postgresql/src/interfaces/libpq/fe-secure.c

892 lines
22 KiB
C
Raw Normal View History

UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
/*-------------------------------------------------------------------------
*
* fe-secure.c
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
* functions related to setting up a secure connection to the backend.
* Secure connections are expected to provide confidentiality,
* message integrity and endpoint authentication.
*
*
2002-06-20 22:29:54 +02:00
* Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
* Portions Copyright (c) 1994, Regents of the University of California
*
*
* IDENTIFICATION
* $Header: /cvsroot/pgsql/src/interfaces/libpq/fe-secure.c,v 1.10 2002/07/20 05:43:31 momjian Exp $
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
*
* NOTES
* The client *requires* a valid server certificate. Since
* SSH tunnels provide anonymous confidentiality, the presumption
* is that sites that want endpoint authentication will use the
* direct SSL support, while sites that are comfortable with
* anonymous connections will use SSH tunnels.
*
* This code verifies the server certificate, to detect simple
* "man-in-the-middle" and "impersonation" attacks. The
* server certificate, or better yet the CA certificate used
* to sign the server certificate, should be present in the
* "$HOME/.postgresql/root.crt" file. If this file isn't
* readable, or the server certificate can't be validated,
* pqsecure_open_client() will return an error code.
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
*
* Additionally, the server certificate's "common name" must
* resolve to the other end of the socket. This makes it
* substantially harder to pull off a "man-in-the-middle" or
* "impersonation" attack even if the server's private key
* has been stolen. This check limits acceptable network
* layers to Unix sockets (weird, but legal), TCPv4 and TCPv6.
*
* Unfortunately neither the current front- or back-end handle
* failure gracefully, resulting in the backend hiccupping.
* This points out problems in each (the frontend shouldn't even
* try to do SSL if pqsecure_initialize() fails, and the backend
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
* shouldn't crash/recover if an SSH negotiation fails. The
* backend definitely needs to be fixed, to prevent a "denial
* of service" attack, but I don't know enough about how the
* backend works (especially that pre-SSL negotiation) to identify
* a fix.
*
* ...
*
* Unlike the server's static private key, the client's
* static private key ($HOME/.postgresql/postgresql.key)
* should normally be stored encrypted. However we still
* support EPH since it's useful for other reasons.
*
* ...
*
* Client certificates are supported, if the server requests
* or requires them. Client certificates can be used for
* authentication, to prevent sessions from being hijacked,
* or to allow "road warriors" to access the database while
* keeping it closed to everyone else.
*
* The user's certificate and private key are located in
* $HOME/.postgresql/postgresql.crt
* and
* $HOME/.postgresql/postgresql.key
* respectively.
*
* ...
*
* We don't provide informational callbacks here (like
* info_cb() in be-secure.c), since there's mechanism to
* display that information to the client.
*
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
* OS DEPENDENCIES
* The code currently assumes a POSIX password entry. How should
* Windows and Mac users be handled?
*
*-------------------------------------------------------------------------
*/
#include "postgres_fe.h"
#include <sys/types.h>
#include <signal.h>
#include <fcntl.h>
#include <errno.h>
#include <ctype.h>
#include <string.h>
#include "libpq-fe.h"
#include "libpq-int.h"
#include "fe-auth.h"
#include "pqsignal.h"
#ifdef WIN32
#include "win32.h"
#else
#include <sys/socket.h>
#include <unistd.h>
#include <netdb.h>
#include <netinet/in.h>
#ifdef HAVE_NETINET_TCP_H
#include <netinet/tcp.h>
#endif
#include <arpa/inet.h>
#endif
#ifndef HAVE_STRDUP
#include "strdup.h"
#endif
#ifndef WIN32
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
#include <pwd.h>
#endif
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
#include <sys/stat.h>
#ifdef USE_SSL
#include <openssl/ssl.h>
#include <openssl/e_os.h>
#endif /* USE_SSL */
#ifdef USE_SSL
static int verify_cb(int ok, X509_STORE_CTX *ctx);
static int verify_peer(PGconn *);
static DH *load_dh_file(int keylength);
static DH *load_dh_buffer(const char *, size_t);
static DH *tmp_dh_cb(SSL *s, int is_export, int keylength);
static int client_cert_cb(SSL *, X509 **, EVP_PKEY **);
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
static int initialize_SSL(PGconn *);
static void destroy_SSL(void);
static int open_client_SSL(PGconn *);
static void close_SSL(PGconn *);
static const char *SSLerrmessage(void);
#endif
#ifdef USE_SSL
static SSL_CTX *SSL_context = NULL;
#endif
/* ------------------------------------------------------------ */
/* Hardcoded values */
/* ------------------------------------------------------------ */
/*
* Hardcoded DH parameters, used in empheral DH keying.
