0
Fork 0
mirror of https://github.com/caddyserver/caddy.git synced 2024-12-23 22:27:38 -05:00
caddy/caddytls/config.go

512 lines
16 KiB
Go
Raw Normal View History

// Copyright 2015 Light Code Labs, LLC
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
package caddytls
import (
"crypto/tls"
"crypto/x509"
"fmt"
"io/ioutil"
"github.com/xenolf/lego/challenge/tlsalpn01"
"github.com/klauspost/cpuid"
"github.com/mholt/caddy"
"github.com/mholt/certmagic"
"github.com/xenolf/lego/certcrypto"
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
)
// Config describes how TLS should be configured and used.
type Config struct {
// The hostname or class of hostnames this config is
// designated for; can contain wildcard characters
// according to RFC 6125 §6.4.3 - this field MUST
// be set in order for things to work as expected
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
Hostname string
// Whether TLS is enabled
Enabled bool
// Minimum and maximum protocol versions to allow
ProtocolMinVersion uint16
ProtocolMaxVersion uint16
// The list of cipher suites; first should be
// TLS_FALLBACK_SCSV to prevent degrade attacks
Ciphers []uint16
// Whether to prefer server cipher suites
PreferServerCipherSuites bool
// The list of preferred curves
CurvePreferences []tls.CurveID
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
// Client authentication policy
ClientAuth tls.ClientAuthType
// List of client CA certificates to allow, if
// client authentication is enabled
ClientCerts []string
// Manual means user provides own certs and keys
Manual bool
// Managed means this config should be managed
// by the CertMagic Config (Manager field)
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
Managed bool
// Manager is how certificates are managed
Manager *certmagic.Config
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
// SelfSigned means that this hostname is
// served with a self-signed certificate
// that we generated in memory for convenience
SelfSigned bool
// The email address to use when creating or
// using an ACME account (fun fact: if this
// is set to "off" then this config will not
// qualify for managed TLS)
ACMEEmail string
// The list of protocols to choose from for Application Layer
// Protocol Negotiation (ALPN).
ALPN []string
// The final tls.Config created with
// buildStandardTLSConfig()
tlsConfig *tls.Config
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
}
tls: Restructure and improve certificate management - Expose the list of Caddy instances through caddy.Instances() - Added arbitrary storage to caddy.Instance - The cache of loaded certificates is no longer global; now scoped per-instance, meaning upon reload (like SIGUSR1) the old cert cache will be discarded entirely, whereas before, aggressively reloading config that added and removed lots of sites would cause unnecessary build-up in the cache over time. - Key certificates in the cache by their SHA-256 hash instead of by their names. This means certificates will not be duplicated in memory (within each instance), making Caddy much more memory-efficient for large-scale deployments with thousands of sites sharing certs. - Perform name-to-certificate lookups scoped per caddytls.Config instead of a single global lookup. This prevents certificates from stepping on each other when they overlap in their names. - Do not allow TLS configurations keyed by the same hostname to be different; this now throws an error. - Updated relevant tests, with a stark awareness that more tests are needed. - Change the NewContext function signature to include an *Instance. - Strongly recommend (basically require) use of caddytls.NewConfig() to create a new *caddytls.Config, to ensure pointers to the instance certificate cache are initialized properly. - Update the TLS-SNI challenge solver (even though TLS-SNI is disabled currently on the CA side). Store temporary challenge cert in instance cache, but do so directly by the ACME challenge name, not the hash. Modified the getCertificate function to check the cache directly for a name match if one isn't found otherwise. This will allow any caddytls.Config to be able to help solve a TLS-SNI challenge, with one extra side-effect that might actually be kind of interesting (and useless): clients could send a certificate's hash as the SNI and Caddy would be able to serve that certificate for the handshake. - Do not attempt to match a "default" (random) certificate when SNI is present but unrecognized; return no certificate so a TLS alert happens instead. - Store an Instance in the list of instances even while the instance is still starting up (this allows access to the cert cache for performing renewals at startup, etc). Will be removed from list again if instance startup fails. - Laid groundwork for ACMEv2 and Let's Encrypt wildcard support. Server type plugins will need to be updated slightly to accommodate minor adjustments to their API (like passing in an Instance). This commit includes the changes for the HTTP server. Certain Caddyfile configurations might error out with this change, if they configured different TLS settings for the same hostname. This change trades some complexity for other complexity, but ultimately this new complexity is more correct and robust than earlier logic. Fixes #1991 Fixes #1994 Fixes #1303
2018-02-04 02:58:27 -05:00
// NewConfig returns a new Config with a pointer to the instance's
// certificate cache. You will usually need to set Other fields on
// the returned Config for successful practical use.
