NSS currently uses the default Curl_none_data_pending() method which
always returns false, causing TLS buffered input data to be missed.
The current commit implements the nss_data_pending() method that properly
monitors the presence of available TLS data.
Ref:#10077
Closes#10225
SecureTransport expects result code errSSLWouldBlock when the requested
length could not be sent/recieved in full. The previous code returned
noErr, which let SecureTransport to believe that the IO had terminated
prematurely.
Fixes#10227Closes#10235
- by using BIT() instead of bool
- imap: shrink struct
- ftp: make state 'unsigned char'
- ftp: sort ftp_conn struct entries on size
- urldata: use smaller fields for SSL version info storage
- pop3: reduce the pop3_conn struct size
- smtp: reduce the size of the smtp structs
Closes#10186
- they are mostly pointless in all major jurisdictions
- many big corporations and projects already don't use them
- saves us from pointless churn
- git keeps history for us
- the year range is kept in COPYING
checksrc is updated to allow non-year using copyright statements
Closes#10205
The only TLS auth type libcurl ever supported is SRP and that is the
default type. Since nobody ever sets any other type, there is no point
in wasting space to store the set type and code to check the type.
If TLS auth is used, SRP is now implied.
Closes#10181
Refactoring of connection setup and happy eyeballing. Move
nghttp2. ngtcp2, quiche and msh3 into connection filters.
- eyeballing cfilter that uses sub-filters for performing parallel connects
- socket cfilter for all transport types, including QUIC
- QUIC implementations in cfilter, can now participate in eyeballing
- connection setup is more dynamic in order to adapt to what filter did
really connect. Relevant to see if a SSL filter needs to be added or
if SSL has already been provided
- HTTP/3 test cases similar to HTTP/2
- multiuse of parallel transfers for HTTP/3, tested for ngtcp2 and quiche
- Fix for data attach/detach in VTLS filters that could lead to crashes
during parallel transfers.
- Eliminating setup() methods in cfilters, no longer needed.
- Improving Curl_conn_is_alive() to replace Curl_connalive() and
integrated ssl alive checks into cfilter.
- Adding CF_CNTRL_CONN_INFO_UPDATE to tell filters to update
connection into and persist it at the easy handle.
- Several more cfilter related cleanups and moves:
- stream_weigth and dependency info is now wrapped in struct
Curl_data_priority
- Curl_data_priority members depend is available in HTTP2|HTTP3
- Curl_data_priority members depend on NGHTTP2 support
- handling init/reset/cleanup of priority part of url.c
- data->state.priority same struct, but shallow copy for compares only
- PROTOPT_STREAM has been removed
- Curl_conn_is_mulitplex() now available to check on capability
- Adding query method to connection filters.
- ngtcp2+quiche: implementing query for max concurrent transfers.
- Adding is_alive and keep_alive cfilter methods. Adding DATA_SETUP event.
- setting keepalive timestamp on connect
- DATA_SETUP is called after the connection has been completely
setup (but may not connected yet) to allow filters to initialize
data members they use.
- there is no socket to be had with msh3, it is unclear how select
shall work
- manual test via "curl --http3 https://curl.se" fail with "empty
reply from server".
- Various socket/conn related cleanups:
- Curl_socket is now Curl_socket_open and in cf-socket.c
- Curl_closesocket is now Curl_socket_close and in cf-socket.c
- Curl_ssl_use has been replaced with Cur_conn_is_ssl
- Curl_conn_tcp_accepted_set has been split into
Curl_conn_tcp_listen_set and Curl_conn_tcp_accepted_set
with a clearer purpose
Closes#10141
- keeping the "current" easy handle registered at SSL* is no longer
necessary, since the "calling" data object is already stored in the
cfilter's context (and used by other SSL backends from there).
- The "detach" of an easy handle that goes out of scope is then avoided.
- using SSL_set0_wbio for clear reference counting where available.
Closes#10151
When utf8asn1str fails there is no allocation returned, so freeing
the return pointer in **to is at best a no-op and at worst a double-
free bug waiting to happen. The current coding isn't hiding any such
bugs but to future proof, avoid freeing the return value pointer iff
the function failed.