* As discussed above, EDH protects the confidentiality of
* sessions even if the static private key is compromised,
* so we are *highly* motivated to ensure that we can use
* EDH even if the user... or an attacker... deletes the
* $HOME/.postgresql/dh*.pem files.
*
* It's not critical that users have EPH keys, but it doesn't
* hurt and if it's missing someone will demand it, so....
*/
static const char file_dh512[] =
"-----BEGIN DH PARAMETERS-----\n\
MEYCQQD1Kv884bEpQBgRjXyEpwpy1obEAxnIByl6ypUM2Zafq9AKUJsCRtMIPWak\n\
XUGfnHy9iUsiGSa6q6Jew1XpKgVfAgEC\n\
-----END DH PARAMETERS-----\n";
static const char file_dh1024[] =
"-----BEGIN DH PARAMETERS-----\n\
MIGHAoGBAPSI/VhOSdvNILSd5JEHNmszbDgNRR0PfIizHHxbLY7288kjwEPwpVsY\n\
jY67VYy4XTjTNP18F1dDox0YbN4zISy1Kv884bEpQBgRjXyEpwpy1obEAxnIByl6\n\
ypUM2Zafq9AKUJsCRtMIPWakXUGfnHy9iUsiGSa6q6Jew1XpL3jHAgEC\n\
-----END DH PARAMETERS-----\n";
static const char file_dh2048[] =
"-----BEGIN DH PARAMETERS-----\n\
MIIBCAKCAQEA9kJXtwh/CBdyorrWqULzBej5UxE5T7bxbrlLOCDaAadWoxTpj0BV\n\
89AHxstDqZSt90xkhkn4DIO9ZekX1KHTUPj1WV/cdlJPPT2N286Z4VeSWc39uK50\n\
T8X8dryDxUcwYc58yWb/Ffm7/ZFexwGq01uejaClcjrUGvC/RgBYK+X0iP1YTknb\n\
zSC0neSRBzZrM2w4DUUdD3yIsxx8Wy2O9vPJI8BD8KVbGI2Ou1WMuF040zT9fBdX\n\
Q6MdGGzeMyEstSr/POGxKUAYEY18hKcKctaGxAMZyAcpesqVDNmWn6vQClCbAkbT\n\
CD1mpF1Bn5x8vYlLIhkmuquiXsNV6TILOwIBAg==\n\
-----END DH PARAMETERS-----\n";
static const char file_dh4096[] =
"-----BEGIN DH PARAMETERS-----\n\
MIICCAKCAgEA+hRyUsFN4VpJ1O8JLcCo/VWr19k3BCgJ4uk+d+KhehjdRqNDNyOQ\n\
l/MOyQNQfWXPeGKmOmIig6Ev/nm6Nf9Z2B1h3R4hExf+zTiHnvVPeRBhjdQi81rt\n\
Xeoh6TNrSBIKIHfUJWBh3va0TxxjQIs6IZOLeVNRLMqzeylWqMf49HsIXqbcokUS\n\
Vt1BkvLdW48j8PPv5DsKRN3tloTxqDJGo9tKvj1Fuk74A+Xda1kNhB7KFlqMyN98\n\
VETEJ6c7KpfOo30mnK30wqw3S8OtaIR/maYX72tGOno2ehFDkq3pnPtEbD2CScxc\n\
alJC+EL7RPk5c/tgeTvCngvc1KZn92Y//EI7G9tPZtylj2b56sHtMftIoYJ9+ODM\n\
sccD5Piz/rejE3Ome8EOOceUSCYAhXn8b3qvxVI1ddd1pED6FHRhFvLrZxFvBEM9\n\
ERRMp5QqOaHJkM+Dxv8Cj6MqrCbfC4u+ZErxodzuusgDgvZiLF22uxMZbobFWyte\n\
OvOzKGtwcTqO/1wV5gKkzu1ZVswVUQd5Gg8lJicwqRWyyNRczDDoG9jVDxmogKTH\n\
AaqLulO7R8Ifa1SwF2DteSGVtgWEN8gDpN3RBmmPTDngyF2DHb5qmpnznwtFKdTL\n\
KWbuHn491xNO25CQWMtem80uKw+pTnisBRF/454n1Jnhub144YRBoN8CAQI=\n\
-----END DH PARAMETERS-----\n";
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
/* ------------------------------------------------------------ */
/* Procedures common to all secure sessions */
/* ------------------------------------------------------------ */
/*
* Initialize global context
*/
int
pqsecure_initialize (PGconn *conn)
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
{
int r = 0;
#ifdef USE_SSL
r = initialize_SSL(conn);
#endif
return r;
}
/*
* Destroy global context
*/
void
pqsecure_destroy (void)
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
{
#ifdef USE_SSL
destroy_SSL();
#endif
}
/*
* Attempt to negotiate secure session.