func NewConfig(inst *caddy.Instance) *Config {
inst.StorageMu.RLock()
certCache, ok := inst.Storage[CertCacheInstStorageKey].(*certmagic.Cache)
tls: Restructure and improve certificate management - Expose the list of Caddy instances through caddy.Instances() - Added arbitrary storage to caddy.Instance - The cache of loaded certificates is no longer global; now scoped per-instance, meaning upon reload (like SIGUSR1) the old cert cache will be discarded entirely, whereas before, aggressively reloading config that added and removed lots of sites would cause unnecessary build-up in the cache over time. - Key certificates in the cache by their SHA-256 hash instead of by their names. This means certificates will not be duplicated in memory (within each instance), making Caddy much more memory-efficient for large-scale deployments with thousands of sites sharing certs. - Perform name-to-certificate lookups scoped per caddytls.Config instead of a single global lookup. This prevents certificates from stepping on each other when they overlap in their names. - Do not allow TLS configurations keyed by the same hostname to be different; this now throws an error. - Updated relevant tests, with a stark awareness that more tests are needed. - Change the NewContext function signature to include an *Instance. - Strongly recommend (basically require) use of caddytls.NewConfig() to create a new *caddytls.Config, to ensure pointers to the instance certificate cache are initialized properly. - Update the TLS-SNI challenge solver (even though TLS-SNI is disabled currently on the CA side). Store temporary challenge cert in instance cache, but do so directly by the ACME challenge name, not the hash. Modified the getCertificate function to check the cache directly for a name match if one isn't found otherwise. This will allow any caddytls.Config to be able to help solve a TLS-SNI challenge, with one extra side-effect that might actually be kind of interesting (and useless): clients could send a certificate's hash as the SNI and Caddy would be able to serve that certificate for the handshake. - Do not attempt to match a "default" (random) certificate when SNI is present but unrecognized; return no certificate so a TLS alert happens instead. - Store an Instance in the list of instances even while the instance is still starting up (this allows access to the cert cache for performing renewals at startup, etc). Will be removed from list again if instance startup fails. - Laid groundwork for ACMEv2 and Let's Encrypt wildcard support. Server type plugins will need to be updated slightly to accommodate minor adjustments to their API (like passing in an Instance). This commit includes the changes for the HTTP server. Certain Caddyfile configurations might error out with this change, if they configured different TLS settings for the same hostname. This change trades some complexity for other complexity, but ultimately this new complexity is more correct and robust than earlier logic. Fixes #1991 Fixes #1994 Fixes #1303
2018-02-04 02:58:27 -05:00
inst.StorageMu.RUnlock()
if !ok || certCache == nil {
certCache = certmagic.NewCache(certmagic.FileStorage{Path: caddy.AssetsPath()})
inst.OnShutdown = append(inst.OnShutdown, func() error {
certCache.Stop()
return nil
})
tls: Restructure and improve certificate management - Expose the list of Caddy instances through caddy.Instances() - Added arbitrary storage to caddy.Instance - The cache of loaded certificates is no longer global; now scoped per-instance, meaning upon reload (like SIGUSR1) the old cert cache will be discarded entirely, whereas before, aggressively reloading config that added and removed lots of sites would cause unnecessary build-up in the cache over time. - Key certificates in the cache by their SHA-256 hash instead of by their names. This means certificates will not be duplicated in memory (within each instance), making Caddy much more memory-efficient for large-scale deployments with thousands of sites sharing certs. - Perform name-to-certificate lookups scoped per caddytls.Config instead of a single global lookup. This prevents certificates from stepping on each other when they overlap in their names. - Do not allow TLS configurations keyed by the same hostname to be different; this now throws an error. - Updated relevant tests, with a stark awareness that more tests are needed. - Change the NewContext function signature to include an *Instance. - Strongly recommend (basically require) use of caddytls.NewConfig() to create a new *caddytls.Config, to ensure pointers to the instance certificate cache are initialized properly. - Update the TLS-SNI challenge solver (even though TLS-SNI is disabled currently on the CA side). Store temporary challenge cert in instance cache, but do so directly by the ACME challenge name, not the hash. Modified the getCertificate function to check the cache directly for a name match if one isn't found otherwise. This will allow any caddytls.Config to be able to help solve a TLS-SNI challenge, with one extra side-effect that might actually be kind of interesting (and useless): clients could send a certificate's hash as the SNI and Caddy would be able to serve that certificate for the handshake. - Do not attempt to match a "default" (random) certificate when SNI is present but unrecognized; return no certificate so a TLS alert happens instead. - Store an Instance in the list of instances even while the instance is still starting up (this allows access to the cert cache for performing renewals at startup, etc). Will be removed from list again if instance startup fails. - Laid groundwork for ACMEv2 and Let's Encrypt wildcard support. Server type plugins will need to be updated slightly to accommodate minor adjustments to their API (like passing in an Instance). This commit includes the changes for the HTTP server. Certain Caddyfile configurations might error out with this change, if they configured different TLS settings for the same hostname. This change trades some complexity for other complexity, but ultimately this new complexity is more correct and robust than earlier logic. Fixes #1991 Fixes #1994 Fixes #1303
2018-02-04 02:58:27 -05:00
inst.StorageMu.Lock()
inst.Storage[CertCacheInstStorageKey] = certCache
inst.StorageMu.Unlock()
}
return &Config{
Manager: certmagic.NewWithCache(certCache, certmagic.Config{}), // TODO
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
}
}
// buildStandardTLSConfig converts cfg (*caddytls.Config) to a *tls.Config
// and stores it in cfg so it can be used in servers. If TLS is disabled,
// no tls.Config is created.
