| CVE |
Vendors |
Products |
Updated |
CVSS v3.1 |
| In the Linux kernel, the following vulnerability has been resolved:
dmaengine: fix use-after-free in dma_chan_put() and dma_release_channel()
When dma_device_put() drops the last reference on chan->device->ref,
dma_device_release() runs and may free the dma_device along with its
channels.
dma_chan_put() then still reads chan->device->owner via
dma_chan_to_owner() for the trailing module_put(). KASAN catches it:
slab-use-after-free in dma_chan_put+0x3e6/0x4c0
Read of size 8 by task insmod/6319
Freed by task 6319:
kfree+0x225/0x470
dma_chan_put+0x395/0x4c0
dmaengine_put+0xf8/0x160
Cache the module owner in dma_chan_put() before the put so the trailing
module_put() does not need chan->device. |
| In the Linux kernel, the following vulnerability has been resolved:
dmaengine: Fix device kref underflow in dma_chan_put()
dma_chan_get() takes chan->device->ref only on the slow path:
/* no kref on fast path */
if (chan->client_count) {
__module_get(owner);
chan->client_count++;
return 0;
}
if (!try_module_get(owner))
return -ENODEV;
if (!dma_device_get(chan->device)) { // calls kref_get_unless_zero()
dma_chan_put() drops the ref unconditionally, so every fast-path
get/put pair drops one extra device reference.
The bug fires when two conditions hold together: a non-private
provider has a persistent client holding chan->client_count > 0
and another client cycles dmaengine_get()/dmaengine_put().
When the kref hits zero, the subsequent dma_find_channel() returns
NULL even though the provider module is still loaded.
Fix this by dropping device->ref only on the last put, matching the
single slow-path get. |
| In the Linux kernel, the following vulnerability has been resolved:
RDMA/rtrs-clt: Fix CQ pool leak when connect is interrupted
The client borrows shared CQ credits in the ADDR_RESOLVED handler via
ib_cq_pool_get(), before the peer is connected. create_cm() can return
-ERESTARTSYS from wait_event_interruptible_timeout() without destroying
the CM ID. The init_conns() and stop-and-destroy paths then call
destroy_con_cq_qp() while cq is still NULL (no PUT) and only afterwards
rdma_destroy_id().
CMA serializes the handler against rdma_destroy_id() with handler_mutex,
but that does not order the GET against destroy_con_cq_qp(). If
ADDR_RESOLVED has already passed the DESTROYING check, it can take
con_mutex, GET credits, and then lose the con to kfree. Device
unregister later hits WARN_ON(cq->cqe_used) in ib_cq_pool_cleanup().
Set a per-connection flag under con_mutex before CQ/QP teardown so a
racing ADDR_RESOLVED cannot borrow credits after teardown has begun. |
| In the Linux kernel, the following vulnerability has been resolved:
IB/iser: reject a remote invalidation of an unregistered direction
A write command whose data is sent entirely as immediate data is not
registered. iser_reg_mem_fastreg() takes the DMA key path and leaves
rdma_reg[ISER_DIR_OUT].desc at NULL, while iser_dma_map_task_data() has
already set dir[ISER_DIR_OUT].
iser_check_remote_inv() looks at dir[] alone and hands the descriptor to
iser_inv_desc(), which reads desc->sig_protected. A target that answers
such a command with IB_WR_SEND_WITH_INV faults the initiator.
Leaving those commands unregistered is deliberate.
The same function already terminates the connection when a target sends
a remote invalidation the initiator did not ask for. A target that
invalidates a direction that was never registered is in the same class,
so give it the same answer.