*/
int
pqsecure_open_client (PGconn *conn)
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
{
int r = 0;
#ifdef USE_SSL
r = open_client_SSL(conn);
#endif
return r;
}
/*
* Close secure session.
*/
void
pqsecure_close (PGconn *conn)
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
{
#ifdef USE_SSL
if (conn->ssl)
close_SSL(conn);
#endif
}
/*
* Read data from a secure connection.
*/
ssize_t
pqsecure_read (PGconn *conn, void *ptr, size_t len)
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
{
ssize_t n;
#ifdef USE_SSL
if (conn->ssl)
{
n = SSL_read(conn->ssl, ptr, len);
switch (SSL_get_error(conn->ssl, n))
{
case SSL_ERROR_NONE:
break;
case SSL_ERROR_WANT_READ:
break;
case SSL_ERROR_SYSCALL:
SOCK_ERRNO = get_last_socket_error();
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("SSL SYSCALL error: %s\n"),
SOCK_STRERROR(SOCK_ERRNO));
break;
case SSL_ERROR_SSL:
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("SSL error: %s\n"), SSLerrmessage());
/* fall through */
case SSL_ERROR_ZERO_RETURN:
pqsecure_close(conn);
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
SOCK_ERRNO = ECONNRESET;
n = -1;
break;
}
}
else
#endif
n = recv(conn->sock, ptr, len, 0);
return n;
}
/*
* Write data to a secure connection.
*/
ssize_t
pqsecure_write (PGconn *conn, const void *ptr, size_t len)
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
{
ssize_t n;
#ifndef WIN32
pqsigfunc oldsighandler = pqsignal(SIGPIPE, SIG_IGN);
#endif
#ifdef USE_SSL
if (conn->ssl)
{
n = SSL_write(conn->ssl, ptr, len);
switch (SSL_get_error(conn->ssl, n))
{
case SSL_ERROR_NONE:
break;
case SSL_ERROR_WANT_WRITE:
break;
case SSL_ERROR_SYSCALL:
SOCK_ERRNO = get_last_socket_error();
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("SSL SYSCALL error: %s\n"),
SOCK_STRERROR(SOCK_ERRNO));
break;
case SSL_ERROR_SSL:
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("SSL error: %s\n"), SSLerrmessage());
/* fall through */
case SSL_ERROR_ZERO_RETURN:
pqsecure_close(conn);
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
SOCK_ERRNO = ECONNRESET;
n = -1;
break;
}
}
else
#endif
n = send(conn->sock, ptr, len, 0);
#ifndef WIN32
pqsignal(SIGPIPE, oldsighandler);
#endif
return n;
}
/* ------------------------------------------------------------ */
/* SSL specific code */
/* ------------------------------------------------------------ */
#ifdef USE_SSL
/*
* Certificate verification callback
*
* This callback allows us to log intermediate problems during
* verification, but there doesn't seem to be a clean way to get
* our PGconn * structure. So we can't log anything!
*
* This callback also allows us to override the default acceptance
* criteria (e.g., accepting self-signed or expired certs), but
* for now we accept the default checks.
*/
static int
verify_cb (int ok, X509_STORE_CTX *ctx)
{
return ok;
}
/*
* Verify that common name resolves to peer.