2017-04-29 19:28:18 -05:00
func (c *Config) buildStandardTLSConfig() error {
if !c.Enabled {
return nil
}
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
config := new(tls.Config)
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
ciphersAdded := make(map[uint16]struct{})
curvesAdded := make(map[tls.CurveID]struct{})
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
// add cipher suites
2017-04-29 19:28:18 -05:00
for _, ciph := range c.Ciphers {
if _, ok := ciphersAdded[ciph]; !ok {
ciphersAdded[ciph] = struct{}{}
config.CipherSuites = append(config.CipherSuites, ciph)
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
}
}
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
2017-04-29 19:28:18 -05:00
config.PreferServerCipherSuites = c.PreferServerCipherSuites
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
// add curve preferences
2017-04-29 19:28:18 -05:00
for _, curv := range c.CurvePreferences {
if _, ok := curvesAdded[curv]; !ok {
curvesAdded[curv] = struct{}{}
config.CurvePreferences = append(config.CurvePreferences, curv)
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
}
}
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
// ensure ALPN includes the ACME TLS-ALPN protocol
var alpnFound bool
for _, a := range c.ALPN {
if a == tlsalpn01.ACMETLS1Protocol {
alpnFound = true
break
}
}
if !alpnFound {
c.ALPN = append(c.ALPN, tlsalpn01.ACMETLS1Protocol)
}
2017-04-29 19:28:18 -05:00
config.MinVersion = c.ProtocolMinVersion
config.MaxVersion = c.ProtocolMaxVersion
config.ClientAuth = c.ClientAuth
config.NextProtos = c.ALPN
config.GetCertificate = c.Manager.GetCertificate
// set up client authentication if enabled
if config.ClientAuth != tls.NoClientCert {
pool := x509.NewCertPool()
clientCertsAdded := make(map[string]struct{})
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
2017-04-29 19:28:18 -05:00
for _, caFile := range c.ClientCerts {
// don't add cert to pool more than once
if _, ok := clientCertsAdded[caFile]; ok {
continue
}
clientCertsAdded[caFile] = struct{}{}
// Any client with a certificate from this CA will be allowed to connect
caCrt, err := ioutil.ReadFile(caFile)
if err != nil {
return err
}
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
if !pool.AppendCertsFromPEM(caCrt) {
return fmt.Errorf("error loading client certificate '%s': no certificates were successfully parsed", caFile)
}
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
}
config.ClientCAs = pool
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
}
// default cipher suites
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
if len(config.CipherSuites) == 0 {
config.CipherSuites = getPreferredDefaultCiphers()
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
}
// for security, ensure TLS_FALLBACK_SCSV is always included first
2016-08-25 18:13:27 -05:00
if len(config.CipherSuites) == 0 || config.CipherSuites[0] != tls.TLS_FALLBACK_SCSV {
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
config.CipherSuites = append([]uint16{tls.TLS_FALLBACK_SCSV}, config.CipherSuites...)
}
// store the resulting new tls.Config
2017-04-29 19:28:18 -05:00
c.tlsConfig = config
return nil
}
// MakeTLSConfig makes a tls.Config from configs. The returned
// tls.Config is programmed to load the matching caddytls.Config
tls: Restructure and improve certificate management - Expose the list of Caddy instances through caddy.Instances() - Added arbitrary storage to caddy.Instance - The cache of loaded certificates is no longer global; now scoped per-instance, meaning upon reload (like SIGUSR1) the old cert cache will be discarded entirely, whereas before, aggressively reloading config that added and removed lots of sites would cause unnecessary build-up in the cache over time. - Key certificates in the cache by their SHA-256 hash instead of by their names. This means certificates will not be duplicated in memory (within each instance), making Caddy much more memory-efficient for large-scale deployments with thousands of sites sharing certs. - Perform name-to-certificate lookups scoped per caddytls.Config instead of a single global lookup. This prevents certificates from stepping on each other when they overlap in their names. - Do not allow TLS configurations keyed by the same hostname to be different; this now throws an error. - Updated relevant tests, with a stark awareness that more tests are needed. - Change the NewContext function signature to include an *Instance. - Strongly recommend (basically require) use of caddytls.NewConfig() to create a new *caddytls.Config, to ensure pointers to the instance certificate cache are initialized properly. - Update the TLS-SNI challenge solver (even though TLS-SNI is disabled currently on the CA side). Store temporary challenge cert in instance cache, but do so directly by the ACME challenge name, not the hash. Modified the getCertificate function to check the cache directly for a name match if one isn't found otherwise. This will allow any caddytls.Config to be able to help solve a TLS-SNI challenge, with one extra side-effect that might actually be kind of interesting (and useless): clients could send a certificate's hash as the SNI and Caddy would be able to serve that certificate for the handshake. - Do not attempt to match a "default" (random) certificate when SNI is present but unrecognized; return no certificate so a TLS alert happens instead. - Store an Instance in the list of instances even while the instance is still starting up (this allows access to the cert cache for performing renewals at startup, etc). Will be removed from list again if instance startup fails. - Laid groundwork for ACMEv2 and Let's Encrypt wildcard support. Server type plugins will need to be updated slightly to accommodate minor adjustments to their API (like passing in an Instance). This commit includes the changes for the HTTP server. Certain Caddyfile configurations might error out with this change, if they configured different TLS settings for the same hostname. This change trades some complexity for other complexity, but ultimately this new complexity is more correct and robust than earlier logic. Fixes #1991 Fixes #1994 Fixes #1303
2018-02-04 02:58:27 -05:00
// based on the hostname in SNI, but that's all. This is used
// to create a single TLS configuration for a listener (a group
// of sites).
