, hence this dedicated patch for 5.10. Fixes: e93fa44f0714 ("mptcp: fix duplicate data handling") Reported-by: Nathan Chancellor <nathan(a)kernel.org> Closes: https://lore.kernel.org/20240928175524.GA1713144@thelio-3990X Signed-off-by: Matthieu Baerts (NGI0) <matttbe(a)kernel.org> --- net/mptcp/subflow.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c index 8a0ef50c307c..843c61ebd421 100644 --- a/net/mptcp/subflow.c +++ b/net/mptcp/subflow.c @@ -871,7 +871,7 @@ static void mptcp_subflow_discard_data(struct sock *ssk, struct sk_buff *skb, struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(ssk); bool fin = TCP_SKB_CB(skb)->tcp_flags & TCPHDR_FIN; struct tcp_sock *tp = tcp_sk(ssk); - u32 offset, incr, avail_len; + u32 offset, incr = 0, avail_len; offset = tp->copied_seq - TCP_SKB_CB(skb)->seq; if (WARN_ON_ONCE(offset > skb->len)) -- 2.45.2

6 months

2
1
0 0

[PATCH 5.10 v2 0/3] Re-adapt "bpf: Fix DEVMAP_HASH overflow check on 32-bit arches"

by Pu Lehui

Commit 70294d8bc31f ("bpf: Eliminate rlimit-based memory accounting for devmap maps") relies on the v5.11+ basic mechanism of memcg-based memory accounting [0]. The commit cannot be independently backported to the 5.10 stable branch, otherwise the related memory when creating devmap will be unrestricted and the associated bpf selftest map_ptr will fail. Let's roll back to rlimit-based memory accounting mode for devmap and re-adapt the commit 225da02acdc9 ("bpf: Fix DEVMAP_HASH overflow check on 32-bit arches") to the 5.10 stable branch. Link: https://lore.kernel.org/bpf/20201201215900.3569844-1-guro@fb.com [0] Pu Lehui (2): Revert "bpf: Fix DEVMAP_HASH overflow check on 32-bit arches" Revert "bpf: Eliminate rlimit-based memory accounting for devmap maps" Toke Høiland-Jørgensen (1): bpf: Fix DEVMAP_HASH overflow check on 32-bit arches kernel/bpf/devmap.c | 20 ++++++++++++++++++-- 1 file changed, 18 insertions(+), 2 deletions(-) -- 2.34.1

6 months

2
4
0 0

[PATCH 5.10.y 0/3] x86: Complete backports for x86_match_cpu()

by Ricardo Neri

Hi, Upstream commit 93022482b294 ("x86/cpu: Fix x86_match_cpu() to match just X86_VENDOR_INTEL") introduced a flags member to struct x86_cpu_id. Bit 0 of x86_cpu.id.flags must be set to 1 for x86_match_cpu() to work correctly. This upstream commit has been backported to 5.10.y. Callers that use the X86_MATCH_*() family of macros to compose the argument of x86_match_cpu() function correctly. Callers that use their own custom mechanisms may not work if they fail to set x86_cpu_id.flags correctly. There are three remaining callers in 5.10.y that use their own mechanisms: a) setup_pcid(), b) rapl_msr_probe(), and c) goodix_add_acpi_gpio_ mappings(). a) caused a regression that Thomas Lindroth reported in [1]. b) works by luck but it still populates its x86_cpu_id[] array incorrectly. I am not aware of any reports on c), but inspecting the code reveals that it will fail to identify INTEL_FAM6_ATOM_SILVERMONT for the reason described above. I backported three patches that fix these bugs in mainline. Hopefully the authors and/or maintainers can ack the backports? I tested patches 2/3 and 3/3 on Tiger Lake, Alder Lake, and Meteor Lake systems as the two involved callers behave differently on these systems. I wrote the testing details in each patch separately. I could not test patch 1/3 because I do not have access to Bay Trail hardware. Thanks and BR, Ricardo [1]. https://lore.kernel.org/all/eb709d67-2a8d-412f-905d-f3777d897bfa@gmail.com/ Hans de Goede (1): Input: goodix - use the new soc_intel_is_byt() helper Sumeet Pawnikar (1): powercap: RAPL: fix invalid initialization for pl4_supported field Tony Luck (1): x86/mm: Switch to new Intel CPU model defines arch/x86/mm/init.c | 16 ++++++---------- drivers/input/touchscreen/goodix.c | 18 ++---------------- drivers/powercap/intel_rapl_msr.c | 2 +- 3 files changed, 9 insertions(+), 27 deletions(-) -- 2.34.1