* This function is not thread-safe due to gethostbyname2().
*/
static int
verify_peer (PGconn *conn)
{
struct hostent *h = NULL;
struct sockaddr addr;
struct sockaddr_in *sin;
socklen_t len;
char **s;
unsigned long l;
/* get the address on the other side of the socket */
len = sizeof(addr);
if (getpeername(conn->sock, &addr, &len) == -1)
{
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("error querying socket: %s\n"),
SOCK_STRERROR(SOCK_ERRNO));
return -1;
}
/* weird, but legal case */
if (addr.sa_family == AF_UNIX)
return 0;
/* what do we know about the peer's common name? */
if ((h = gethostbyname2(conn->peer_cn, addr.sa_family)) == NULL)
{
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("error getting information about host (%s): %s\n"),
conn->peer_cn, hstrerror(h_errno));
return -1;
}
/* does the address match? */
switch (addr.sa_family)
{
case AF_INET:
sin = (struct sockaddr_in *) &addr;
for (s = h->h_addr_list; *s != NULL; s++)
{
if (!memcmp(&sin->sin_addr.s_addr, *s, h->h_length))
return 0;
}
break;
default:
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("sorry, this protocol not yet supported\n"));
return -1;
}
/* the prior test should be definitive, but in practice
* it sometimes fails. So we also check the aliases. */
for (s = h->h_aliases; *s != NULL; s++)
{
if (strcasecmp(conn->peer_cn, *s) == 0)
return 0;
}
/* generate protocol-aware error message */
switch (addr.sa_family)
{
case AF_INET:
sin = (struct sockaddr_in *) &addr;
l = ntohl(sin->sin_addr.s_addr);
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext(
"server common name '%s' does not resolve to %ld.%ld.%ld.%ld\n"),
conn->peer_cn, (l >> 24) % 0x100, (l >> 16) % 0x100,
(l >> 8) % 0x100, l % 0x100);
break;
default:
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext(
"server common name '%s' does not resolve to peer address\n"),
conn->peer_cn);
}
return -1;
}
/*
* Load precomputed DH parameters.
*
* To prevent "downgrade" attacks, we perform a number of checks
* to verify that the DBA-generated DH parameters file contains
* what we expect it to contain.
*/
static DH *
load_dh_file (int keylength)
{
struct passwd *pwd;
FILE *fp;
char fnbuf[2048];
DH *dh = NULL;
int codes;
if ((pwd = getpwuid(getuid())) == NULL)
return NULL;
/* attempt to open file. It's not an error if it doesn't exist. */
snprintf(fnbuf, sizeof fnbuf, "%s/.postgresql/dh%d.pem",
pwd->pw_dir, keylength);
if ((fp = fopen(fnbuf, "r")) == NULL)
return NULL;
/* flock(fileno(fp), LOCK_SH); */
dh = PEM_read_DHparams(fp, NULL, NULL, NULL);
/* flock(fileno(fp), LOCK_UN); */
fclose(fp);
/* is the prime the correct size? */
if (dh != NULL && 8*DH_size(dh) < keylength)
{
dh = NULL;
}
/* make sure the DH parameters are usable */
if (dh != NULL)
{
if (DH_check(dh, &codes))
{
return NULL;
}
if (codes & DH_CHECK_P_NOT_PRIME)
{
return NULL;
}
if ((codes & DH_NOT_SUITABLE_GENERATOR) &&
(codes & DH_CHECK_P_NOT_SAFE_PRIME))
{
return NULL;
}
}
return dh;
}
/*
* Load hardcoded DH parameters.
*
* To prevent problems if the DH parameters files don't even
* exist, we can load DH parameters hardcoded into this file.
*/
static DH *
load_dh_buffer (const char *buffer, size_t len)
{
BIO *bio;
DH *dh = NULL;
bio = BIO_new_mem_buf((char *) buffer, len);
if (bio == NULL)
return NULL;
dh = PEM_read_bio_DHparams(bio, NULL, NULL, NULL);
BIO_free(bio);
return dh;
}
/*
* Generate an empheral DH key. Because this can take a long
* time to compute, we can use precomputed parameters of the
* common key sizes.
*
* Since few sites will bother to precompute these parameter
* files, we also provide a fallback to the parameters provided
* by the OpenSSL project.