func MakeTLSConfig(configs []*Config) (*tls.Config, error) {
if len(configs) == 0 {
return nil, nil
}
configMap := make(configGroup)
for i, cfg := range configs {
if cfg == nil {
// avoid nil pointer dereference below this loop
configs[i] = new(Config)
continue
}
// can't serve TLS and non-TLS on same port
if i > 0 && cfg.Enabled != configs[i-1].Enabled {
thisConfProto, lastConfProto := "not TLS", "not TLS"
if cfg.Enabled {
thisConfProto = "TLS"
}
if configs[i-1].Enabled {
lastConfProto = "TLS"
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
}
return nil, fmt.Errorf("cannot multiplex %s (%s) and %s (%s) on same listener",
configs[i-1].Hostname, lastConfProto, cfg.Hostname, thisConfProto)
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
}
tls: Restructure and improve certificate management - Expose the list of Caddy instances through caddy.Instances() - Added arbitrary storage to caddy.Instance - The cache of loaded certificates is no longer global; now scoped per-instance, meaning upon reload (like SIGUSR1) the old cert cache will be discarded entirely, whereas before, aggressively reloading config that added and removed lots of sites would cause unnecessary build-up in the cache over time. - Key certificates in the cache by their SHA-256 hash instead of by their names. This means certificates will not be duplicated in memory (within each instance), making Caddy much more memory-efficient for large-scale deployments with thousands of sites sharing certs. - Perform name-to-certificate lookups scoped per caddytls.Config instead of a single global lookup. This prevents certificates from stepping on each other when they overlap in their names. - Do not allow TLS configurations keyed by the same hostname to be different; this now throws an error. - Updated relevant tests, with a stark awareness that more tests are needed. - Change the NewContext function signature to include an *Instance. - Strongly recommend (basically require) use of caddytls.NewConfig() to create a new *caddytls.Config, to ensure pointers to the instance certificate cache are initialized properly. - Update the TLS-SNI challenge solver (even though TLS-SNI is disabled currently on the CA side). Store temporary challenge cert in instance cache, but do so directly by the ACME challenge name, not the hash. Modified the getCertificate function to check the cache directly for a name match if one isn't found otherwise. This will allow any caddytls.Config to be able to help solve a TLS-SNI challenge, with one extra side-effect that might actually be kind of interesting (and useless): clients could send a certificate's hash as the SNI and Caddy would be able to serve that certificate for the handshake. - Do not attempt to match a "default" (random) certificate when SNI is present but unrecognized; return no certificate so a TLS alert happens instead. - Store an Instance in the list of instances even while the instance is still starting up (this allows access to the cert cache for performing renewals at startup, etc). Will be removed from list again if instance startup fails. - Laid groundwork for ACMEv2 and Let's Encrypt wildcard support. Server type plugins will need to be updated slightly to accommodate minor adjustments to their API (like passing in an Instance). This commit includes the changes for the HTTP server. Certain Caddyfile configurations might error out with this change, if they configured different TLS settings for the same hostname. This change trades some complexity for other complexity, but ultimately this new complexity is more correct and robust than earlier logic. Fixes #1991 Fixes #1994 Fixes #1303
2018-02-04 02:58:27 -05:00
// convert this caddytls.Config into a tls.Config
if err := cfg.buildStandardTLSConfig(); err != nil {
return nil, err
}
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
tls: Restructure and improve certificate management - Expose the list of Caddy instances through caddy.Instances() - Added arbitrary storage to caddy.Instance - The cache of loaded certificates is no longer global; now scoped per-instance, meaning upon reload (like SIGUSR1) the old cert cache will be discarded entirely, whereas before, aggressively reloading config that added and removed lots of sites would cause unnecessary build-up in the cache over time. - Key certificates in the cache by their SHA-256 hash instead of by their names. This means certificates will not be duplicated in memory (within each instance), making Caddy much more memory-efficient for large-scale deployments with thousands of sites sharing certs. - Perform name-to-certificate lookups scoped per caddytls.Config instead of a single global lookup. This prevents certificates from stepping on each other when they overlap in their names. - Do not allow TLS configurations keyed by the same hostname to be different; this now throws an error. - Updated relevant tests, with a stark awareness that more tests are needed. - Change the NewContext function signature to include an *Instance. - Strongly recommend (basically require) use of caddytls.NewConfig() to create a new *caddytls.Config, to ensure pointers to the instance certificate cache are initialized properly. - Update the TLS-SNI challenge solver (even though TLS-SNI is disabled currently on the CA side). Store temporary challenge cert in instance cache, but do so directly by the ACME challenge name, not the hash. Modified the getCertificate function to check the cache directly for a name match if one isn't found otherwise. This will allow any caddytls.Config to be able to help solve a TLS-SNI challenge, with one extra side-effect that might actually be kind of interesting (and useless): clients could send a certificate's hash as the SNI and Caddy would be able to serve that certificate for the handshake. - Do not attempt to match a "default" (random) certificate when SNI is present but unrecognized; return no certificate so a TLS alert happens instead. - Store an Instance in the list of instances even while the instance is still starting up (this allows access to the cert cache for performing renewals at startup, etc). Will be removed from list again if instance startup fails. - Laid groundwork for ACMEv2 and Let's Encrypt wildcard support. Server type plugins will need to be updated slightly to accommodate minor adjustments to their API (like passing in an Instance). This commit includes the changes for the HTTP server. Certain Caddyfile configurations might error out with this change, if they configured different TLS settings for the same hostname. This change trades some complexity for other complexity, but ultimately this new complexity is more correct and robust than earlier logic. Fixes #1991 Fixes #1994 Fixes #1303
2018-02-04 02:58:27 -05:00
// if an existing config with this hostname was already
// configured, then they must be identical (or at least
