GHSA-6R34-94WQ-JHRC
Vulnerability from github – Published: 2026-04-06 17:53 – Updated: 2026-05-13 16:20Summary
A signed length truncation bug causes an out-of-bounds read in the default Markdown parse path. Inputs larger than INT_MAX are truncated to a signed int before entering the native parser, allowing the parser to read past the end of the supplied buffer and crash the process
Details
In both public entry points:
ext/rdiscount.c:97ext/rdiscount.c:136
RSTRING_LEN(text) is passed directly into mkd_string():
MMIOT *doc = mkd_string(RSTRING_PTR(text), RSTRING_LEN(text), flags);
mkd_string() accepts int len:
ext/mkdio.c:174
Document * mkd_string(const char *buf, int len, mkd_flag_t flags)
{
struct string_stream about;
about.data = buf;
about.size = len;
return populate((getc_func)__mkd_io_strget, &about, flags & INPUT_MASK);
}
The parser stores the remaining input length in a signed int:
ext/markdown.h:205
struct string_stream {
const char *data;
int size;
};
The read loop stops only when size == 0:
ext/mkdio.c:161
int __mkd_io_strget(struct string_stream *in)
{
if ( !in->size ) return EOF;
--(in->size);
return *(in->data)++;
}
If the Ruby string length exceeds INT_MAX, the value can truncate to a negative int. In that state, the parser continues incrementing data and reading past the end of the original Ruby string, causing an out-of-bounds read and native crash.
Affected APIs:
RDiscount.new(input).to_htmlRDiscount.new(input).toc_content
PoC
Crash via to_html:
RUBYLIB=lib:ext ruby -e 'require "rdiscount"; n=2_200_000_000; s = "a" * n; warn "built=#{s.bytesize}"; RDiscount.new(s).to_html"'
result:
built=2200000000- Ruby terminates with
[BUG] Segmentation fault - top control frame:
CFUNC :to_html
same result with toc_content
Impact
This is an out-of-bounds read with the main issue being reliable denial-of-service. Impacted is limited to deployments parses attacker-controlled Markdown and permits multi-GB inputs.
Fix
just add a checked length guard before the mkd_string() call in both public entry points:
ext/rdiscount.c:97ext/rdiscount.c:136ex:
VALUE text = rb_funcall(self, rb_intern("text"), 0);
long text_len = RSTRING_LEN(text);
VALUE buf = rb_str_buf_new(1024);
Check_Type(text, T_STRING);
if (text_len > INT_MAX) {
rb_raise(rb_eArgError, "markdown input too large");
}
MMIOT *doc = mkd_string(RSTRING_PTR(text), (int)text_len, flags);
The same guard should be applied in rb_rdiscount_toc_content() before its mkd_string() call.
{
"affected": [
{
"package": {
"ecosystem": "RubyGems",
"name": "rdiscount"
},
"ranges": [
{
"events": [
{
"introduced": "1.3.1.1"
},
{
"fixed": "2.2.7.4"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-35201"
],
"database_specific": {
"cwe_ids": [
"CWE-125"
],
"github_reviewed": true,
"github_reviewed_at": "2026-04-06T17:53:59Z",
"nvd_published_at": "2026-04-06T20:16:27Z",
"severity": "MODERATE"
},
"details": "### Summary\n\nA signed length truncation bug causes an out-of-bounds read in the default Markdown parse path. Inputs larger than `INT_MAX` are truncated to a signed `int` before entering the native parser, allowing the parser to read past the end of the supplied buffer and crash the process\n\n### Details\n\nIn both public entry points:\n\n- `ext/rdiscount.c:97`\n- `ext/rdiscount.c:136`\n\n`RSTRING_LEN(text)` is passed directly into `mkd_string()`:\n\n```c\nMMIOT *doc = mkd_string(RSTRING_PTR(text), RSTRING_LEN(text), flags);\n```\n\n`mkd_string()` accepts `int len`:\n\n- `ext/mkdio.c:174`\n\n```c\nDocument * mkd_string(const char *buf, int len, mkd_flag_t flags)\n{\n struct string_stream about;\n\n about.data = buf;\n about.size = len;\n\n return populate((getc_func)__mkd_io_strget, \u0026about, flags \u0026 INPUT_MASK);\n}\n```\n\nThe parser stores the remaining input length in a signed `int`:\n\n- `ext/markdown.h:205`\n\n```c\nstruct string_stream {\n const char *data;\n int size;\n};\n```\n\nThe read loop stops only when `size == 0`:\n\n- `ext/mkdio.c:161`\n\n```c\nint __mkd_io_strget(struct string_stream *in)\n{\n if ( !in-\u003esize ) return EOF;\n\n --(in-\u003esize);\n\n return *(in-\u003edata)++;\n}\n```\n\nIf the Ruby string length exceeds `INT_MAX`, the value can truncate to a negative `int`. In that state, the parser continues incrementing `data` and reading past the end of the original Ruby string, causing an out-of-bounds read and native crash.