*
* These values can be static (once loaded or computed) since
* the OpenSSL library can efficiently generate random keys from
* the information provided.
*/
static DH *
tmp_dh_cb (SSL *s, int is_export, int keylength)
{
DH *r = NULL;
static DH *dh = NULL;
static DH *dh512 = NULL;
static DH *dh1024 = NULL;
static DH *dh2048 = NULL;
static DH *dh4096 = NULL;
switch (keylength)
{
case 512:
if (dh512 == NULL)
dh512 = load_dh_file(keylength);
if (dh512 == NULL)
dh512 = load_dh_buffer(file_dh512, sizeof file_dh512);
r = dh512;
break;
case 1024:
if (dh1024 == NULL)
dh1024 = load_dh_file(keylength);
if (dh1024 == NULL)
dh1024 = load_dh_buffer(file_dh1024, sizeof file_dh1024);
r = dh1024;
break;
case 2048:
if (dh2048 == NULL)
dh2048 = load_dh_file(keylength);
if (dh2048 == NULL)
dh2048 = load_dh_buffer(file_dh2048, sizeof file_dh2048);
r = dh2048;
break;
case 4096:
if (dh4096 == NULL)
dh4096 = load_dh_file(keylength);
if (dh4096 == NULL)
dh4096 = load_dh_buffer(file_dh4096, sizeof file_dh4096);
r = dh4096;
break;
default:
if (dh == NULL)
dh = load_dh_file(keylength);
r = dh;
}
/* this may take a long time, but it may be necessary... */
if (r == NULL || 8*DH_size(r) < keylength)
{
r = DH_generate_parameters(keylength, DH_GENERATOR_2, NULL, NULL);
}
return r;
}
/*
* Callback used by SSL to load client cert and key.
* This callback is only called when the server wants a
* client cert.
*
* Returns 1 on success, 0 on no data, -1 on error.
*/
static int
client_cert_cb (SSL *ssl, X509 **x509, EVP_PKEY **pkey)
{
struct passwd *pwd;
struct stat buf, buf2;
char fnbuf[2048];
FILE *fp;
PGconn *conn = (PGconn *) SSL_get_app_data(ssl);
int (*cb)() = NULL; /* how to read user password */
if ((pwd = getpwuid(getuid())) == NULL)
{
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("unable to get user information\n"));
return -1;
}
/* read the user certificate */
snprintf(fnbuf, sizeof fnbuf, "%s/.postgresql/postgresql.crt",
pwd->pw_dir);
if (stat(fnbuf, &buf) == -1)
return 0;
if ((fp = fopen(fnbuf, "r")) == NULL)
{
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("unable to open certificate (%s): %s\n"),
fnbuf, strerror(errno));
return -1;
}
if (PEM_read_X509(fp, x509, NULL, NULL) == NULL)
{
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("unable to read certificate (%s): %s\n"),
fnbuf, SSLerrmessage());
fclose(fp);
return -1;
}
fclose(fp);
/* read the user key */
snprintf(fnbuf, sizeof fnbuf, "%s/.postgresql/postgresql.key",
pwd->pw_dir);
if (stat(fnbuf, &buf) == -1)
{
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("certificate present, but not private key (%s)\n"),
fnbuf);
X509_free(*x509);
return 0;
}
if (!S_ISREG(buf.st_mode) || (buf.st_mode & 0077) ||
buf.st_uid != getuid())
{
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("private key has bad permissions (%s)\n"), fnbuf);
X509_free(*x509);
return -1;
}
if ((fp = fopen(fnbuf, "r")) == NULL)
{
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("unable to open private key file (%s): %s\n"),
fnbuf, strerror(errno));
X509_free(*x509);
return -1;
}
if (fstat(fileno(fp), &buf2) == -1 ||
buf.st_dev != buf2.st_dev || buf.st_ino != buf2.st_ino)
{
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("private key changed under us (%s)\n"), fnbuf);
X509_free(*x509);
return -1;
}
if (PEM_read_PrivateKey(fp, pkey, cb, NULL) == NULL)
{
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("unable to read private key (%s): %s\n"),
fnbuf, SSLerrmessage());
X509_free(*x509);
fclose(fp);
return -1;
}
fclose(fp);
/* verify that the cert and key go together */
if (!X509_check_private_key(*x509, *pkey))
{
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("certificate/private key mismatch (%s): %s\n"),
fnbuf, SSLerrmessage());
X509_free(*x509);
EVP_PKEY_free(*pkey);
return -1;
}
return 1;
}
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
/*
* Initialize global SSL context.
