Where
-Infinity
0
Severity
9.9
OS Command Injection, Input Validation
AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H

A trivial sandbox (enabled with the -dSAFER option) escape flaw was found in the ghostscript interpreter by injecting a specially crafted pipe command. This flaw allows a specially crafted document to execute arbitrary commands on the system in the context of the ghostscript interpreter. The highest threat from this vulnerability is to confidentiality, integrity, as well as system availability.

1 / 2
Source: MITRE
First published (updated )
Severity
9.8
EPSS
0.06%
Buffer Overflow
AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

An issue was discovered in Artifex Ghostscript before 10.05.0. The BJ10V device has a Print buffer overflow in contrib/japanese/gdev10v.c.

1 / 2
Source: NVD
First published (updated )
Severity
9.8
EPSS
0.06%
Buffer Overflow
AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

An issue was discovered in Artifex Ghostscript before 10.05.0. The NPDL device has a Compression buffer overflow for contrib/japanese/gdevnpdl.c.

1 / 3
Source: Debian
First published (updated )
Severity
9.8
EPSS
0.06%
Buffer Overflow
AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

An issue was discovered in Artifex Ghostscript before 10.05.0. The DOCXWRITE TXTWRITE device has a text buffer overflow via long characters to devices/vector/doccommon.c.

1 / 2
Source: MITRE
First published (updated )
Severity
9.8
EPSS
0.07%
Path Traversal
AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

An issue was discovered in Artifex Ghostscript before 10.05.0. Access to arbitrary files can occur through a truncated path with invalid UTF-8 characters, for base/gpmswin.c and base/winrtsup.cpp.

First published (updated )
Severity
9.8
Input Validation
AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H

Summary

In vLLM 0.30.0, ImageMediaIO passes attacker-controlled image bytes to Pillow's format autodetection without restricting the accepted formats; the request's MIME label does not constrain decoder selection. An EPS payload labeled image/png reached the host's Ghostscript executable. On a host with Ghostscript affected by CVE-2024-29510, this exposes the known Ghostscript sandbox escape through vLLM's image-loading path and can execute commands with the vLLM process's privileges. A bounded marker-command payload executed through the PyPI wheel's actual ImageMediaIO.loadbase64() implementation using Debian Ghostscript 10.03.0. The loader-level execution is reproduced; the HTTP request path is source-traced but was not exercised end to end. This report concerns vLLM's untrusted-format dispatch and does not claim a new Ghostscript vulnerability. Exploitation requires an image-capable request path, a caller able to reach it under the deployment's authentication policy, and a vulnerable host Ghostscript build.

Details

ImageMediaIO is the boundary where attacker-controlled bytes enter Pillow's format autodetection: the declared MIME type does not constrain decoder selection, and the loader calls Image.open(BytesIO(data)) without a Pillow format allowlist. Pillow recognizes EPS by its file signature, and vLLM's subsequent image.load() call enters Pillow's EPS processing path. The OpenAI chat multimodal parser reaches this path through parseimage() and fetchimage() and the media connector; the data URL path extracts the MIME label and forwards the bytes to the loader (source). This is the vLLM-specific issue: untrusted image bytes can select a Pillow-supported format parser. Each downstream impact depends on the relevant parser, component version, and host configuration.

Pillow recognizes EPS by its file signature and invokes Ghostscript with -dSAFER, -sDEVICE=pnmraw, and the untrusted EPS input (Pillow EPS plugin, load path). The command execution in this report uses the already-known Ghostscript CVE-2024-29510, which allows crafted PostScript to bypass -dSAFER; Ghostscript lists 10.03.1 as the upstream fixed version (Ghostscript CVE details). This is an N-day dependency chain: CVE-2024-29510 provides the command-execution primitive, while the vLLM-specific issue is that untrusted, remotely supplied image bytes can reach the affected interpreter. No new Ghostscript flaw is claimed here.

The vLLM Python requirements declare Pillow but not Ghostscript, so this execution chain depends on a Ghostscript executable supplied by the host or deployment image. I reproduced command execution with the PyPI vllm==0.30.0 wheel, Pillow 12.3.0, and Debian amd64 Ghostscript 10.03.0~dfsg-1. The Codean Labs CVE-2024-29510 testkit identified that Ghostscript build as vulnerable. The adapted EPS payload ran only touch on a temporary marker. The marker existed after the loader call, although Pillow then raised ValueError because the exploit changes Ghostscript's output device. The API route was traced in source but a model-backed HTTP request was not run. Only PyPI vllm==0.30.0 was checked; no wider affected release range or other Ghostscript vendor build is claimed.