Oops: general protection fault, probably for non-canonical address 0xdffffc0000000004: 0000 [#1] SMP KASAN NOPTI
KASAN: null-ptr-deref in range [0x0000000000000020-0x0000000000000027]
CPU: 0 UID: 0 PID: 40 Comm: kworker/u8:2 Not tainted 7.2.0-rc5-ISERHOST-gf5098b6bae76-dirty #3 PREEMPT(lazy)
Hardware name: QEMU Ubuntu 24.04 PC v2 (i440FX + PIIX, arch_caps fix, 1996), BIOS 1.16.3-debian-1.16.3-2 04/01/2014
Workqueue: rxe_wq do_work
RIP: 0010:iser_task_rsp+0x6d6/0xec0
Code: 48 c1 ea 03 80 3c 02 00 0f 85 ba 06 00 00 48 8b 9b 78 01 00 00 48 b8 00 00 00 00 00 fc ff df 48 8d 7b 20 48 89 fa 48 c1 ea 03 <0f> b6 04 02 84 c0 74 06 0f 8e 76 06 00 00 80 7b 20 00 0f 84 3d 04
RSP: 0018:ffff88811b008db8 EFLAGS: 00010202
RAX: dffffc0000000000 RBX: 0000000000000000 RCX: 0000000000001848
RDX: 0000000000000004 RSI: 1ffff11021587b12 RDI: 0000000000000020
RBP: ffff88810adc1ae4 R08: ffff888109b7f860 R09: ffffffff90a922c0
R10: ffff88810adc1a1c R11: 000000000000003c R12: ffff888109b7f800
R13: ffff88810adc1acc R14: ffff888109b7f820 R15: 0000000000000000
FS: 0000000000000000(0000) GS:ffff88818a676000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00000000005afe2b CR3: 000000010af23005 CR4: 0000000000770ef0
PKRU: 55555554
Call Trace:
<IRQ>
__ib_process_cq+0xe1/0x390
ib_poll_handler+0x6e/0x200
irq_poll_softirq+0x1df/0x480
? clockevents_program_event+0x2ba/0x860
? __pfx_irq_poll_softirq+0x10/0x10
handle_softirqs+0x18e/0x590
? __pfx_handle_softirqs+0x10/0x10
? __hrtimer_rearm_deferred+0x156/0x450
do_softirq+0x3b/0x60
</IRQ>
<TASK>
__local_bh_enable_ip+0x61/0x70
__alloc_skb+0x732/0x890
? _raw_spin_lock_irqsave+0x85/0xe0
? __pfx___alloc_skb+0x10/0x10
? _raw_read_unlock_irqrestore+0x16/0x50
rxe_init_packet+0x16b/0x4f0
prepare_ack_packet+0xb8/0x830
rxe_receiver+0x499/0x9980
? __pfx_rxe_receiver+0x10/0x10
? rxe_completer+0x29e5/0x38c0
? hrtimer_start_range_ns_common+0x75f/0x1730
? hrtimer_start_range_ns+0xa6/0x2c0
? __pfx__raw_spin_lock_irqsave+0x10/0x10
? __pfx_rxe_receiver+0x10/0x10
do_work+0x144/0x470
process_one_work+0x633/0x1030
? assign_work+0x11d/0x370
worker_thread+0x45b/0xd10
? __pfx_worker_thread+0x10/0x10
kthread+0x2c6/0x3b0
? recalc_sigpending+0x15c/0x1e0
? __pfx_kthread+0x10/0x10
ret_from_fork+0x36e/0x5a0
? __pfx_ret_from_fork+0x10/0x10
? __switch_to+0x572/0xdd0
? __pfx_kthread+0x10/0x10
ret_from_fork_asm+0x1a/0x30
</TASK>
Modules linked in:
---[ end trace 0000000000000000 ]--- |
| In the Linux kernel, the following vulnerability has been resolved:
RDMA/rxe: Restore HMM_PFN_WRITE check in ODP write paths
Commit 0b261d7c1cd3 ("RDMA/rxe: Break endless pagefault loop for RO
pages") dropped the access permission test from rxe_check_pagefault()
and left only HMM_PFN_VALID. A page faulted in read-only, for example
a page-cache folio behind a PROT_READ file mapping, then satisfies the
check and ODP write operations (RDMA WRITE, RDMA READ response, SEND
payload, atomics) modify it through kmap without ever breaking CoW.
An unprivileged user can register an ODP MR over such a mapping and
have incoming RDMA traffic overwrite the page cache of a file it only
holds O_RDONLY, including /etc/passwd or setuid binaries. This is the
same primitive class as Dirty COW and CVE-2022-2590.
mlx5 has the missing invariant: its ODP path sets the device write bit
only for pfns that carry HMM_PFN_WRITE. Restore it in rxe by requiring
HMM_PFN_WRITE in rxe_check_pagefault() for every operation except
RXE_PAGEFAULT_RDONLY. A write to a non-writable VMA now fails the one
fault attempt with -EPERM from hmm_vma_fault() instead of re-faulting
forever. For a writable VMA the fault breaks CoW and the write lands
in the private page.