6 months

2
4
0 0

Fix regression from "drm/amd/display: Fix MST BW calculation Regression"

by Mario Limonciello

Hello, The commit 338567d17627 ("drm/amd/display: Fix MST BW calculation Regression") caused a regression with some MST displays due to a mistake. For 6.11.y here is the series of commits that fixes it: commit 4599aef0c97b ("drm/amd/display: Fix Synaptics Cascaded Panamera DSC Determination") commit ecc4038ec1de ("drm/amd/display: Add DSC Debug Log") commit b2b4afb9cf07 ("drm/amdgpu/display: Fix a mistake in revert commit") Can you please bring them back to 6.11.y? Thanks!

6 months

2
3
0 0

[PATCH 1/2] drm/xe/ct: prevent UAF in send_recv()

by Matthew Auld

Ensure we serialize with completion side to prevent UAF with fence going out of scope on the stack, since we have no clue if it will fire after the timeout before we can erase from the xa. Also we have some dependent loads and stores for which we need the correct ordering, and we lack the needed barriers. Fix this by grabbing the ct->lock after the wait, which is also held by the completion side. Fixes: dd08ebf6c352 ("drm/xe: Introduce a new DRM driver for Intel GPUs") Signed-off-by: Matthew Auld <matthew.auld(a)intel.com> Cc: Matthew Brost <matthew.brost(a)intel.com> Cc: Badal Nilawar <badal.nilawar(a)intel.com> Cc: <stable(a)vger.kernel.org> # v6.8+ --- drivers/gpu/drm/xe/xe_guc_ct.c | 17 ++++++++++++++++- 1 file changed, 16 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/xe/xe_guc_ct.c b/drivers/gpu/drm/xe/xe_guc_ct.c index 4b95f75b1546..232eb69bd8e4 100644 --- a/drivers/gpu/drm/xe/xe_guc_ct.c +++ b/drivers/gpu/drm/xe/xe_guc_ct.c @@ -903,16 +903,26 @@ static int guc_ct_send_recv(struct xe_guc_ct *ct, const u32 *action, u32 len, } ret = wait_event_timeout(ct->g2h_fence_wq, g2h_fence.done, HZ); + + /* + * Ensure we serialize with completion side to prevent UAF with fence going out of scope on + * the stack, since we have no clue if it will fire after the timeout before we can erase + * from the xa. Also we have some dependent loads and stores below for which we need the + * correct ordering, and we lack the needed barriers. + */ + mutex_lock(&ct->lock); if (!ret) { xe_gt_err(gt, "Timed out wait for G2H, fence %u, action %04x", g2h_fence.seqno, action[0]); xa_erase_irq(&ct->fence_lookup, g2h_fence.seqno); + mutex_unlock(&ct->lock); return -ETIME; } if (g2h_fence.retry) { xe_gt_dbg(gt, "H2G action %#x retrying: reason %#x\n", action[0], g2h_fence.reason); + mutex_unlock(&ct->lock); goto retry; } if (g2h_fence.fail) { @@ -921,7 +931,12 @@ static int guc_ct_send_recv(struct xe_guc_ct *ct, const u32 *action, u32 len, ret = -EIO; } - return ret > 0 ? response_buffer ? g2h_fence.response_len : g2h_fence.response_data : ret; + if (ret > 0) + ret = response_buffer ? g2h_fence.response_len : g2h_fence.response_data; + + mutex_unlock(&ct->lock); + + return ret; } /** -- 2.46.2

