caddy/caddytls/setup.go

277 lines
7.7 KiB
Go
Raw Normal View History

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!!
2016-06-05 07:00:29 +08:00
package caddytls
import (
"bytes"
"crypto/tls"
"encoding/pem"
"fmt"
"io/ioutil"
"log"
"os"
"path/filepath"
"strconv"
"strings"
"github.com/mholt/caddy"
)
func init() {
caddy.RegisterPlugin("tls", caddy.Plugin{Action: setupTLS})
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!!
2016-06-05 07:00:29 +08:00
}
// setupTLS sets up the TLS configuration and installs certificates that
// are specified by the user in the config file. All the automatic HTTPS
// stuff comes later outside of this function.
func setupTLS(c *caddy.Controller) error {
configGetter, ok := configGetters[c.ServerType()]
if !ok {
return fmt.Errorf("no caddytls.ConfigGetter for %s server type; must call RegisterConfigGetter", c.ServerType())
}
config := configGetter(c)
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!!
2016-06-05 07:00:29 +08:00
if config == nil {
return fmt.Errorf("no caddytls.Config to set up for %s", c.Key)
}
config.Enabled = true
for c.Next() {
var certificateFile, keyFile, loadDir, maxCerts string
args := c.RemainingArgs()
switch len(args) {
case 1:
// even if the email is one of the special values below,
// it is still necessary for future analysis that we store
// that value in the ACMEEmail field.
config.ACMEEmail = args[0]
// user can force-disable managed TLS this way
if args[0] == "off" {
config.Enabled = false
return nil
}
// user might want a temporary, in-memory, self-signed cert
if args[0] == "self_signed" {
config.SelfSigned = true
}
case 2:
certificateFile = args[0]
keyFile = args[1]
config.Manual = true
}
// Optional block with extra parameters
var hadBlock bool
for c.NextBlock() {
hadBlock = true
switch c.Val() {
case "key_type":
arg := c.RemainingArgs()
value, ok := supportedKeyTypes[strings.ToUpper(arg[0])]
if !ok {
return c.Errf("Wrong key type name or key type not supported: '%s'", c.Val())
}
config.KeyType = value
case "protocols":
args := c.RemainingArgs()
if len(args) != 2 {
return c.ArgErr()
}
value, ok := supportedProtocols[strings.ToLower(args[0])]
if !ok {
return c.Errf("Wrong protocol name or protocol not supported: '%s'", args[0])
}
config.ProtocolMinVersion = value
value, ok = supportedProtocols[strings.ToLower(args[1])]
if !ok {
return c.Errf("Wrong protocol name or protocol not supported: '%s'", args[1])
}
config.ProtocolMaxVersion = value
if config.ProtocolMinVersion > config.ProtocolMaxVersion {
return c.Errf("Minimum protocol version cannot be higher than maximum (reverse the order)")
}
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!!
2016-06-05 07:00:29 +08:00
case "ciphers":
for c.NextArg() {
value, ok := supportedCiphersMap[strings.ToUpper(c.Val())]
if !ok {
return c.Errf("Wrong cipher name or cipher not supported: '%s'", c.Val())
}
config.Ciphers = append(config.Ciphers, value)
}
case "clients":
clientCertList := c.RemainingArgs()
if len(clientCertList) == 0 {
return c.ArgErr()
}
listStart, mustProvideCA := 1, true
switch clientCertList[0] {
case "request":
config.ClientAuth = tls.RequestClientCert
mustProvideCA = false
case "require":
config.ClientAuth = tls.RequireAnyClientCert
mustProvideCA = false
case "verify_if_given":
config.ClientAuth = tls.VerifyClientCertIfGiven
default:
config.ClientAuth = tls.RequireAndVerifyClientCert
listStart = 0
}
if mustProvideCA && len(clientCertList) <= listStart {
return c.ArgErr()
}
config.ClientCerts = clientCertList[listStart:]
case "load":
c.Args(&loadDir)
config.Manual = true
case "max_certs":
c.Args(&maxCerts)
config.OnDemand = true
case "dns":
args := c.RemainingArgs()
if len(args) != 1 {
return c.ArgErr()
}
dnsProvName := args[0]
if _, ok := dnsProviders[dnsProvName]; !ok {
return c.Errf("Unsupported DNS provider '%s'", args[0])
}
config.DNSProvider = args[0]
default:
return c.Errf("Unknown keyword '%s'", c.Val())
}
}
// tls requires at least one argument if a block is not opened
if len(args) == 0 && !hadBlock {
return c.ArgErr()
}
// set certificate limit if on-demand TLS is enabled
if maxCerts != "" {
maxCertsNum, err := strconv.Atoi(maxCerts)
if err != nil || maxCertsNum < 1 {
return c.Err("max_certs must be a positive integer")
}
config.OnDemandState.MaxObtain = int32(maxCertsNum)
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!!