The same input boundary also exposes a separate Pillow-only denial-of-service condition when the installed Pillow version is 12.0.0–12.2.0. CVE-2026-59203 is triggered by a negative %%BeginBinary byte count and causes Pillow's EPS parser to loop during Image.open(); it does not invoke Ghostscript or provide code execution (Pillow advisory). The published vllm==0.30.0 requirements leave Pillow unpinned, so an existing environment can retain an affected Pillow version; the source at commit bc21cba9673cfc2256a2726b4bc32f062237cd35 requires pillow>=12.3.0. In an isolated local check with Pillow 12.2.0, the official sample hung through ImageMediaIO.loadbytes() and MediaConnector.fetchimageasync(). Eight concurrent image loads occupied the connector's eight media-loading threads and left a benign image load queued. Pillow 12.3.0 rejected the sample promptly. This confirms a conditional media-processing DoS path, separate from the Ghostscript RCE chain; full HTTP-server availability was not measured.

Impact

When a vLLM image-processing endpoint accepts attacker-supplied content and the host has Ghostscript affected by CVE-2024-29510, a caller who can reach the endpoint can trigger command execution as the vLLM process user. API-key authentication and network access controls determine who can reach the request path. Deployments without Ghostscript or with a build fixed for CVE-2024-29510 are not affected by this execution chain. Separately, existing environments that run vLLM 0.30.0 with Pillow 12.0.0–12.2.0 are exposed to CVE-2026-59203 through the same image-format dispatch boundary; the demonstrated impact is exhaustion of media-loading workers, and this Pillow CVE does not cause RCE. The report asks vLLM to assess and address the unsafe dispatch of untrusted image bytes and the affected dependency combinations.

This advisory was fixed in MR60022.

1 / 2
Source: GitHub
First published (updated )
Severity
9.8
Buffer Overflow
AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

In Artifex Ghostscript through 10.01.0, there is a buffer overflow leading to potential corruption of data internal to the PostScript interpreter, in base/sbcp.c. This affects BCPEncode, BCPDecode, TBCPEncode, and TBCPDecode. If the write buffer is filled to one byte less than full, and one then tries to write an escaped character, two bytes are written.

1 / 2
First published (updated )
Severity
9.8
AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

In Artifex Ghostscript through 10.01.2, gdevijs.c in GhostPDL can lead to remote code execution via crafted PostScript documents because they can switch to the IJS device, or change the IjsServer parameter, after SAFER has been activated. NOTE: it is a documented risk that the IJS server can be specified on a gs command line (the IJS device inherently must execute a command to start the IJS server).

1 / 3
First published (updated )
Severity
9.8
Integer Underflow
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

A memory corruption issue was found in Artifex Ghostscript 9.50 and 9.52. Use of a non-standard PostScript operator can allow overriding of file access controls. The 'rsearch' calculation for the 'post' size resulted in a size that was too large, and could underflow to max uint32t. This was fixed in commit 5d499272b95a6b890a1397e11d20937de000d31b.

1 / 2
Source: Launchpad
First published (updated )
Severity
9.8
AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:L

A flaw was found in ghostscript, versions 9.x before 9.50, in the setsystemparams procedure where it did not properly secure its privileged calls, enabling scripts to bypass -dSAFER restrictions. A specially crafted PostScript file could disable security protection and then have access to the file system, or execute arbitrary commands.

1 / 2
First published (updated )
Severity
9.8
AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

An issue was discovered in Artifex Ghostscript before 9.26. LockSafetyParams is not checked correctly if another device is used.

1 / 3
Source: Launchpad
First published (updated )
Severity
9.8
Use After Free, Double Free
AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

An incorrect reference count was found in .setdevice. This leads to a use-after-free, triggering a Denial of Service, or possibly, arbitrary code execution.

Upstream bug : - Bug 697179 - double free with .setdevice http://bugs.ghostscript.com/showbug.cgi?id=697179

upstream fix : - Bug 697179: Reference count device icc profile when copying a device http://git.ghostscript.com/?p=user/chrisl/ghostpdl.git;a=commitdiff;h=d5ad1e02

Reference : http://seclists.org/oss-sec/2016/q4/37

1 / 2
Source: Red Hat
First published (updated )
Severity
9.8
Incorrect Type Cast
AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

.initializedscparser doesn't validate the parameter is a dict type before using it.