// compatible), otherwise that is a configuration error
if otherConfig, ok := configMap[cfg.Hostname]; ok {
if err := assertConfigsCompatible(cfg, otherConfig); err != nil {
return nil, fmt.Errorf("incompatible TLS configurations for the same SNI "+
tls: Restructure and improve certificate management - Expose the list of Caddy instances through caddy.Instances() - Added arbitrary storage to caddy.Instance - The cache of loaded certificates is no longer global; now scoped per-instance, meaning upon reload (like SIGUSR1) the old cert cache will be discarded entirely, whereas before, aggressively reloading config that added and removed lots of sites would cause unnecessary build-up in the cache over time. - Key certificates in the cache by their SHA-256 hash instead of by their names. This means certificates will not be duplicated in memory (within each instance), making Caddy much more memory-efficient for large-scale deployments with thousands of sites sharing certs. - Perform name-to-certificate lookups scoped per caddytls.Config instead of a single global lookup. This prevents certificates from stepping on each other when they overlap in their names. - Do not allow TLS configurations keyed by the same hostname to be different; this now throws an error. - Updated relevant tests, with a stark awareness that more tests are needed. - Change the NewContext function signature to include an *Instance. - Strongly recommend (basically require) use of caddytls.NewConfig() to create a new *caddytls.Config, to ensure pointers to the instance certificate cache are initialized properly. - Update the TLS-SNI challenge solver (even though TLS-SNI is disabled currently on the CA side). Store temporary challenge cert in instance cache, but do so directly by the ACME challenge name, not the hash. Modified the getCertificate function to check the cache directly for a name match if one isn't found otherwise. This will allow any caddytls.Config to be able to help solve a TLS-SNI challenge, with one extra side-effect that might actually be kind of interesting (and useless): clients could send a certificate's hash as the SNI and Caddy would be able to serve that certificate for the handshake. - Do not attempt to match a "default" (random) certificate when SNI is present but unrecognized; return no certificate so a TLS alert happens instead. - Store an Instance in the list of instances even while the instance is still starting up (this allows access to the cert cache for performing renewals at startup, etc). Will be removed from list again if instance startup fails. - Laid groundwork for ACMEv2 and Let's Encrypt wildcard support. Server type plugins will need to be updated slightly to accommodate minor adjustments to their API (like passing in an Instance). This commit includes the changes for the HTTP server. Certain Caddyfile configurations might error out with this change, if they configured different TLS settings for the same hostname. This change trades some complexity for other complexity, but ultimately this new complexity is more correct and robust than earlier logic. Fixes #1991 Fixes #1994 Fixes #1303
2018-02-04 02:58:27 -05:00
"name (%s) on the same listener: %v",
cfg.Hostname, err)
}
}
// key this config by its hostname (overwrites
// configs with the same hostname pattern; should
// be OK since we already asserted they are roughly
// the same); during TLS handshakes, configs are
// loaded based on the hostname pattern, according
// to client's SNI
configMap[cfg.Hostname] = cfg
}
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
// Is TLS disabled? By now, we know that all
// configs agree whether it is or not, so we
// can just look at the first one. If so,
// we're done here.
if len(configs) == 0 || !configs[0].Enabled {
return nil, nil
}
return &tls.Config{
GetConfigForClient: configMap.GetConfigForClient,
}, nil
}
tls: Restructure and improve certificate management - Expose the list of Caddy instances through caddy.Instances() - Added arbitrary storage to caddy.Instance - The cache of loaded certificates is no longer global; now scoped per-instance, meaning upon reload (like SIGUSR1) the old cert cache will be discarded entirely, whereas before, aggressively reloading config that added and removed lots of sites would cause unnecessary build-up in the cache over time. - Key certificates in the cache by their SHA-256 hash instead of by their names. This means certificates will not be duplicated in memory (within each instance), making Caddy much more memory-efficient for large-scale deployments with thousands of sites sharing certs. - Perform name-to-certificate lookups scoped per caddytls.Config instead of a single global lookup. This prevents certificates from stepping on each other when they overlap in their names. - Do not allow TLS configurations keyed by the same hostname to be different; this now throws an error. - Updated relevant tests, with a stark awareness that more tests are needed. - Change the NewContext function signature to include an *Instance. - Strongly recommend (basically require) use of caddytls.NewConfig() to create a new *caddytls.Config, to ensure pointers to the instance certificate cache are initialized properly. - Update the TLS-SNI challenge solver (even though TLS-SNI is disabled currently on the CA side). Store temporary challenge cert in instance cache, but do so directly by the ACME challenge name, not the hash. Modified the getCertificate function to check the cache directly for a name match if one isn't found otherwise. This will allow any caddytls.Config to be able to help solve a TLS-SNI challenge, with one extra side-effect that might actually be kind of interesting (and useless): clients could send a certificate's hash as the SNI and Caddy would be able to serve that certificate for the handshake. - Do not attempt to match a "default" (random) certificate when SNI is present but unrecognized; return no certificate so a TLS alert happens instead. - Store an Instance in the list of instances even while the instance is still starting up (this allows access to the cert cache for performing renewals at startup, etc). Will be removed from list again if instance startup fails. - Laid groundwork for ACMEv2 and Let's Encrypt wildcard support. Server type plugins will need to be updated slightly to accommodate minor adjustments to their API (like passing in an Instance). This commit includes the changes for the HTTP server. Certain Caddyfile configurations might error out with this change, if they configured different TLS settings for the same hostname. This change trades some complexity for other complexity, but ultimately this new complexity is more correct and robust than earlier logic. Fixes #1991 Fixes #1994 Fixes #1303
2018-02-04 02:58:27 -05:00
// assertConfigsCompatible returns an error if the two Configs
// do not have the same (or roughly compatible) configurations.
