Service

PHP Upgrades & Emergency Patching

Short-turnaround PHP version upgrades for applications stuck on an unsupported branch, plus emergency patching when a critical fix lands and you need to be on it this week rather than next quarter.

Detail

Compatibility assessment, a staged upgrade path, and the deprecation and breaking-change work needed to move a live PHP application onto a supported, actively patched branch, without a rewrite and without an open-ended project.

What we cover

Exposure check

Which PHP version each of your hosts actually runs, and whether the current critical fixes have reached it.

Compatibility audit

A pass over your codebase and Composer dependencies for the deprecations and breaking changes between your version and a supported one.

Upgrade path

A staged plan setting out which branch to target, in what order, and what has to change at each step.

Emergency patching

Where upgrading cannot happen immediately, the mitigations and configuration changes that reduce exposure in the meantime.

Dependency work

Abandoned or unmaintained packages that block the upgrade, with replacements or a plan to vendor them.

Verification

A re-check after the work, confirming the running version is patched and the mitigations are actually in place.

On 21 October 2026 the PHP project shipped emergency releases for two critical vulnerabilities, CVE-2026-109618 and CVE-2026-109619, collectively nicknamed php2shell. Both are unauthenticated remote code execution, both carry a CVSS v4.0 score of 9.2, and within days CISA had added them to its Known Exploited Vulnerabilities catalogue on the strength of exploitation observed in the wild.

The fixes landed in PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35.

The important part of that list is what is missing from it. The oldest branch to receive a fix was 8.2. If your application runs on PHP 8.1 or anything earlier, no official patch was published for you, and none is coming. For those versions, upgrading is the remediation.

Why this is an upgrade problem, not a patching problem

Plenty of working PHP applications are still on 8.1, 7.4, or older. They were stable, they were doing their job, and upgrading was never urgent enough to win an argument against shipping features. That calculation changed the moment an actively exploited, unauthenticated RCE landed with no backport for the branch they sit on.

The usual blocker is not the language version. It is the dependency that has not been touched since 2019, the extension that was never ported, and the handful of behaviours the application quietly relies on that changed between major versions. That work is finite and knowable, but somebody has to go and find it.

Short turnaround

This service exists because "we should really upgrade PHP" tends to become a project with no end date. We scope it as a short, bounded engagement instead: establish what you are actually running, identify precisely what breaks between here and a supported branch, and give you a staged path with the risky parts called out.

Where an immediate upgrade genuinely is not possible, perhaps a vendor application you cannot modify or a release window weeks away, we document the interim mitigations that reduce exposure, and we are explicit that they are a holding position rather than a fix.

New features on an old runtime

There is a pattern worth naming, because we see it repeatedly. An organisation adds something genuinely new to an application, an AI assistant, a chatbot, an agent wired into internal data, and ships it on top of a codebase running a PHP version that stopped receiving security patches some time ago.

The new feature gets the scrutiny. The runtime underneath it does not, because it has been working for years. The result is a brand new, internet-facing, often unauthenticated entry point sitting on an interpreter with no vendor fixes available. The feature did not introduce the vulnerability, but it did widen the door to it considerably.

If that describes your situation, the upgrade and the AI and LLM application security review are worth scoping together. Fixing the runtime without reviewing the new surface, or the reverse, leaves half the problem in place.

A note on the follow-up advisory

Two days after the php2shell releases, PHP disclosed a further RCE issue, CVE-2026-115560, in the interaction between session serialisation and the session upload progress feature. It is worth knowing about for two reasons. First, the php2shell releases do not remediate it. Second, at the time of writing no patch has shipped. The fix is queued for a routine release, and the mitigation is a single INI directive, session.upload_progress.enabled=0.

PHP's own assessment is that exploitation requires an unusual session configuration and is unlikely in practice. We check for it anyway, because "unusual" is not the same as "absent", and a one-line INI change is a cheap thing to be certain about.

What you get

01

A written exposure summary covering every host and the PHP version it runs

02

A staged upgrade plan with the breaking changes identified per step

03

Interim mitigations where an immediate upgrade is not possible

04

A verification pass confirming the patched version is live

Every engagement starts with a conversation about scope. Tell us what you have built and what worries you about it, and we will tell you what a useful assessment looks like.