Expand only known placeholders in TLS regexp patterns, matching HTTP regexp behavior. Test numeric quantifiers and selection of the client-certificate-verifying fallback policy.
Addresses scan report b6c9c0a1-79a7-4254-b057-b1d8fc96b412.
Code written by GitHub Copilot (GPT-6.1 Sol, Claude Opus 5.5); reviewed by mholt.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Policy subjects and handshake names were compared without the IDNA
Lookup normalization CertMagic applies to SNI, so a Unicode or mapped
SNI could select the fallback issuer. Normalize both the same way,
preserving wildcard labels and non-DNS subjects.
Addresses scan report e804f22226c85b87.
Code written by GitHub Copilot (GPT-6.1 Sol, Claude Opus 5.5); reviewed by mholt.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
An empty on_demand object with no permission module caused a nil
dereference during the handshake. Treat it like absent permission
settings.
Addresses scan report 01e4330f22e4a986.
Code written by GitHub Copilot (GPT-6.1 Sol, Claude Opus 5.5); reviewed by mholt.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
For wildcard or default policies with explicit certificate managers but
no on-demand permission module, a manager miss fell through to issuance
for any client-chosen SNI. Fail closed after managers are consulted.
Addresses scan report a6eb136fbbaa01c0.
Code written by GitHub Copilot (GPT-6.1 Sol, Claude Opus 5.5); reviewed by mholt.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
The http CA pool's tls.server_name was never applied, so the endpoint
was verified against the URL host instead. Use the configured name,
including placeholders.
Addresses scan report 8b7d4c9c392321ab.
Code written by GitHub Copilot (GPT-6.1 Sol, Claude Opus 5.5); reviewed by mholt.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* caddytls: require a trust pool for client cert verification modes
Reject verify_if_given and require_and_verify without a trust pool so they stop falling back to system roots.
* Apply suggestion from @francislavoie
Signed-off-by: Francis Lavoie <lavofr@gmail.com>
---------
Signed-off-by: Francis Lavoie <lavofr@gmail.com>
Co-authored-by: Francis Lavoie <lavofr@gmail.com>
* caddytls: keep ACME issuer network proxy across re-provisioning
LoadModule clears NetworkProxyRaw, so when the HTTP app provisions an
ACME issuer a second time (fillInACMEIssuer), the rebuilt template has
no HTTPProxy. Since #7510 that template is used via SetConfig, and ACME
requests bypass the configured proxy.
Keep the loaded proxy func on the issuer, like the DNS solver.
Fixes#8283
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
(cherry picked from commit 2366825f2380383dacd8adeb1508e34c3e25634d)
* caddytls: test that ACME requests use the proxy after RebuildCertMagic
Replace the provision-twice check with a regression for the path the
HTTP app takes: Provision, re-provision in fillInACMEIssuer, then
RebuildCertMagic. The test issues against a local proxy and asserts the
ACME request goes through it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: k4lleanka <257361691+k4lleanka@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
CertMagic assembles an event's data before emit() can look at OnEvent,
because Go evaluates arguments first, so tls_get_certificate costs a map
and a ClientHello copy on every handshake even when nothing will read
them. caddyserver/certmagic#406 added Config.ShouldEmitFunc so an
embedder can answer that question first.
Export App.ShouldEmit and pass it along with OnEvent. The predicate has
to account for every observer, not just subscribed handlers: Emit writes
a debug line for each event, so an event is observable whenever debug
logging is on. Emit now asks the same method it exports, so the two
cannot drift.
Benchmarks are included; measurements are in the pull request.
* caddytls: synchronize storage cleaner with TLS.Stop via context and WaitGroup
* Unify the storage cleanup synchronization to both tls and ech.
* Cannot embed the sync.WaitGroup directly as the TLS struct is copied.
* caddytls: propagate cancellable context to ECH rotation and add sync tests
Pass cancellable context to ECH key rotation so in-flight storage locks
and operations unblock when TLS.Stop is invoked. Add comprehensive tests
verifying TLS.Stop cleanly unblocks and waits for storage cleaner and
ECH workers.
---------
Co-authored-by: Zen Dodd <mail@steadytao.com>
SNI is always ASCII on the wire (RFC 6066), and most config
patterns are also ASCII. For pure ASCII input, idna.ToASCII
only validates and lowercases, which is equivalent to a simple
strings.ToLower. Add a fast path to avoid the overhead of
idna.ToASCII in the common case.