6 months

2
2
0 0

[PATCH net 03/10] ice: Fix improper handling of refcount in ice_sriov_set_msix_vec_count()

by Tony Nguyen

From: Gui-Dong Han <hanguidong02(a)outlook.com> This patch addresses an issue with improper reference count handling in the ice_sriov_set_msix_vec_count() function. First, the function calls ice_get_vf_by_id(), which increments the reference count of the vf pointer. If the subsequent call to ice_get_vf_vsi() fails, the function currently returns an error without decrementing the reference count of the vf pointer, leading to a reference count leak. The correct behavior, as implemented in this patch, is to decrement the reference count using ice_put_vf(vf) before returning an error when vsi is NULL. Second, the function calls ice_sriov_get_irqs(), which sets vf->first_vector_idx. If this call returns a negative value, indicating an error, the function returns an error without decrementing the reference count of the vf pointer, resulting in another reference count leak. The patch addresses this by adding a call to ice_put_vf(vf) before returning an error when vf->first_vector_idx < 0. This bug was identified by an experimental static analysis tool developed by our team. The tool specializes in analyzing reference count operations and identifying potential mismanagement of reference counts. In this case, the tool flagged the missing decrement operation as a potential issue, leading to this patch. Fixes: 4035c72dc1ba ("ice: reconfig host after changing MSI-X on VF") Fixes: 4d38cb44bd32 ("ice: manage VFs MSI-X using resource tracking") Cc: stable(a)vger.kernel.org Signed-off-by: Gui-Dong Han <hanguidong02(a)outlook.com> Reviewed-by: Simon Horman <horms(a)kernel.org> Tested-by: Rafal Romanowski <rafal.romanowski(a)intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen(a)intel.com> --- drivers/net/ethernet/intel/ice/ice_sriov.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/drivers/net/ethernet/intel/ice/ice_sriov.c b/drivers/net/ethernet/intel/ice/ice_sriov.c index e34fe2516ccc..c2d6b2a144e9 100644 --- a/drivers/net/ethernet/intel/ice/ice_sriov.c +++ b/drivers/net/ethernet/intel/ice/ice_sriov.c @@ -1096,8 +1096,10 @@ int ice_sriov_set_msix_vec_count(struct pci_dev *vf_dev, int msix_vec_count) return -ENOENT; vsi = ice_get_vf_vsi(vf); - if (!vsi) + if (!vsi) { + ice_put_vf(vf); return -ENOENT; + } prev_msix = vf->num_msix; prev_queues = vf->num_vf_qs; @@ -1142,8 +1144,10 @@ int ice_sriov_set_msix_vec_count(struct pci_dev *vf_dev, int msix_vec_count) vf->num_msix = prev_msix; vf->num_vf_qs = prev_queues; vf->first_vector_idx = ice_sriov_get_irqs(pf, vf->num_msix); - if (vf->first_vector_idx < 0) + if (vf->first_vector_idx < 0) { + ice_put_vf(vf); return -EINVAL; + } if (needs_rebuild) { ice_vf_reconfig_vsi(vf); -- 2.42.0