Keep pmem flushes on the read-only check. arch_wb_cache_pmem() never
modifies memory, and the FLUSH access bits do not make the umem
writable, so classifying flushes as writes would make every flush
against a flush-only MR fail. |
| When a private repository is transferred to a user who lacks access, Gitea grants that recipient temporary read access as a collaborator so they can review the repository. Rejecting or cancelling the transfer did not revoke this collaboration, so the named recipient kept persistent read access to the private repository, including its code, issues, pull requests and wiki, and could clone it. The repository owner was not notified. Transfer-granted access is now removed while collaborations that existed before the transfer are preserved. |
| Gitea expanded a workflow's static `strategy.matrix` into its full Cartesian product without a size limit when creating a run, before the fork pull request approval gate applied. A user who can open a pull request from a fork could submit a small workflow file whose matrix expands to a very large number of jobs, consuming server memory and potentially terminating the Gitea process. No runner is required. Static matrices above 256 combinations are now rejected before expansion. |
| When Gitea's web installer is reachable against a database that already contains users, such as after `INSTALL_LOCK` has been reset to `false`, submitting the install form with an administrator username matching an existing account issued an authenticated session for that account without verifying its password. If the account is an administrator, the session grants full administrative access, including changing the account's password. Databases with a single user also did not require the reinstall confirmation. |
| With `[migrations] ALLOWED_DOMAINS` set to a matching entry such as `*` or a hostname wildcard, Gitea's migration URL validation could permit reserved and link-local addresses, such as `169.254.169.254`, even when `ALLOW_LOCALNETWORKS = false`. The local-network block list did not cover these ranges, and a hostname matching the allow list was accepted regardless of its resolved address. A user who can start migrations on such an instance could reach these addresses from the Gitea server; the default empty `ALLOWED_DOMAINS` configuration is not affected. |
| A repository's external issue tracker regular expression containing alternating capture groups could produce invalid slice indexes when Gitea rendered issue references, causing a runtime panic that terminated the Gitea process. A user who can edit a repository's external issue tracker settings could make any later rendering of matching content, such as viewing a README, crash the instance for all users. |
| When processing issue and comment bodies, Gitea scanned the entire preceding text for action keywords such as "closes" or "fixes" once per Markdown link, giving processing time quadratic in the input size. An authenticated user able to submit issue or comment content could send a crafted body of about 1 MB that keeps a CPU core busy for several minutes while holding a database transaction open. |
| Gitea Actions decided whether a fork pull request run needed approval based on the user who triggered the event rather than the pull request author. For `pull_request` activity triggered by a maintainer during ordinary triage, such as adding a label, the run was created without requiring approval, while the workflow definition was still taken from the fork head. Where Actions is enabled and a matching runner is registered, fork-controlled workflow code could run on the base repository's runners without an explicit approval. |
| Vim is an open source, command line text editor. Prior to 9.2.0840, runtime/plugin/netrwPlugin.vim loads netrw and runtime/pack/dist/opt/netrw/autoload/netrw.vim constructs Bookmarks, History, and Targets menu entries by interpolating attacker-controlled directory paths into executed :menu commands. s:NetrwBookmarkMenu(), s:NetrwTgtMenu(), g:netrw_menu_escape, EX_TRLBAR, and netrw#MakeTgt() fail to neutralize the | command separator or single quotes at five construction sites, allowing a crafted path browsed or bookmarked in GUI Vim to execute arbitrary Ex and operating-system commands. This issue is fixed in version 9.2.0840. |
| When a push was authenticated with a deploy key, Gitea recorded the repository owner as the pusher, so permission checks in the push hook pipeline evaluated the owner instead of the deploy key. A holder of a writable deploy key could create protected tags without being on the tag allow list and change repository visibility through push options, for example making a private repository public. Pull requests created through the AGit flow with a deploy key were also attributed to the owner. |
| Gitea validates a repository migration hostname against its network allow and block lists before invoking Git, but the Git subprocess independently resolves the hostname when connecting. An attacker who can start a migration and control the destination's DNS can change the address between validation and connection to reach a blocked internal address. The affected path is the Git clone operation; validation in the migration HTTP client's dialer does not protect the independently connecting Git subprocess. |
| Gitea validated the initial remote URL for push mirrors, wiki remote checks, and fetches of migrated pull request heads, but the subsequent raw Git operations followed HTTP redirects without revalidating the destination. A repository administrator using a policy-allowed endpoint that redirects could make Gitea's Git client send requests to an address that the outbound host policy would otherwise block. The impact depends on the configured policy and the internal services reachable from the server. |
| A flaw was found in CRI-O's `bind_mount_prefix` handling. When configured with a non-empty `bind_mount_prefix`, a malicious container or local attacker could use a Container Runtime Interface (CRI) hostPath containing an intermediate absolute symlink. This could cause the bind mount source to resolve outside the intended prefixed root, potentially leading to unauthorized access to files or privilege escalation on the host system. |
| Changing an organization team's permission through the API with only the `permission` field did not rebuild the team's per-unit access, and the requested level was not applied as a cap. After an organization owner demoted a team, for example from admin to read, the team's members kept their previous unit permissions, including write access to the team's repositories. The web form was not affected. |
| Gitea accepted pushed Git trees containing two entries with the same name, which Git's own consistency checks reject. Gitea's web views resolved such a path to the first entry, while `git checkout`, Gitea Actions, and release archives use the last. A contributor could open a pull request whose diff and file views show benign content while CI and checkouts at the same commit use different, attacker-controlled content. Incoming objects are now checked for consistency; objects already stored in existing repositories are not rescanned. |
| Gitea validated a push mirror's remote address against the `[migrations]` allow and block lists only when the mirror was created. Each synchronization passed the stored address directly to `git push`, so a name that later resolved to a blocked or internal address was still reached. A user with administrator access to a repository, which includes repositories they create themselves, could aim push mirror synchronization at internal Git services and force-push the repository's contents to them. |