Closes: #10087
Reviewed-by: Daniel Stenberg <daniel@haxx.se>
- fixes `Curl_ssl_cf_get_ssl()` to detect also the first filter instance
as ssl (refs #10053)
- replaces `Curl_ssl_use()` with the correct `Curl_conn_is_ssl()`
Closes#10054Fixes#10053
Reported-by: Patrick Monnerat
BIO_read and BIO_write return negative numbers on error, including
retryable ones. A regression from 55807e6. Both branches should be
returning -1.
The APIs are patterned after POSIX read and write which, similarly,
return -1 on errors, not zero, with EAGAIN treated as an error.
Bug: https://github.com/curl/curl/issues/10013#issuecomment-1335308146
Reported-by: David Benjamin
Closes#10021
- `Curl_ssl_get_config()` now returns the first config if no SSL proxy
filter is active
- socket filter starts connection only on first invocation of its
connect method
Fixes#9982Closes#9983
In non-IPv6 builds the conn parameter is unused, and compilers which
run with "-Werror=unused-parameter" (or similar) warnings turned on
fails to build. Below is an excerpt from a CI job:
vtls/openssl.c: In function ‘Curl_ossl_verifyhost’:
vtls/openssl.c:2016:75: error: unused parameter ‘conn’ [-Werror=unused-parameter]
2016 | CURLcode Curl_ossl_verifyhost(struct Curl_easy *data, struct connectdata *conn,
| ~~~~~~~~~~~~~~~~~~~~^~~~
Closes: #9970
Reviewed-by: Daniel Stenberg <daniel@haxx.se>
- almost all backend calls pass the Curl_cfilter intance instead of
connectdata+sockindex
- ssl_connect_data is remove from struct connectdata and made internal
to vtls
- ssl_connect_data is allocated in the added filter, kept at cf->ctx
- added function to let a ssl filter access its ssl_primary_config and
ssl_config_data this selects the propert subfields in conn and data,
for filters added as plain or proxy
- adjusted all backends to use the changed api
- adjusted all backends to access config data via the exposed
functions, no longer using conn or data directly
cfilter renames for clear purpose:
- methods `Curl_conn_*(data, conn, sockindex)` work on the complete
filter chain at `sockindex` and connection `conn`.
- methods `Curl_cf_*(cf, ...)` work on a specific Curl_cfilter
instance.
- methods `Curl_conn_cf()` work on/with filter instances at a
connection.
- rebased and resolved some naming conflicts
- hostname validation (und session lookup) on SECONDARY use the same
name as on FIRST (again).
new debug macros and removing connectdata from function signatures where not
needed.
adapting schannel for new Curl_read_plain paramter.
Closes#9919
- Adding Curl_conn_is_ip_connected() to check if network connectivity
has been reached
- having ftp wait for network connectivity before proceeding with
transfers.
Fixes test failures 1631 and 1632 with hyper.
Closes#9952
Prior to this change Curl_read_plain would attempt to read the
socket directly. On Windows that's a problem because recv data may be
cached by libcurl and that data is only drained using Curl_recv_plain.
Rather than rewrite Curl_read_plain to handle cached recv data, I
changed it to wrap Curl_recv_plain, in much the same way that
Curl_write_plain already wraps Curl_send_plain.
Curl_read_plain -> Curl_recv_plain
Curl_write_plain -> Curl_send_plain
This fixes a bug in the schannel backend where decryption of arbitrary
TLS records fails because cached recv data is never drained. We send
data (TLS records formed by Schannel) using Curl_write_plain, which
calls Curl_send_plain, and that may do a recv-before-send
("pre-receive") to cache received data. The code calls Curl_read_plain
to read data (TLS records from the server), which prior to this change
did not call Curl_recv_plain and therefore cached recv data wasn't
retrieved, resulting in malformed TLS records and decryption failure
(SEC_E_DECRYPT_FAILURE).
The bug has only been observed during Schannel TLS 1.3 handshakes. Refer
to the issue and PR for more information.
--
This is take 2 of the original fix. It preserves the original behavior
of Curl_read_plain to write 0 to the bytes read parameter on error,
since apparently some callers expect that (SOCKS tests were hanging).
The original fix which landed in 12e1def5 and was later reverted in
18383fbf failed to work properly because it did not do that.
Also, it changes Curl_write_plain the same way to complement
Curl_read_plain, and it changes Curl_send_plain to return -1 instead of
0 on CURLE_AGAIN to complement Curl_recv_plain.