6 months

2
1
0 0

[PATCH net 09/10] idpf: use actual mbx receive payload length

by Tony Nguyen

From: Joshua Hay <joshua.a.hay(a)intel.com> When a mailbox message is received, the driver is checking for a non 0 datalen in the controlq descriptor. If it is valid, the payload is attached to the ctlq message to give to the upper layer. However, the payload response size given to the upper layer was taken from the buffer metadata which is _always_ the max buffer size. This meant the API was returning 4K as the payload size for all messages. This went unnoticed since the virtchnl exchange response logic was checking for a response size less than 0 (error), not less than exact size, or not greater than or equal to the max mailbox buffer size (4K). All of these checks will pass in the success case since the size provided is always 4K. However, this breaks anyone that wants to validate the exact response size. Fetch the actual payload length from the value provided in the descriptor data_len field (instead of the buffer metadata). Unfortunately, this means we lose some extra error parsing for variable sized virtchnl responses such as create vport and get ptypes. However, the original checks weren't really helping anyways since the size was _always_ 4K. Fixes: 34c21fa894a1 ("idpf: implement virtchnl transaction manager") Cc: stable(a)vger.kernel.org # 6.9+ Signed-off-by: Joshua Hay <joshua.a.hay(a)intel.com> Reviewed-by: Przemek Kitszel <przemyslaw.kitszel(a)intel.com> Tested-by: Krishneil Singh <krishneil.k.singh(a)intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen(a)intel.com> --- drivers/net/ethernet/intel/idpf/idpf_virtchnl.c | 9 +-------- 1 file changed, 1 insertion(+), 8 deletions(-) diff --git a/drivers/net/ethernet/intel/idpf/idpf_virtchnl.c b/drivers/net/ethernet/intel/idpf/idpf_virtchnl.c index 70986e12da28..3c0f97650d72 100644 --- a/drivers/net/ethernet/intel/idpf/idpf_virtchnl.c +++ b/drivers/net/ethernet/intel/idpf/idpf_virtchnl.c @@ -666,7 +666,7 @@ idpf_vc_xn_forward_reply(struct idpf_adapter *adapter, if (ctlq_msg->data_len) { payload = ctlq_msg->ctx.indirect.payload->va; - payload_size = ctlq_msg->ctx.indirect.payload->size; + payload_size = ctlq_msg->data_len; } xn->reply_sz = payload_size; @@ -1295,10 +1295,6 @@ int idpf_send_create_vport_msg(struct idpf_adapter *adapter, err = reply_sz; goto free_vport_params; } - if (reply_sz < IDPF_CTLQ_MAX_BUF_LEN) { - err = -EIO; - goto free_vport_params; - } return 0; @@ -2602,9 +2598,6 @@ int idpf_send_get_rx_ptype_msg(struct idpf_vport *vport) if (reply_sz < 0) return reply_sz; - if (reply_sz < IDPF_CTLQ_MAX_BUF_LEN) - return -EIO; - ptypes_recvd += le16_to_cpu(ptype_info->num_ptypes); if (ptypes_recvd > max_ptype) return -EINVAL; -- 2.42.0

6 months

1
0
0 0

[PATCH 1/4] ACPI: resource: Remove duplicate Asus E1504GAB IRQ override

by Hans de Goede

Commit d2aaf1996504 ("ACPI: resource: Add DMI quirks for ASUS Vivobook E1504GA and E1504GAB") does exactly what the subject says, adding DMI matches for both the E1504GA and E1504GAB. But DMI_MATCH() does a substring match, so checking for E1504GA will also match E1504GAB. Drop the unnecessary E1504GAB entry since that is covered already by the E1504GA entry. Fixes: d2aaf1996504 ("ACPI: resource: Add DMI quirks for ASUS Vivobook E1504GA and E1504GAB") Cc: Ben Mayo <benny1091(a)gmail.com> Cc: stable(a)vger.kernel.org Signed-off-by: Hans de Goede <hdegoede(a)redhat.com> --- drivers/acpi/resource.c | 9 +-------- 1 file changed, 1 insertion(+), 8 deletions(-) diff --git a/drivers/acpi/resource.c b/drivers/acpi/resource.c index 8a4726e2eb69..1ff251fd1901 100644 --- a/drivers/acpi/resource.c +++ b/drivers/acpi/resource.c @@ -511,19 +511,12 @@ static const struct dmi_system_id irq1_level_low_skip_override[] = { }, }, { - /* Asus Vivobook E1504GA */ + /* Asus Vivobook E1504GA* */ .matches = { DMI_MATCH(DMI_SYS_VENDOR, "ASUSTeK COMPUTER INC."), DMI_MATCH(DMI_BOARD_NAME, "E1504GA"), }, }, - { - /* Asus Vivobook E1504GAB */ - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "ASUSTeK COMPUTER INC."), - DMI_MATCH(DMI_BOARD_NAME, "E1504GAB"), - }, - }, { /* Asus Vivobook Pro N6506MV */ .matches = { -- 2.46.0