* caddytls: fix data race in session ticket key rotation
stayUpdated copies the map header (configs := s.configs) under the
lock, then iterates the original map after releasing it. Concurrent
calls to register/unregister mutate the same map.
Hold the lock for the entire iteration instead.
* caddytls: fix data race in AllMatchingCertificates
AllMatchingCertificates reads the package-level certCache without
acquiring certCacheMu, while Cleanup sets certCache to nil under
the write lock. The adjacent HasCertificateForSubject correctly
acquires certCacheMu.RLock.
Add the missing RLock/RUnlock to match.
* caddytls: fix ECH key rotation stopping permanently on error
When rotateECHKeys returns an error, the rotation goroutine returns
immediately, stopping all future key rotation for the lifetime of
the process.
Change return to continue, matching the error handling for
publishECHConfigs two lines below.
* caddytls: Fix wildcard race in auto-HTTPS launch
When evaluating whether to skip managing an individual subdomain
due to an existing wildcard configuration, we now explicitly consult
the automate loader.
Because Caddy apps can start in any order, relying strictly on the
TLS app's internal management state was non-deterministic if the
HTTP app started first. Checking the automate loader guarantees
predictable behavior since it is fully populated during the
Provision phase, well before any apps are started.
* respond to review comments
1. update requested comment
2. remove personal domain from test
3. add regression test
* remove unnecessary mutex lock
* refactor: -integration test, +explicit cases
* refactor: remove redundant test, add comment
* rename file and add header
* update copyright year
* tls: add alpn to managed HTTPS records
* tls: centralise HTTPS RR ALPN defaults and registration
Reuse shared protocol defaults instead of repeating the default HTTP protocol list, unify server name registration to carry ALPN in one experimental API and reuse the TLS default ALPN ordering for HTTPS RR publication
* http: centralise effective protocol resolution for HTTPS RR ALPN
* caddytls: Expand ACME credentials
This allows using global placeholders such as {file./run/secrets/key_id}
when setting up the tls configuration.
* chore(formatting): gofmt on acmeissuer_test
* feat: add system and combined CA pool modules
* fix: combining pools using `CertificateProvider`
* fix: lint issue
* chore: caddyfiletests
* doing it for first time, so not sure if its right.
* fix: use `x509` native addCert
* chore: explicit err handling
* Apply suggestion from @mohammed90
---------
Co-authored-by: Mohammed Al Sahaf <mohammed@caffeinatedwonders.com>
Two error returns in ClientAuthentication.provision() were
returning nil instead of the actual error, silently swallowing
failures when converting PEM files to DER and when provisioning
the CA pool. This could cause mTLS client authentication to
silently fall back to the system trust store, accepting any
client certificate signed by a public CA instead of restricting
to the configured trust anchors.
* caddypki: Add support for multiple intermediates in signing chain
* Move intermediate lifetime configuration check
In #7272 a check was changed to ensure that generated intermediate
certificates would always use a lifetime that falls within the
lifetime of the root. However, when a root and intermediate(s)
are supplied, the configuration value was being used instead of
the actual lifetimes of the certificates. The check was moved to
only be performed when an intermediate is generated; not when
loaded from disk.
* Add tests for `pemDecodeCertificateChain` and `pemDecodeCertificate`
* Use `crypto.Signer` instead of `any` in appropriate places
* Use latest Smallstep packages
---------
Co-authored-by: Matt Holt <mholt@users.noreply.github.com>
* caddytls: correct documentation of `LeafFolderLoader`
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* fmt...
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
---------
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* client_auth: wire up leaf verifier Caddyfile
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* review feedback + tests
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
---------
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* chore: upgrade .golangci.yml and workflow to v2
run `golangci-lint fmt`
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* run `golangci-lint run --fix`
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* more lint fixes
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* bring back comments to .golangci.yml
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* appease the linter some more
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* oops
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* use embedded structs
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* use embedded structs where they were used before
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* disable rule `-QF1006`
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* missed a spot
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
---------
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
Matthew HoltandCopilot
Faiyaz RahmanandFrancis Lavoie
Akinyemi Toluwalase
awss
videah
Simon Sudarushkin
Y.Horie
Jens-Uwe MagerandZen Dodd
gelsomino
cbro
Rijul
Felix Eckhofer
Harsh PatelandMohammed Al Sahaf
Pieter Berkel
Pavel SiomachkinandFrancis Lavoie
moscowchill
Paul B
okrc
Herman SlatmanandMatt Holt
Mohammed Al Sahaf
avery
mountdisk
tongjicoder