// If one of the tlsConfig pointers on either Config is nil,
// an error will be returned. If both are nil, no error.
func assertConfigsCompatible(cfg1, cfg2 *Config) error {
c1, c2 := cfg1.tlsConfig, cfg2.tlsConfig
if (c1 == nil && c2 != nil) || (c1 != nil && c2 == nil) {
return fmt.Errorf("one config is not made")
}
if c1 == nil && c2 == nil {
return nil
}
if len(c1.CipherSuites) != len(c2.CipherSuites) {
return fmt.Errorf("different number of allowed cipher suites")
}
for i, ciph := range c1.CipherSuites {
if c2.CipherSuites[i] != ciph {
return fmt.Errorf("different cipher suites or different order")
}
}
if len(c1.CurvePreferences) != len(c2.CurvePreferences) {
return fmt.Errorf("different number of allowed cipher suites")
}
for i, curve := range c1.CurvePreferences {
if c2.CurvePreferences[i] != curve {
return fmt.Errorf("different curve preferences or different order")
}
}
if len(c1.NextProtos) != len(c2.NextProtos) {
return fmt.Errorf("different number of ALPN (NextProtos) values")
}
for i, proto := range c1.NextProtos {
if c2.NextProtos[i] != proto {
return fmt.Errorf("different ALPN (NextProtos) values or different order")
}
}
if c1.PreferServerCipherSuites != c2.PreferServerCipherSuites {
return fmt.Errorf("one prefers server cipher suites, the other does not")
}
if c1.MinVersion != c2.MinVersion {
return fmt.Errorf("minimum TLS version mismatch")
}
if c1.MaxVersion != c2.MaxVersion {
return fmt.Errorf("maximum TLS version mismatch")
}
if c1.ClientAuth != c2.ClientAuth {
return fmt.Errorf("client authentication policy mismatch")
}
if c1.ClientAuth != tls.NoClientCert && c2.ClientAuth != tls.NoClientCert && c1.ClientCAs != c2.ClientCAs {
// Two hosts defined on the same listener are not compatible if they
// have ClientAuth enabled, because there's no guarantee beyond the
// hostname which config will be used (because SNI only has server name).
// To prevent clients from bypassing authentication, require that
// ClientAuth be configured in an unambiguous manner.
return fmt.Errorf("multiple hosts requiring client authentication ambiguously configured")
}
tls: Restructure and improve certificate management - Expose the list of Caddy instances through caddy.Instances() - Added arbitrary storage to caddy.Instance - The cache of loaded certificates is no longer global; now scoped per-instance, meaning upon reload (like SIGUSR1) the old cert cache will be discarded entirely, whereas before, aggressively reloading config that added and removed lots of sites would cause unnecessary build-up in the cache over time. - Key certificates in the cache by their SHA-256 hash instead of by their names. This means certificates will not be duplicated in memory (within each instance), making Caddy much more memory-efficient for large-scale deployments with thousands of sites sharing certs. - Perform name-to-certificate lookups scoped per caddytls.Config instead of a single global lookup. This prevents certificates from stepping on each other when they overlap in their names. - Do not allow TLS configurations keyed by the same hostname to be different; this now throws an error. - Updated relevant tests, with a stark awareness that more tests are needed. - Change the NewContext function signature to include an *Instance. - Strongly recommend (basically require) use of caddytls.NewConfig() to create a new *caddytls.Config, to ensure pointers to the instance certificate cache are initialized properly. - Update the TLS-SNI challenge solver (even though TLS-SNI is disabled currently on the CA side). Store temporary challenge cert in instance cache, but do so directly by the ACME challenge name, not the hash. Modified the getCertificate function to check the cache directly for a name match if one isn't found otherwise. This will allow any caddytls.Config to be able to help solve a TLS-SNI challenge, with one extra side-effect that might actually be kind of interesting (and useless): clients could send a certificate's hash as the SNI and Caddy would be able to serve that certificate for the handshake. - Do not attempt to match a "default" (random) certificate when SNI is present but unrecognized; return no certificate so a TLS alert happens instead. - Store an Instance in the list of instances even while the instance is still starting up (this allows access to the cert cache for performing renewals at startup, etc). Will be removed from list again if instance startup fails. - Laid groundwork for ACMEv2 and Let's Encrypt wildcard support. Server type plugins will need to be updated slightly to accommodate minor adjustments to their API (like passing in an Instance). This commit includes the changes for the HTTP server. Certain Caddyfile configurations might error out with this change, if they configured different TLS settings for the same hostname. This change trades some complexity for other complexity, but ultimately this new complexity is more correct and robust than earlier logic. Fixes #1991 Fixes #1994 Fixes #1303
2018-02-04 02:58:27 -05:00
return nil
}
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
// ConfigGetter gets a Config keyed by key.
type ConfigGetter func(c *caddy.Controller) *Config
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
var configGetters = make(map[string]ConfigGetter)
// RegisterConfigGetter registers fn as the way to get a
// Config for server type serverType.
func RegisterConfigGetter(serverType string, fn ConfigGetter) {
configGetters[serverType] = fn
}
// SetDefaultTLSParams sets the default TLS cipher suites, protocol versions,
// and server preferences of a server.Config if they were not previously set
// (it does not overwrite; only fills in missing values).
func SetDefaultTLSParams(config *Config) {
// If no ciphers provided, use default list
if len(config.Ciphers) == 0 {
config.Ciphers = getPreferredDefaultCiphers()
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
}
// Not a cipher suite, but still important for mitigating protocol downgrade attacks
// (prepend since having it at end breaks http2 due to non-h2-approved suites before it)
config.Ciphers = append([]uint16{tls.TLS_FALLBACK_SCSV}, config.Ciphers...)
// If no curves provided, use default list
if len(config.CurvePreferences) == 0 {
config.CurvePreferences = defaultCurves
}
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
// Set default protocol min and max versions - must balance compatibility and security
if config.ProtocolMinVersion == 0 {
config.ProtocolMinVersion = tls.VersionTLS12
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
}
if config.ProtocolMaxVersion == 0 {
config.ProtocolMaxVersion = tls.VersionTLS12
}
// Prefer server cipher suites
config.PreferServerCipherSuites = true
}
// Map of supported key types
var supportedKeyTypes = map[string]certcrypto.KeyType{
"P384": certcrypto.EC384,
"P256": certcrypto.EC256,
"RSA8192": certcrypto.RSA8192,
"RSA4096": certcrypto.RSA4096,
"RSA2048": certcrypto.RSA2048,
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
}
// SupportedProtocols is a map of supported protocols.