6 months

3
8
0 0

[PATCH] pidfs: check for valid pid namespace

by Christian Brauner

When we access a no-current task's pid namespace we need check that the task hasn't been reaped in the meantime and it's pid namespace isn't accessible anymore. The user namespace is fine because it is only released when the last reference to struct task_struct is put and exit_creds() is called. Fixes: 5b08bd408534 ("pidfs: allow retrieval of namespace file descriptors") CC: stable(a)vger.kernel.org # v6.11 Signed-off-by: Christian Brauner <brauner(a)kernel.org> --- fs/pidfs.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/fs/pidfs.c b/fs/pidfs.c index 7ffdc88dfb52..80675b6bf884 100644 --- a/fs/pidfs.c +++ b/fs/pidfs.c @@ -120,6 +120,7 @@ static long pidfd_ioctl(struct file *file, unsigned int cmd, unsigned long arg) struct nsproxy *nsp __free(put_nsproxy) = NULL; struct pid *pid = pidfd_pid(file); struct ns_common *ns_common = NULL; + struct pid_namespace *pid_ns; if (arg) return -EINVAL; @@ -202,7 +203,9 @@ static long pidfd_ioctl(struct file *file, unsigned int cmd, unsigned long arg) case PIDFD_GET_PID_NAMESPACE: if (IS_ENABLED(CONFIG_PID_NS)) { rcu_read_lock(); - ns_common = to_ns_common( get_pid_ns(task_active_pid_ns(task))); + pid_ns = task_active_pid_ns(task); + if (pid_ns) + ns_common = to_ns_common(get_pid_ns(pid_ns)); rcu_read_unlock(); } break; -- 2.45.2

6 months

2
1
0 0

[PATCH] scsi: fnic: move flush_work initialization out of if block

by Martin Wilck

After commit 379a58caa199 ("scsi: fnic: Move fnic_fnic_flush_tx() to a work queue"), it can happen that a work item is sent to an uninitialized work queue. This may has the effect that the item being queued is never actually queued, and any further actions depending on it will not proceed. The following warning is observed while the fnic driver is loaded: kernel: WARNING: CPU: 11 PID: 0 at ../kernel/workqueue.c:1524 __queue_work+0x373/0x410 kernel: <IRQ> kernel: queue_work_on+0x3a/0x50 kernel: fnic_wq_copy_cmpl_handler+0x54a/0x730 [fnic 62fbff0c42e7fb825c60a55cde2fb91facb2ed24] kernel: fnic_isr_msix_wq_copy+0x2d/0x60 [fnic 62fbff0c42e7fb825c60a55cde2fb91facb2ed24] kernel: __handle_irq_event_percpu+0x36/0x1a0 kernel: handle_irq_event_percpu+0x30/0x70 kernel: handle_irq_event+0x34/0x60 kernel: handle_edge_irq+0x7e/0x1a0 kernel: __common_interrupt+0x3b/0xb0 kernel: common_interrupt+0x58/0xa0 kernel: </IRQ> It has been observed that this may break the rediscovery of fibre channel devices after a temporary fabric failure. This patch fixes it by moving the work queue initialization out of an if block in fnic_probe(). Signed-off-by: Martin Wilck <mwilck(a)suse.com> Fixes: 379a58caa199 ("scsi: fnic: Move fnic_fnic_flush_tx() to a work queue") Cc: stable(a)vger.kernel.org --- drivers/scsi/fnic/fnic_main.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/scsi/fnic/fnic_main.c b/drivers/scsi/fnic/fnic_main.c index 0044717d4486..adec0df24bc4 100644 --- a/drivers/scsi/fnic/fnic_main.c +++ b/drivers/scsi/fnic/fnic_main.c @@ -830,7 +830,6 @@ static int fnic_probe(struct pci_dev *pdev, const struct pci_device_id *ent) spin_lock_init(&fnic->vlans_lock); INIT_WORK(&fnic->fip_frame_work, fnic_handle_fip_frame); INIT_WORK(&fnic->event_work, fnic_handle_event); - INIT_WORK(&fnic->flush_work, fnic_flush_tx); skb_queue_head_init(&fnic->fip_frame_queue); INIT_LIST_HEAD(&fnic->evlist); INIT_LIST_HEAD(&fnic->vlans); @@ -948,6 +947,7 @@ static int fnic_probe(struct pci_dev *pdev, const struct pci_device_id *ent) INIT_WORK(&fnic->link_work, fnic_handle_link); INIT_WORK(&fnic->frame_work, fnic_handle_frame); + INIT_WORK(&fnic->flush_work, fnic_flush_tx); skb_queue_head_init(&fnic->frame_queue); skb_queue_head_init(&fnic->tx_queue); -- 2.46.1

6 months

3
3
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror September 2024