[dependabot]

Bumps github.com/jackc/pgx/v5 from 5.9.0 to 5.9.2.

Changelog

Sourced from github.com/jackc/pgx/v5's changelog.

5.9.2 (April 18, 2026)

Fix SQL Injection via placeholder confusion with dollar quoted string literals (GHSA-j88v-2chj-qfwx)

SQL injection can occur when:

1. The non-default simple protocol is used.
2. A dollar quoted string literal is used in the SQL query.
3. That query contains text that would be would be interpreted outside as a placeholder outside of a string literal.
4. The value of that placeholder is controllable by the attacker.

e.g.

attackValue := `$tag$; drop table canary; --`
_, err = tx.Exec(ctx, `select $tag$ $1 $tag$, $1`, pgx.QueryExecModeSimpleProtocol, attackValue)

This is unlikely to occur outside of a contrived scenario.

5.9.1 (March 22, 2026)

• Fix: batch result format corruption when using cached prepared statements (reported by Dirkjan Bussink)

Commits

• 0aeabbc Release v5.9.2
• 60644f8 Fix SQL sanitizer bugs with dollar-quoted strings and placeholder overflow
• a5680bc Merge pull request #2531 from dolmen-go/godoc-add-links
• e34e452 doc: Add godoc links
• 08c9bb1 Fix Stringer types encoded as text instead of numeric value in composite fields
• 96b4dbd Remove unstable test
• acf88e0 Merge pull request #2526 from abrightwell/abrightwell-min-proto
• 2f81f1f Update max_protocol_version and min_protocol_version defaults
• 4e4eaed Release v5.9.1
• 6273188 Fix batch result format corruption when using cached prepared statements
• Additional commits viewable in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

---

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

• @dependabot rebase will rebase this PR
• @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
• @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
• @dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)
  You can disable automated security fix PRs for this repo from the Security Alerts page.

[steadytao]

Non-breaking change. Patches GHSA-j88v-2chj-qfwx SQL Injection Vulnerability + misc fixes.

[vrothberg]

@steadytao any chance you could cut a new patch release of caddy with this fix? https://nvd.nist.gov/vuln/detail/CVE-2026-33815 (and CVE-2026-33816) is showing up in security scanners at the moment.

[mohammed90]

@steadytao any chance you could cut a new patch release of caddy with this fix? https://nvd.nist.gov/vuln/detail/CVE-2026-33815 (and CVE-2026-33816) is showing up in security scanners at the moment.

Which scanner is this? Can you report to the scanner how their scanner is noisy? The reported vuln isn't in Caddy execution path.

[vrothberg]

@steadytao any chance you could cut a new patch release of caddy with this fix? https://nvd.nist.gov/vuln/detail/CVE-2026-33815 (and CVE-2026-33816) is showing up in security scanners at the moment.

Which scanner is this? Can you report to the scanner how their scanner is noisy? The reported vuln isn't in Caddy execution path.

I think pretty much all scanners will report that. I understand how frustrating such CVEs are but it's an uphill battle. Cutting patch releases is less work going forward than explaining why N CVEs aren't exploitable.

[steadytao]

I understand the scanner pressure and the dependency bump is already on master so this should clear with the next normal Caddy release.

However, I do not think this should drive a dedicated patch release by itself though. The pgx advisory is not in Caddy's execution path so releasing specifically for it would imply a Caddy security impact that does not exist. Bump was merged due to it being a non-breaking bump with misc changes and for the exact scenario where scanners will falsely report the CVEs simply because the dep exists. If someone can show a reachable Caddy path for CVE-2026-33815 or CVE-2026-33816, that would change the situation.

[vrothberg]

I under your point of view. I've been a Podman maintainer for many years and was quite annoyed by CVEs getting filed against it where the code path wasn't even vendored.

What I am currently working on is Project Hummingbird providing 0 CVE images to users including caddy. The pressure of CVE management is getting a real burden to users, so we are shifting the responsibility as far left as possible. That will protect each user from trying to figure out whether a CVE is exploitable or not.

I've been working in this space for a while now and I think the path of least resistance and work is keep dependencies up to date and cutting releases as often as possible.

I will go ahead and close the CVE for Hummingbird.

[steadytao]

I also understand your point of view and appreciate you closing this on Hummingbirds end. I think that is fair from the downstream image-maintainer side. If users are blocked by scanner gates, a patched release is often the cleanest operational answer.

My concern comes more so from the release signal on Caddy’s end. If there is no known reachable Caddy path, I would rather include the bump in the next normal release than cut a dedicated patch release that would imply Caddy itself held the security impact. I would honestly prefer to not let scanner noise influence our release structure. Unfortunately that is just the ecosystem surrounding CVEs with scanners being noisy and practically anyone being allowed to create them, it causes quite a lot of headache in most cases.