Behavior on error with these changes:
Curl_recv_plain returns -1 and *code receives error code.
Curl_send_plain returns -1 and *code receives error code.
Curl_read_plain returns error code and *n (bytes read) receives 0.
Curl_write_plain returns error code and *written receives 0.
--
Ref: https://github.com/curl/curl/issues/9431#issuecomment-1312420361
Assisted-by: Joel Depooter
Reported-by: Egor Pugin
Fixes https://github.com/curl/curl/issues/9431
Closes https://github.com/curl/curl/pull/9949
Prior to this change Curl_read_plain would attempt to read the
socket directly. On Windows that's a problem because recv data may be
cached by libcurl and that data is only drained using Curl_recv_plain.
Rather than rewrite Curl_read_plain to handle cached recv data, I
changed it to wrap Curl_recv_plain, in much the same way that
Curl_write_plain already wraps Curl_send_plain.
Curl_read_plain -> Curl_recv_plain
Curl_write_plain -> Curl_send_plain
This fixes a bug in the schannel backend where decryption of arbitrary
TLS records fails because cached recv data is never drained. We send
data (TLS records formed by Schannel) using Curl_write_plain, which
calls Curl_send_plain, and that may do a recv-before-send
("pre-receive") to cache received data. The code calls Curl_read_plain
to read data (TLS records from the server), which prior to this change
did not call Curl_recv_plain and therefore cached recv data wasn't
retrieved, resulting in malformed TLS records and decryption failure
(SEC_E_DECRYPT_FAILURE).
The bug has only been observed during Schannel TLS 1.3 handshakes. Refer
to the issue and PR for more information.
Ref: https://github.com/curl/curl/issues/9431#issuecomment-1312420361
Assisted-by: Joel Depooter
Reported-by: Egor Pugin
Fixes https://github.com/curl/curl/issues/9431
Closes https://github.com/curl/curl/pull/9904
- general construct/destroy in connectdata
- default implementations of callback functions
- connect: cfilters for connect and accept
- socks: cfilter for socks proxying
- http_proxy: cfilter for http proxy tunneling
- vtls: cfilters for primary and proxy ssl
- change in general handling of data/conn
- Curl_cfilter_setup() sets up filter chain based on data settings,
if none are installed by the protocol handler setup
- Curl_cfilter_connect() boot straps filters into `connected` status,
used by handlers and multi to reach further stages
- Curl_cfilter_is_connected() to check if a conn is connected,
e.g. all filters have done their work
- Curl_cfilter_get_select_socks() gets the sockets and READ/WRITE
indicators for multi select to work
- Curl_cfilter_data_pending() asks filters if the have incoming
data pending for recv
- Curl_cfilter_recv()/Curl_cfilter_send are the general callbacks
installed in conn->recv/conn->send for io handling
- Curl_cfilter_attach_data()/Curl_cfilter_detach_data() inform filters
and addition/removal of a `data` from their connection
- adding vtl functions to prevent use of Curl_ssl globals directly
in other parts of the code.
Reviewed-by: Daniel Stenberg
Closes#9855
Adds a new option to control the maximum time that a cached
certificate store may be retained for.
Currently only the OpenSSL backend implements support for
caching certificate stores.
Closes#9620
This patch aims to cleanup the use of `process.h` header and the macro
`HAVE_PROCESS_H` associated with it.
- `process.h` is always available on Windows. In curl, it is required
only for `_beginthreadex()` in `lib/curl_threads.c`.
- `process.h` is also available in MS-DOS. In curl, its only use was in
`lib/smb.c` for `getpid()`. But `getpid()` is in fact declared by
`unistd.h`, which is always enabled via `lib/config-dos.h`. So the
header is not necessary.
- `HAVE_PROCESS_H` was detected by CMake, forced to 1 on Windows and
left to real detection for other platforms.
It was also set to always-on in `lib/config-win32.h` and
`lib/config-dos.h`.
In autotools builds, there was no detection and the macro was never
set.
Based on these observations, in this patch we:
- Rework Windows `getpid` logic in `lib/smb.c` to always use the
equivalent direct Win32 API function `GetCurrentProcessId()`, as we
already did for Windows UWP apps. This makes `process.h` unnecessary
here on Windows.
- Stop #including `process.h` into files where it was not necessary.
This is everywhere, except `lib/curl_threads.c`.