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
// HTTP/2 only supports TLS 1.2 and higher.
// If updating this map, also update tlsProtocolStringToMap in caddyhttp/fastcgi/fastcgi.go
var SupportedProtocols = map[string]uint16{
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
"tls1.0": tls.VersionTLS10,
"tls1.1": tls.VersionTLS11,
"tls1.2": tls.VersionTLS12,
}
// GetSupportedProtocolName returns the protocol name
func GetSupportedProtocolName(protocol uint16) (string, error) {
for k, v := range SupportedProtocols {
if v == protocol {
return k, nil
}
}
return "", fmt.Errorf("name: unsuported protocol")
}
// SupportedCiphersMap has supported ciphers, used only for parsing config.
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
//
// Note that, at time of writing, HTTP/2 blacklists 276 cipher suites,
// including all but four of the suites below (the four GCM suites).
// See https://http2.github.io/http2-spec/#BadCipherSuites
//
// TLS_FALLBACK_SCSV is not in this list because we manually ensure
// it is always added (even though it is not technically a cipher suite).
//
// This map, like any map, is NOT ORDERED. Do not range over this map.
var SupportedCiphersMap = map[string]uint16{
"ECDHE-ECDSA-AES256-GCM-SHA384": tls.TLS_ECDHE_ECDSA_WITH_AES_256_GCM_SHA384,
"ECDHE-RSA-AES256-GCM-SHA384": tls.TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384,
"ECDHE-ECDSA-AES128-GCM-SHA256": tls.TLS_ECDHE_ECDSA_WITH_AES_128_GCM_SHA256,
"ECDHE-RSA-AES128-GCM-SHA256": tls.TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256,
"ECDHE-ECDSA-WITH-CHACHA20-POLY1305": tls.TLS_ECDHE_ECDSA_WITH_CHACHA20_POLY1305,
"ECDHE-RSA-WITH-CHACHA20-POLY1305": tls.TLS_ECDHE_RSA_WITH_CHACHA20_POLY1305,
"ECDHE-RSA-AES256-CBC-SHA": tls.TLS_ECDHE_RSA_WITH_AES_256_CBC_SHA,
"ECDHE-RSA-AES128-CBC-SHA": tls.TLS_ECDHE_RSA_WITH_AES_128_CBC_SHA,
"ECDHE-ECDSA-AES256-CBC-SHA": tls.TLS_ECDHE_ECDSA_WITH_AES_256_CBC_SHA,
"ECDHE-ECDSA-AES128-CBC-SHA": tls.TLS_ECDHE_ECDSA_WITH_AES_128_CBC_SHA,
"RSA-AES256-CBC-SHA": tls.TLS_RSA_WITH_AES_256_CBC_SHA,
"RSA-AES128-CBC-SHA": tls.TLS_RSA_WITH_AES_128_CBC_SHA,
"ECDHE-RSA-3DES-EDE-CBC-SHA": tls.TLS_ECDHE_RSA_WITH_3DES_EDE_CBC_SHA,
"RSA-3DES-EDE-CBC-SHA": tls.TLS_RSA_WITH_3DES_EDE_CBC_SHA,
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
}
// GetSupportedCipherName returns the cipher name
func GetSupportedCipherName(cipher uint16) (string, error) {
for k, v := range SupportedCiphersMap {
if v == cipher {
return k, nil
}
}
return "", fmt.Errorf("name: unsuported cipher")
}
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
// List of all the ciphers we want to use by default
var defaultCiphers = []uint16{
tls.TLS_ECDHE_ECDSA_WITH_AES_256_GCM_SHA384,
tls.TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384,
tls.TLS_ECDHE_ECDSA_WITH_AES_128_GCM_SHA256,
tls.TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256,
tls.TLS_ECDHE_ECDSA_WITH_CHACHA20_POLY1305,
tls.TLS_ECDHE_RSA_WITH_CHACHA20_POLY1305,
Rewrote Caddy from the ground up; initial commit of 0.9 branch These changes span work from the last ~4 months in an effort to make Caddy more extensible, reduce the coupling between its components, and lay a more robust foundation of code going forward into 1.0. A bunch of new features have been added, too, with even higher future potential. The most significant design change is an overall inversion of dependencies. Instead of the caddy package knowing about the server and the notion of middleware and config, the caddy package exposes an interface that other components plug into. This does introduce more indirection when reading the code, but every piece is very modular and pluggable. Even the HTTP server is pluggable. The caddy package has been moved to the top level, and main has been pushed into a subfolder called caddy. The actual logic of the main file has been pushed even further into caddy/caddymain/run.go so that custom builds of Caddy can be 'go get'able. The HTTPS logic was surgically separated into two parts to divide the TLS-specific code and the HTTPS-specific code. The caddytls package can now be used by any type of server that needs TLS, not just HTTP. I also added the ability to customize nearly every aspect of TLS at the site level rather than all sites sharing the same TLS configuration. Not all of this flexibility is exposed in the Caddyfile yet, but it may be in the future. Caddy can also generate self-signed certificates in memory for the convenience of a developer working on localhost who wants HTTPS. And Caddy now supports the DNS challenge, assuming at least one DNS provider is plugged in. Dozens, if not hundreds, of other minor changes swept through the code base as I literally started from an empty main function, copying over functions or files as needed, then adjusting them to fit in the new design. Most tests have been restored and adapted to the new API, but more work is needed there. A lot of what was "impossible" before is now possible, or can be made possible with minimal disruption of the code. For example, it's fairly easy to make plugins hook into another part of the code via callbacks. Plugins can do more than just be directives; we now have plugins that customize how the Caddyfile is loaded (useful when you need to get your configuration from a remote store). Site addresses no longer need be just a host and port. They can have a path, allowing you to scope a configuration to a specific path. There is no inheretance, however; each site configuration is distinct. Thanks to amazing work by Lucas Clemente, this commit adds experimental QUIC support. Turn it on using the -quic flag; your browser may have to be configured to enable it. Almost everything is here, but you will notice that most of the middle- ware are missing. After those are transferred over, we'll be ready for beta tests. I'm very excited to get this out. Thanks for everyone's help and patience these last few months. I hope you like it!!