\n\nAffected APIs:\n\n- `RDiscount.new(input).to_html`\n- `RDiscount.new(input).toc_content`\n\n### PoC\n\nCrash via `to_html`:\n\n```sh\nRUBYLIB=lib:ext ruby -e \u0027require \"rdiscount\"; n=2_200_000_000; s = \"a\" * n; warn \"built=#{s.bytesize}\"; RDiscount.new(s).to_html\"\u0027\n```\nresult:\n\n- `built=2200000000`\n- Ruby terminates with `[BUG] Segmentation fault`\n- top control frame: `CFUNC :to_html`\n\nsame result with `toc_content`\n\n### Impact\n\nThis is an out-of-bounds read with the main issue being reliable denial-of-service. Impacted is limited to deployments parses attacker-controlled Markdown and permits multi-GB inputs.\n\n### Fix\n\njust add a checked length guard before the `mkd_string()` call in both public entry points:\n\n- `ext/rdiscount.c:97`\n- `ext/rdiscount.c:136`\nex: \n```c\nVALUE text = rb_funcall(self, rb_intern(\"text\"), 0);\nlong text_len = RSTRING_LEN(text);\nVALUE buf = rb_str_buf_new(1024);\nCheck_Type(text, T_STRING);\n\nif (text_len \u003e INT_MAX) {\n rb_raise(rb_eArgError, \"markdown input too large\");\n}\n\nMMIOT *doc = mkd_string(RSTRING_PTR(text), (int)text_len, flags);\n```\n\nThe same guard should be applied in `rb_rdiscount_toc_content()` before its `mkd_string()` call.",
"id": "GHSA-6r34-94wq-jhrc",
"modified": "2026-05-13T16:20:38Z",
"published": "2026-04-06T17:53:59Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/davidfstr/rdiscount/security/advisories/GHSA-6r34-94wq-jhrc"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-35201"
},
{
"type": "WEB",
"url": "https://github.com/davidfstr/rdiscount/commit/b1a16445e92e0d12c07594dedcdc56f80b317761"
},
{
"type": "PACKAGE",
"url": "https://github.com/davidfstr/rdiscount"
},
{
"type": "WEB",
"url": "https://github.com/rubysec/ruby-advisory-db/blob/master/gems/rdiscount/CVE-2026-35201.yml"
},
{
"type": "WEB",
"url": "http://github.com/davidfstr/rdiscount/releases/tag/2.2.7.4"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:N/A:H",
"type": "CVSS_V3"
}
],
"summary": "rdiscount has an Out-of-bounds Read"
}
Drag nodes to arrange them. Double-click a related vulnerability to open it, or a source to visit its link. The Physics button on the left tunes the layout and the simulation; the legend at the bottom left hides or shows nodes by kind and by sighting type.
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
Vulnerability timeline
Sightings and recorded dates, grouped by day · UTC
How durations are calculated
Following the VTEM framework (Black Hat USA 2015), time to patch starts when the vendor or a coordinator was informed, total time to patch at discovery, and disclosure to patch at public disclosure; vendor response time runs from the report to its acknowledgement. Time to exploit and time unpatched start at the earliest discovery or report, falling back to public disclosure. Unpatched time ends at the first recorded patch, or the calculation time when no patch is recorded. Exploitation to patch shows the interval between those two events; a published proof of concept is a public exploit, not exploitation.
Each result shows its source dates. A negative interval means the end event came first; when date-only inputs leave the order open, the result says so. Date-only inputs are approximate. The earliest evidence wins: KEV report dates and sightings are labelled as such. Records describing another vulnerability are listed but never calculated from, and a PoC alone does not establish exploitation. Missing patch evidence does not prove that no patch exists. Filters below change the event list only.
Loading timeline…
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.
Browse all ATT&CK techniques and the vulnerabilities related to each.
Related by attack behaviour
Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.