This is a security issue, because it can be abused to escape the -dSAFER sandbox.

This allows a Denial of Service, arbitrary code execution.

Upstream bug : - Bug 697190 - .initializedscparser doesn't validate the parameter is a dict type before using it. http://bugs.ghostscript.com/showbug.cgi?id=697190

Upstream patch : - DSC parser - validate parameters http://git.ghostscript.com/?p=ghostpdl.git;h=875a0095f37626a721c7ff57d606a0f95af03913

Reference : http://seclists.org/oss-sec/2016/q4/37

1 / 2
Source: Red Hat
First published (updated )
Severity
9.3
Buffer Overflow
AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X

Ghostscript before 10.08.0 contains a heap-based buffer overflow vulnerability in the JPEG 2000 output adapter (base/sjpxopenjpeg.c) that allows attackers to cause memory corruption by supplying a crafted PDF containing a JPEG 2000 image with mismatched component subsampling factors. When image components declare different subsampling values, the non-samescale sub-byte-depth output path allocates a row buffer sized for packed output but writes a full byte per output column regardless of bit depth, overflowing the allocation and corrupting internal chunk-allocator metadata to achieve code execution.

1 / 2
Source: MITRE
First published (updated )
Severity
9.3
Buffer Overflow
AV:N/AC:M/Au:N/C:C/I:C/A:C

Created attachment 373324 [details] ghostscript backtrace

Description of problem: When cups prints a PDF, it passes it at some point through pstoraster which calls gs. I noticed a PDF I was working on caused printing to fail, and it appears that some of the debugging output (which pstoraster turns on) from the gs cups driver can trigger a buffer overflow in the gs debug handling (which uses vsprintf - bad bad bad).

I think there are two issues here. One is that somehow gs ended up with MediaType argument without null termination (at least that's how it appears). The other is that ghostscript's print handling code uses (v)sprintf when it should be using (v)snprintf. In fact, ghostscript uses (v)sprintf with an "%s" format modifier over 50 times in the source code, several instances of which are probably unsafe (such as here).

Consider the following code from base/gsmisc.c: int errprintf(const gsmemoryt mem, const char fmt, ...) { ... char buf[PRINTFBUFLENGTH]; ... count = vsprintf(buf, fmt, args); errwrite(mem, buf, count); if (count >= PRINTFBUFLENGTH) { count = sprintf(buf, "PANIC: printf exceeded %d bytes. Stack has been corrupted.\n", PRINTFBUFLENGTH); errwrite(mem, buf, count); } ...

Just marvellous -- ghostscript expects vsprintf to return sanely if it overflows (which it doesn't), and then expects to print a nice panic message. It would be a 5 line patch to fix this so that it uses vsnprintf, and if count+1 exceeds the buffer length then it can print a warning that the message was truncated instead. Same goes for outprintf.

I will attach the stack trace from gdb, the cups error log and the PDF which caused the problem.

I wasn't sure if this was security-sensitive or not, but I ticked the box just to be on the safe side. With all the sanity checking that gcc adds to (v)sprintf these days, it's probably very hard to exploit.

Version-Release number of selected component (if applicable): ghostscript-8.70-1.fc12.i686

How reproducible: No idea - works every time on this PDF though.

Steps to Reproduce: 1. Open PDF in evince 2. Print PDF to any printer 3. Wait for message saying "printing failed" or somesuch.

Additional info: To attach gdb, I put a 'sleep 20' command in the pstoraster shell script (/usr/lib/cups/filter/pstoraster) just before it executed gs, then I "printed" the PDF, found the sleeping bash process and attached gdb to it.

1 / 2
Source: Red Hat
First published (updated )
Severity
8.8
Code Injection
AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H

An issue was discovered in Artifex Ghostscript before 10.03.1. contrib/opvp/gdevopvp.c allows arbitrary code execution via a custom Driver library, exploitable via a crafted PostScript document. This occurs because the Driver parameter for opvp (and oprp) devices can have an arbitrary name for a dynamic library; this library is then loaded.

1 / 3
Source: NVD
First published (updated )
Severity
8.8
EPSS
1.45%
AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H

Artifex Ghostscript before 10.03.0 has a heap-based overflow when PDFPassword (e.g., for runpdf) has a \000 byte in the middle.

1 / 2
Source: MITRE
First published (updated )
Severity
8.8
EPSS
0.91%
Buffer Overflow
AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H

Artifex Ghostscript before 10.03.0 has a stack-based buffer overflow in the pdfiapplyfilter() function via a long PDF filter name.