*/
static int
initialize_SSL (PGconn *conn)
{
struct stat buf;
struct passwd *pwd;
char fnbuf[2048];
if (!SSL_context)
{
SSL_library_init();
SSL_load_error_strings();
SSL_context = SSL_CTX_new(TLSv1_method());
if (!SSL_context)
{
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("could not create SSL context: %s\n"),
SSLerrmessage());
return -1;
}
}
if ((pwd = getpwuid(getuid())) != NULL)
{
snprintf(fnbuf, sizeof fnbuf, "%s/.postgresql/root.crt",
pwd->pw_dir);
if (stat(fnbuf, &buf) == -1)
{
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("could not read root cert list(%s): %s"),
fnbuf, strerror(errno));
return -1;
}
if (!SSL_CTX_load_verify_locations(SSL_context, fnbuf, 0))
{
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("could not read root cert list (%s): %s"),
fnbuf, SSLerrmessage());
return -1;
}
}
SSL_CTX_set_verify(SSL_context,
SSL_VERIFY_PEER | SSL_VERIFY_FAIL_IF_NO_PEER_CERT, verify_cb);
SSL_CTX_set_verify_depth(SSL_context, 1);
/* set up empheral DH keys */
SSL_CTX_set_tmp_dh_callback(SSL_context, tmp_dh_cb);
SSL_CTX_set_options(SSL_context, SSL_OP_SINGLE_DH_USE);
/* set up mechanism to provide client certificate, if available */
SSL_CTX_set_client_cert_cb(SSL_context, client_cert_cb);
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
return 0;
}
/*
* Destroy global SSL context.
*/
static void
destroy_SSL (void)
{
if (SSL_context)
{
SSL_CTX_free(SSL_context);
SSL_context = NULL;
}
}
/*
* Attempt to negotiate SSL connection.
*/
static int
open_client_SSL (PGconn *conn)
{
int r;
if (!(conn->ssl = SSL_new(SSL_context)) ||
!SSL_set_app_data(conn->ssl, conn) ||
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
!SSL_set_fd(conn->ssl, conn->sock) ||
SSL_connect(conn->ssl) <= 0)
{
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("could not establish SSL connection: %s\n"),
SSLerrmessage());
close_SSL(conn);
return -1;
}
/* check the certificate chain of the server */
/* this eliminates simple man-in-the-middle attacks and
* simple impersonations */
r = SSL_get_verify_result(conn->ssl);
if (r != X509_V_OK)
{
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("certificate could not be validated: %s\n"),
X509_verify_cert_error_string(r));
close_SSL(conn);
return -1;
}
/* pull out server distinguished and common names */
conn->peer = SSL_get_peer_certificate(conn->ssl);
if (conn->peer == NULL)
{
printfPQExpBuffer(&conn->errorMessage,
libpq_gettext("certificate could not be obtained: %s\n"),
SSLerrmessage());
close_SSL(conn);
return -1;
}
X509_NAME_oneline(X509_get_subject_name(conn->peer),
conn->peer_dn, sizeof(conn->peer_dn));
conn->peer_dn[sizeof(conn->peer_dn)-1] = '\0';
X509_NAME_get_text_by_NID(X509_get_subject_name(conn->peer),
NID_commonName, conn->peer_cn, SM_USER);
conn->peer_cn[SM_USER] = '\0';
/* verify that the common name resolves to peer */
/* this is necessary to eliminate man-in-the-middle attacks
* and impersonations where the attacker somehow learned
* the server's private key */
if (verify_peer(conn) == -1)
{
close_SSL(conn);
return -1;
}
return 0;
}
/*
* Close SSL connection.
*/
static void
close_SSL (PGconn *conn)
{
if (conn->ssl)
{
SSL_shutdown(conn->ssl);
SSL_free(conn->ssl);
conn->ssl = NULL;
}
}
/*
* Obtain reason string for last SSL error
*
* Some caution is needed here since ERR_reason_error_string will
* return NULL if it doesn't recognize the error code. We don't
* want to return NULL ever.