2016-06-05 07:00:29 +08:00
}
// don't try to load certificates unless we're supposed to
if !config.Enabled || !config.Manual {
continue
}
// load a single certificate and key, if specified
if certificateFile != "" && keyFile != "" {
err := cacheUnmanagedCertificatePEMFile(certificateFile, keyFile)
if err != nil {
return c.Errf("Unable to load certificate and key files for '%s': %v", c.Key, err)
}
log.Printf("[INFO] Successfully loaded TLS assets from %s and %s", certificateFile, keyFile)
}
// load a directory of certificates, if specified
if loadDir != "" {
err := loadCertsInDir(c, loadDir)
if err != nil {
return err
}
}
}
SetDefaultTLSParams(config)
// generate self-signed cert if needed
if config.SelfSigned {
err := makeSelfSignedCert(config)
if err != nil {
return fmt.Errorf("self-signed: %v", err)
}
}
return nil
}
// loadCertsInDir loads all the certificates/keys in dir, as long as
// the file ends with .pem. This method of loading certificates is
// modeled after haproxy, which expects the certificate and key to
// be bundled into the same file:
// https://cbonte.github.io/haproxy-dconv/configuration-1.5.html#5.1-crt
//
// This function may write to the log as it walks the directory tree.
func loadCertsInDir(c *caddy.Controller, dir string) error {
return filepath.Walk(dir, func(path string, info os.FileInfo, err error) error {
if err != nil {
log.Printf("[WARNING] Unable to traverse into %s; skipping", path)
return nil
}
if info.IsDir() {
return nil
}
if strings.HasSuffix(strings.ToLower(info.Name()), ".pem") {
certBuilder, keyBuilder := new(bytes.Buffer), new(bytes.Buffer)
var foundKey bool // use only the first key in the file
bundle, err := ioutil.ReadFile(path)
if err != nil {
return err
}
for {
// Decode next block so we can see what type it is
var derBlock *pem.Block
derBlock, bundle = pem.Decode(bundle)
if derBlock == nil {
break
}
if derBlock.Type == "CERTIFICATE" {
// Re-encode certificate as PEM, appending to certificate chain
pem.Encode(certBuilder, derBlock)
} else if derBlock.Type == "EC PARAMETERS" {
// EC keys generated from openssl can be composed of two blocks:
// parameters and key (parameter block should come first)
if !foundKey {
// Encode parameters
pem.Encode(keyBuilder, derBlock)
// Key must immediately follow
derBlock, bundle = pem.Decode(bundle)
if derBlock == nil || derBlock.Type != "EC PRIVATE KEY" {
return c.Errf("%s: expected elliptic private key to immediately follow EC parameters", path)
}
pem.Encode(keyBuilder, derBlock)
foundKey = true
}
} else if derBlock.Type == "PRIVATE KEY" || strings.HasSuffix(derBlock.Type, " PRIVATE KEY") {
// RSA key
if !foundKey {
pem.Encode(keyBuilder, derBlock)
foundKey = true
}
} else {
return c.Errf("%s: unrecognized PEM block type: %s", path, derBlock.Type)
}
}
certPEMBytes, keyPEMBytes := certBuilder.Bytes(), keyBuilder.Bytes()
if len(certPEMBytes) == 0 {
return c.Errf("%s: failed to parse PEM data", path)
}
if len(keyPEMBytes) == 0 {
return c.Errf("%s: no private key block found", path)
}
err = cacheUnmanagedCertificatePEMBytes(certPEMBytes, keyPEMBytes)
if err != nil {
return c.Errf("%s: failed to load cert and key for '%s': %v", path, c.Key, err)
}
log.Printf("[INFO] Successfully loaded TLS assets from %s", path)
}
return nil
})
}