From 39d368612ac737d06921b88d4ce4236fc0da32bf Mon Sep 17 00:00:00 2001
From: Mikael Mattsson <62009528+mikael-mattsson@users.noreply.github.com>
Date: Tue, 20 Oct 2026 11:56:23 +0200
Subject: [PATCH] Improve GHSA-rj75-hqrm-r3gf

---
 .../GHSA-rj75-hqrm-r3gf.json                  | 37 +++++++++++++++++--
 1 file changed, 34 insertions(+), 3 deletions(-)

diff --git a/advisories/github-reviewed/2026/10/GHSA-rj75-hqrm-r3gf/GHSA-rj75-hqrm-r3gf.json b/advisories/github-reviewed/2026/10/GHSA-rj75-hqrm-r3gf/GHSA-rj75-hqrm-r3gf.json
index 97171b374a2f..8eff9949095a 100644
--- a/advisories/github-reviewed/2026/10/GHSA-rj75-hqrm-r3gf/GHSA-rj75-hqrm-r3gf.json
+++ b/advisories/github-reviewed/2026/10/GHSA-rj75-hqrm-r3gf/GHSA-rj75-hqrm-r3gf.json
@@ -1,13 +1,13 @@
 {
   "schema_version": "1.4.0",
   "id": "GHSA-rj75-hqrm-r3gf",
-  "modified": "2026-10-05T22:53:54Z",
+  "modified": "2026-10-05T22:53:57Z",
   "published": "2026-10-05T22:53:54Z",
   "aliases": [
     "CVE-2026-104844"
   ],
   "summary": "PostCSS: Quadratic complexity in flat selector parsing allows CPU exhaustion",
-  "details": "### Impact\n\n`.` and `#` are not word delimiters in the tokenizer, so a flat selector such as\n`.a.a.a...` reaches `splitWord()` as a single word token carrying n class or id\nindexes. Three passes scanned those index arrays linearly for every index,\nmaking the parse O(n^2) in the number of indexes rather than in input length:\n`uniqs()`, the `indices.forEach` loop, and the Sass-interpolation filter.\nParsing a 400 KB flat selector took ~34 s on a modern laptop, fully occupying a\nsingle thread. A benign selector of identical byte size parses in tens of\nmilliseconds, so the cost is driven by the index count, not the input size.\nThe nesting depth of such a selector is 0, so the `maxNestingDepth` guard added\nin 7.1.3 offers no protection.\n\nReachability is deployment dependent. Only consumers that parse untrusted,\nattacker-supplied selectors synchronously in a request path are exposed, for\nexample CSS sanitizers, CSS-in-JS services and online playgrounds. Ordinary\nbuild-time use on trusted sources is not affected.\n\n### Patches\n\nFixed in 7.1.6. The three passes now use Set membership tests, making parsing\nlinear in the number of indexes. There is no behaviour change: parsing is\nbyte-identical on a differential corpus of 8413 selectors.\n\n### Workarounds\n\nCap the size of selectors accepted from untrusted sources before parsing.",
+  "details": "### Impact\n\n`.` and `#` are not word delimiters in the tokenizer, so a flat selector such as\n`.a.a.a...` reaches `splitWord()` as a single word token carrying n class or id\nindexes. Three passes scanned those index arrays linearly for every index,\nmaking the parse O(n^2) in the number of indexes rather than in input length:\n`uniqs()`, the `indices.forEach` loop, and the Sass-interpolation filter.\nParsing a 400 KB flat selector took ~34 s on a modern laptop, fully occupying a\nsingle thread. A benign selector of identical byte size parses in tens of\nmilliseconds, so the cost is driven by the index count, not the input size.\nThe nesting depth of such a selector is 0, so the `maxNestingDepth` guard added\nin 7.1.3 offers no protection.\n\nReachability is deployment dependent. Only consumers that parse untrusted,\nattacker-supplied selectors synchronously in a request path are exposed, for\nexample CSS sanitizers, CSS-in-JS services and online playgrounds. Ordinary\nbuild-time use on trusted sources is not affected.\n\n### Patches\n\nFixed in 7.1.6, and for the 6.x line in 6.1.5 (backport of the same change,\npostcss/postcss-selector-parser#346). The three passes now use Set membership tests, making parsing\nlinear in the number of indexes. There is no behaviour change: parsing is\nbyte-identical on a differential corpus of 8413 selectors.\n\n### Workarounds\n\nCap the size of selectors accepted from untrusted sources before parsing.",
   "severity": [
     {
       "type": "CVSS_V3",
@@ -25,7 +25,7 @@
           "type": "ECOSYSTEM",
           "events": [
             {
-              "introduced": "0"
+              "introduced": "7.0.0"
             },
             {
               "fixed": "7.1.6"
@@ -33,6 +33,25 @@
           ]
         }
       ]
+    },
+    {
+      "package": {
+        "ecosystem": "npm",
+        "name": "postcss-selector-parser"
+      },
+      "ranges": [
+        {
+          "type": "ECOSYSTEM",
+          "events": [
+            {
+              "introduced": "0"
+            },
+            {
+              "fixed": "6.1.5"
+            }
+          ]
+        }
+      ]
     }
   ],
   "references": [
@@ -44,6 +63,10 @@
       "type": "ADVISORY",
       "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-104844"
     },
+    {
+      "type": "WEB",
+      "url": "https://github.com/postcss/postcss-selector-parser/pull/346"
+    },
     {
       "type": "WEB",
       "url": "https://github.com/postcss/postcss-selector-parser/commit/62b191792df0a0bc56062e5a875bc74aae2a51cd"
@@ -52,9 +75,17 @@
       "type": "PACKAGE",
       "url": "https://github.com/postcss/postcss-selector-parser"
     },
+    {
+      "type": "WEB",
+      "url": "https://github.com/postcss/postcss-selector-parser/releases/tag/6.1.5"
+    },
     {
       "type": "WEB",
       "url": "https://github.com/postcss/postcss-selector-parser/releases/tag/7.1.6"
+    },
+    {
+      "type": "WEB",
+      "url": "https://www.npmjs.com/package/postcss-selector-parser/v/6.1.5"
     }
   ],
   "database_specific": {