2016-06-04 18:00:29 -05:00
tls.TLS_ECDHE_RSA_WITH_AES_256_CBC_SHA,
tls.TLS_ECDHE_RSA_WITH_AES_128_CBC_SHA,
tls.TLS_ECDHE_ECDSA_WITH_AES_256_CBC_SHA,
tls.TLS_ECDHE_ECDSA_WITH_AES_128_CBC_SHA,
}
// List of ciphers we should prefer if native AESNI support is missing
var defaultCiphersNonAESNI = []uint16{
tls.TLS_ECDHE_ECDSA_WITH_CHACHA20_POLY1305,
tls.TLS_ECDHE_RSA_WITH_CHACHA20_POLY1305,
tls.TLS_ECDHE_ECDSA_WITH_AES_256_GCM_SHA384,
tls.TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384,
tls.TLS_ECDHE_ECDSA_WITH_AES_128_GCM_SHA256,
tls.TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256,
tls.TLS_ECDHE_RSA_WITH_AES_256_CBC_SHA,
tls.TLS_ECDHE_RSA_WITH_AES_128_CBC_SHA,
tls.TLS_ECDHE_ECDSA_WITH_AES_256_CBC_SHA,
tls.TLS_ECDHE_ECDSA_WITH_AES_128_CBC_SHA,
}
// getPreferredDefaultCiphers returns an appropriate cipher suite to use, depending on
// the hardware support available for AES-NI.
//
// See https://github.com/mholt/caddy/issues/1674
func getPreferredDefaultCiphers() []uint16 {
if cpuid.CPU.AesNi() {
return defaultCiphers
}
// Return a cipher suite that prefers ChaCha20
return defaultCiphersNonAESNI
}
// Map of supported curves
// https://golang.org/pkg/crypto/tls/#CurveID
var supportedCurvesMap = map[string]tls.CurveID{
"X25519": tls.X25519,
"P256": tls.CurveP256,
"P384": tls.CurveP384,
"P521": tls.CurveP521,
}
tls: Restructure and improve certificate management - Expose the list of Caddy instances through caddy.Instances() - Added arbitrary storage to caddy.Instance - The cache of loaded certificates is no longer global; now scoped per-instance, meaning upon reload (like SIGUSR1) the old cert cache will be discarded entirely, whereas before, aggressively reloading config that added and removed lots of sites would cause unnecessary build-up in the cache over time. - Key certificates in the cache by their SHA-256 hash instead of by their names. This means certificates will not be duplicated in memory (within each instance), making Caddy much more memory-efficient for large-scale deployments with thousands of sites sharing certs. - Perform name-to-certificate lookups scoped per caddytls.Config instead of a single global lookup. This prevents certificates from stepping on each other when they overlap in their names. - Do not allow TLS configurations keyed by the same hostname to be different; this now throws an error. - Updated relevant tests, with a stark awareness that more tests are needed. - Change the NewContext function signature to include an *Instance. - Strongly recommend (basically require) use of caddytls.NewConfig() to create a new *caddytls.Config, to ensure pointers to the instance certificate cache are initialized properly. - Update the TLS-SNI challenge solver (even though TLS-SNI is disabled currently on the CA side). Store temporary challenge cert in instance cache, but do so directly by the ACME challenge name, not the hash. Modified the getCertificate function to check the cache directly for a name match if one isn't found otherwise. This will allow any caddytls.Config to be able to help solve a TLS-SNI challenge, with one extra side-effect that might actually be kind of interesting (and useless): clients could send a certificate's hash as the SNI and Caddy would be able to serve that certificate for the handshake. - Do not attempt to match a "default" (random) certificate when SNI is present but unrecognized; return no certificate so a TLS alert happens instead. - Store an Instance in the list of instances even while the instance is still starting up (this allows access to the cert cache for performing renewals at startup, etc). Will be removed from list again if instance startup fails. - Laid groundwork for ACMEv2 and Let's Encrypt wildcard support. Server type plugins will need to be updated slightly to accommodate minor adjustments to their API (like passing in an Instance). This commit includes the changes for the HTTP server. Certain Caddyfile configurations might error out with this change, if they configured different TLS settings for the same hostname. This change trades some complexity for other complexity, but ultimately this new complexity is more correct and robust than earlier logic. Fixes #1991 Fixes #1994 Fixes #1303
2018-02-04 02:58:27 -05:00
// List of all the curves we want to use by default.
//
// This list should only include curves which are fast by design (e.g. X25519)
// and those for which an optimized assembly implementation exists (e.g. P256).
// The latter ones can be found here: https://github.com/golang/go/tree/master/src/crypto/elliptic
var defaultCurves = []tls.CurveID{
tls.X25519,
tls.CurveP256,
}
// CertCacheInstStorageKey is the name of the key for
// accessing the certificate storage on the *caddy.Instance.
const CertCacheInstStorageKey = "tls_cert_cache"