> Strangely enough, `lib/curl_threads.c` compiled fine with autotools
> because `process.h` is also indirecty included via `unistd.h`. This
> might have been broken in autotools MSVC builds, where the latter
> header is missing.
- Delete all remaining `HAVE_PROCESS_H` feature guards, for they were
unnecessary.
- Delete `HAVE_PROCESS_H` detection from CMake and predefined values
from `lib/config-*.h` headers.
Reviewed-by: Jay Satiro
Closes#9703
No point in having two entry points for the same functions.
Also merged the *safe* function treatment into these so that they can
also be used when one or both pointers are NULL.
Closes#9837
Even though `PICKY_COMPILER=ON` is the default, warnings were not
enabled when using llvm/clang, because `CMAKE_COMPILER_IS_CLANG` was
always false (in my tests at least).
This is the single use of this variable in curl, and in a different
place we already use `CMAKE_C_COMPILER_ID MATCHES "Clang"`, which works
as expected, so change the condition to use that instead.
Also fix the warnings uncovered by the above:
- lib: add casts to silence clang warnings
- schannel: add casts to silence clang warnings in ALPN code
Assuming the code is correct, solve the warnings with a cast.
This particular build case isn't CI tested.
There is a chance the warning is relevant for some platforms, perhaps
Windows 32-bit ARM7.
Closes#9783
These function pointers will have been set when the initial TLS
handshake was completed. If they are unchanged, there is no need to set
them again. If they have been changed, as is the case with HTTP/2, we
don't want to override that change. That would result in the
http22_recv/send functions being completely bypassed.
Prior to this change a connection that uses Schannel with HTTP/2 would
fail on renegotiation with error "Received HTTP/0.9 when not allowed".
Fixes https://github.com/curl/curl/issues/9451
Closes https://github.com/curl/curl/pull/9756
By default, the PFXImportCertStore API persists the key in the user's
key store (as though the certificate was being imported for permanent,
ongoing use.)
The documentation specifies that keys that are not to be persisted
should be imported with the flag PKCS12_NO_PERSIST_KEY.
NOTE: this flag is only supported on versions of Windows newer than XP
and Server 2003.
--
This is take 2 of the original fix. It extends the lifetime of the
client certificate store to that of the credential handle. The original
fix which landed in 70d010d and was later reverted in aec8d30 failed to
work properly because it did not do that.
Minor changes were made to the schannel credential context to support
closing the client certificate store handle at the end of an SSL session.
--
Reported-by: ShadowZzj@users.noreply.github.com
Fixes https://github.com/curl/curl/issues/9300
Supersedes https://github.com/curl/curl/pull/9363
Closes https://github.com/curl/curl/pull/9460
This is a strcmp() alternative function for comparing "secrets",
designed to take the same time no matter the content to not leak
match/non-match info to observers based on how fast it is.
The time this function takes is only a function of the shortest input
string.
Reported-by: Trail of Bits
Closes#9658
This function is currently located in the lib/http.c module and is
therefore disabled by the CURL_DISABLE_HTTP conditional token.
As it may be called by TLS backends, disabling HTTP results in an
undefined reference error at link time.
Move this function to vauth/vauth.c to always provide it and rename it
as Curl_auth_allowed_to_host() to respect the vauth module naming
convention.
Closes#9600
Curl_ossl_certchain() is now an exported function in lib/vtls/openssl.c that
can also be used from quiche.c and ngtcp2.c to get the cert chain for QUIC
connections as well.
The *certchain function was moved to the top of the file for this reason.
Reported-by: Eloy Degen
Fixes#9584Closes#9597
Prior to this commit, non-persistent pointers were being used to store
sessions. When a WOLFSSL object was then freed, that freed the session
it owned, and thus invalidated the pointer held in curl's cache. This
commit makes it so we get a persistent (deep copied) session pointer
that we then add to the cache. Accordingly, wolfssl_session_free, which
was previously a no-op, now needs to actually call SSL_SESSION_free.
This bug was discovered by a wolfSSL customer.
Closes#9492
Instances of ISSPACE() use that should rather use ISBLANK(). I think
somewhat carelessly used because it sounds as if it checks for space or
whitespace, but also includes %0a to %0d.
For parsing purposes, we should only accept what we must and not be
overly liberal. It leads to surprises and surprises lead to bad things.
Closes#9432