1 / 2
Source: MITRE
First published (updated )
Severity
8.8
Input Validation
AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H

The PS Interpreter in Ghostscript 9.18 and 9.20 allows remote attackers to execute arbitrary code via crafted userparams.

First published (updated )
Severity
8.8
AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:L

A flaw was found in all versions of ghostscript 9.x before 9.50, where the .charkeys procedure, where it did not properly secure its privileged calls, enabling scripts to bypass -dSAFER restrictions. An attacker could abuse this flaw by creating a specially crafted PostScript file that could escalate privileges within the Ghostscript and access files outside of restricted areas or execute commands.

1 / 2
Source: MITRE
First published (updated )
Severity
8.6
AV:L/AC:L/PR:N/UI:R/S:C/C:H/I:H/A:H

Artifex Ghostscript 9.25 and earlier allows attackers to bypass a sandbox protection mechanism via vectors involving the 1Policy operator.

1 / 3
Source: Launchpad
First published (updated )
Severity
8.6
AV:L/AC:L/PR:N/UI:R/S:C/C:H/I:H/A:H

Artifex Ghostscript 9.25 and earlier allows attackers to bypass a sandbox protection mechanism via vectors involving errorhandler setup. NOTE: this issue exists because of an incomplete fix for CVE-2018-17183.

1 / 3
Source: Launchpad
First published (updated )
Severity
8.5
Use After Free
AV:L/AC:L/AT:N/PR:N/UI:P/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X

Ghostscript before 10.08.0 contains a -dSAFER bypass vulnerability in the .putgstringcopy procedure in gsfonts.ps, which calls the privileged .forceput operator without checking operand types. Attackers can supply crafted PostScript or PDF documents that trigger a type error, recover .forceput via a custom error handler, and cause a use-after-free across save/restore to execute arbitrary code.

First published (updated )
Severity
8.4
Buffer Overflow
AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H

An issue was discovered in pdf/pdfxref.c in Artifex Ghostscript before 10.04.0. There is a buffer overflow during handling of a PDF XRef stream (related to W array values).

1 / 3
Source: MITRE
First published (updated )
Severity
8.4
Path Traversal
AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H

An issue was discovered in decodeutf8 in base/gputf8.c in Artifex Ghostscript before 10.04.0. Overlong UTF-8 encoding leads to possible ../ directory traversal.

1 / 3
Source: MITRE
First published (updated )
Severity
8.4
OS Command Injection
AV:L/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

A vulnerability was found in Ghostscript. This flaw occurs due to a mishandled permission validation for pipe devices (with the %pipe% prefix or the | pipe character prefix).

1 / 5
First published (updated )
Severity
7.8
AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H

An issue was discovered in psi/zcolor.c in Artifex Ghostscript before 10.04.0. An unchecked Implementation pointer in Pattern color space could lead to arbitrary code execution.

1 / 3
Source: MITRE
First published (updated )
Severity
7.8
Integer Overflow, Path Traversal
AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H

An issue was discovered in base/gsdevice.c in Artifex Ghostscript before 10.04.0. An integer overflow when parsing the filename format string (for the output filename) results in path truncation, and possible path traversal and code execution.

1 / 3
Source: MITRE
First published (updated )
Severity
7.8
AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H

An issue was discovered in psi/zfile.c in Artifex Ghostscript before 10.04.0. Out-of-bounds data access in filenameforall can lead to arbitrary code execution.

1 / 3
Source: MITRE
First published (updated )
Severity
7.8
EPSS
0.01%
Buffer Overflow
AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H

An issue was discovered in Artifex Ghostscript before 10.05.0. A buffer overflow occurs via an oversized Type 4 function in a PDF document to pdf/pdffunc.c.

1 / 2
Source: NVD
First published (updated )

Contact

SecAlerts Pty Ltd.
132 Wickham Terrace
Fortitude Valley,
QLD 4006, Australia
info@secalerts.co
By using SecAlerts services, you agree to our services end-user license agreement. This website is safeguarded by reCAPTCHA and governed by the Google Privacy Policy and Terms of Service. All names, logos, and brands of products are owned by their respective owners, and any usage of these names, logos, and brands for identification purposes only does not imply endorsement. If you possess any content that requires removal, please get in touch with us.
© 2026 SecAlerts Pty Ltd.
ABN: 70 645 966 203, ACN: 645 966 203