*/
static const char *
SSLerrmessage(void)
{
unsigned long errcode;
const char *errreason;
static char errbuf[32];
errcode = ERR_get_error();
if (errcode == 0)
return "No SSL error reported";
errreason = ERR_reason_error_string(errcode);
if (errreason != NULL)
return errreason;
snprintf(errbuf, sizeof(errbuf), "SSL error code %lu", errcode);
return errbuf;
}
/*
* Return pointer to SSL object.
*/
SSL *
PQgetssl(PGconn *conn)
{
if (!conn)
return NULL;
return conn->ssl;
}
UPDATED PATCH: Attached are a revised set of SSL patches. Many of these patches are motivated by security concerns, it's not just bug fixes. The key differences (from stock 7.2.1) are: *) almost all code that directly uses the OpenSSL library is in two new files, src/interfaces/libpq/fe-ssl.c src/backend/postmaster/be-ssl.c in the long run, it would be nice to merge these two files. *) the legacy code to read and write network data have been encapsulated into read_SSL() and write_SSL(). These functions should probably be renamed - they handle both SSL and non-SSL cases. the remaining code should eliminate the problems identified earlier, albeit not very cleanly. *) both front- and back-ends will send a SSL shutdown via the new close_SSL() function. This is necessary for sessions to work properly. (Sessions are not yet fully supported, but by cleanly closing the SSL connection instead of just sending a TCP FIN packet other SSL tools will be much happier.) *) The client certificate and key are now expected in a subdirectory of the user's home directory. Specifically, - the directory .postgresql must be owned by the user, and allow no access by 'group' or 'other.' - the file .postgresql/postgresql.crt must be a regular file owned by the user. - the file .postgresql/postgresql.key must be a regular file owned by the user, and allow no access by 'group' or 'other'. At the current time encrypted private keys are not supported. There should also be a way to support multiple client certs/keys. *) the front-end performs minimal validation of the back-end cert. Self-signed certs are permitted, but the common name *must* match the hostname used by the front-end. (The cert itself should always use a fully qualified domain name (FDQN) in its common name field.) This means that psql -h eris db will fail, but psql -h eris.example.com db will succeed. At the current time this must be an exact match; future patches may support any FQDN that resolves to the address returned by getpeername(2). Another common "problem" is expiring certs. For now, it may be a good idea to use a very-long-lived self-signed cert. As a compile-time option, the front-end can specify a file containing valid root certificates, but it is not yet required. *) the back-end performs minimal validation of the client cert. It allows self-signed certs. It checks for expiration. It supports a compile-time option specifying a file containing valid root certificates. *) both front- and back-ends default to TLSv1, not SSLv3/SSLv2. *) both front- and back-ends support DSA keys. DSA keys are moderately more expensive on startup, but many people consider them preferable than RSA keys. (E.g., SSH2 prefers DSA keys.) *) if /dev/urandom exists, both client and server will read 16k of randomization data from it. *) the server can read empheral DH parameters from the files $DataDir/dh512.pem $DataDir/dh1024.pem $DataDir/dh2048.pem $DataDir/dh4096.pem if none are provided, the server will default to hardcoded parameter files provided by the OpenSSL project. Remaining tasks: *) the select() clauses need to be revisited - the SSL abstraction layer may need to absorb more of the current code to avoid rare deadlock conditions. This also touches on a true solution to the pg_eof() problem. *) the SIGPIPE signal handler may need to be revisited. *) support encrypted private keys. *) sessions are not yet fully supported. (SSL sessions can span multiple "connections," and allow the client and server to avoid costly renegotiations.) *) makecert - a script that creates back-end certs. *) pgkeygen - a tool that creates front-end certs. *) the whole protocol issue, SASL, etc. *) certs are fully validated - valid root certs must be available. This is a hassle, but it means that you *can* trust the identity of the server. *) the client library can handle hardcoded root certificates, to avoid the need to copy these files. *) host name of server cert must resolve to IP address, or be a recognized alias. This is more liberal than the previous iteration. *) the number of bytes transferred is tracked, and the session key is periodically renegotiated. *) basic cert generation scripts (mkcert.sh, pgkeygen.sh). The configuration files have reasonable defaults for each type of use. Bear Giles
2002-06-14 06:23:17 +02:00
#endif /* USE_SSL */