Share common functions/structs between ngtcp2 HTTP/3 and the proxy
version.
Fix bugs in proxy implementation when it comes to stream and pollset
handling and transfer lifetimes.
Curl_multi_xfer_sockbuf_borrow: work without multi
When a connection gets shutdown by a share, the easy handle used is
share->admin and it does not have a multi handle. In that case let
Curl_multi_xfer_sockbuf_borrow() allocate a buffer to be freed on
release.
This happens when a TLS filter sends its last notify through a HTTP/3
proxy tunnel.
Closes#21871
This fixes a misleading log in verbose mode when ipv6 connectivity isn't
available, presumably also in other cases:
```
* Immediate connect fail for 2a00:1450:4028:806::200e: Network is unreachable
* connect to 2a00:1450:4028:806::200e port 443 from :: port 0 failed: Success
```
Closes#21914
If a SECURE cookie is set for a sub-domain (`example.com`) and is then
attempted to get set again for more specific part of that domain
(`www.example.com`) without the SECURE property, the second occurance
should not be allowed.
Reported-by: Trail of Bits
Verified by test 3305
Closes#21910
Do not send PONG frames unless there is sufficient space left in the
websocket send buffer. A server might be lazy in reading our data and
intermediary PONG frames can be skipped by a client (RFC 6455, ch.
5.5.3).
Add test case measuring no real RSS increase on a server blasting with
PING frames.
Closes#21911
Since the username is decoded when used and control codes are accepted
in HTTP usernames in general, the username encoding for the Digest auth
needs to percent encode such bytes.
Verified by test 3221
Reported-by: Trail of Bits
Closes#21915
With this, IPv6 addresses that end with '%25' with no following zone id are
considered invalid.
Extend test 1560 to verify
Reported-by: Hem Parekh
Closes#21918
Trailers are delivered to the application as headers via
CLIENTWRITE_TRAILER, but unlike regular response headers they skipped
the verify_header() checks, so a server could smuggle a nul byte (or
stray CR) into a header reaching CURLOPT_HEADERFUNCTION and
curl_easy_header().
Run each assembled trailer line through Curl_verify_header(), the same
validation used for normal headers.
Covered by the new test 2106.
Closes#21896
Add bit `native_ca_store_opt` to keep the setting of
CURLOPT_(PROXY_)SSL_OPTIONS and use that to calculate every easy
transfer if a native CA store shall be used or not.
This avoids `native_ca_store` getting stuck on TRUE after being set
once.
Closes#21902
This reverts commit 3e9817cd1b.
The change was incorrect as the check was not for the options the commit
message mentions.
Reported-by: ByteRay on hackerone
Closes#21899
Whack the times reported for a transfer (see
https://curl.se/libcurl/c/curl_easy_getinfo.html#TIMES) into order for
all variations of up-/download, http/ftp etc. Make sure they are
reported in the documented order.
There is still the *possibility* of PRETRANSFER being longer then
POSTTRANSFER, if a server sends a response before an upload is done.
POST is the time the first response byte is received, and PRE is the
time the last byte was sent by curl.
This may happen with more likelihood on HTTP/2 and 3 for a server
rejected upload. But for successful uploads, the answer will almost over
come afterwards.
Undo the previous twists in lib500.c tests, adjust pytest timeline
checks.
Fixes#21828
Reported-by: BazaarAcc32 on github
Closes#21843
- Specify that the content is base64 encoded, rather than rely on
auto-detect.
- Remove unnecessary sanity check of the returned content type.
Closes https://github.com/curl/curl/pull/21760
Supported since libssh2 v1.2.5. Replacing `libssh2_knownhost_add()`,
which was deprecated in that same version.
The new API supports a comment field.
Ref: https://github.com/libssh2/libssh2/pull/1977Closes#21866
With a flood of zero lenght UDP packets to curl, the receive loop might
run longer than intended to. Count such packets against the max to
terminate the loop as intended.
URL: https://hackerone.com/reports/3783438
Reported-by: vectorqueue on hackerone
Closes#21869
- fix GnuTLS function name reference in `Curl_gtls_shared_creds_create()`
error message.
Spotted by GitHub Code Quality.
- unfold a line.
- in `Curl_gtls_verifyserver()`:
- report the failure of `gnutls_x509_crt_import()`.
Spotted by GitHub Code Quality.
- fix a minor inconsistency in error strings.
- drop redundant NULL checks for `config->issuercert`.
Closes#21850
Since aeb1a281ca ("gtls: fix OCSP stapling management"), the function
parses the stapled OCSP response and reads the certificate status via
gnutls_ocsp_resp_get_single(), but never calls gnutls_ocsp_resp_verify()
or gnutls_ocsp_resp_verify_direct(). A response with a forged or
corrupted signature is accepted without question.
Fix by calling gnutls_ocsp_resp_verify() against the trust list obtained
from the session credentials immediately after gnutls_ocsp_resp_import().
This handles both directly-signed responses and delegated OCSP responders
without requiring the issuer certificate to be present in the peer chain.
The missing check only affects the CURLOPT_SSL_VERIFYSTATUS code path
when CURLOPT_SSL_VERIFYPEER is disabled. With peer verification enabled,
gnutls_certificate_verify_peers2() independently catches the invalid
response via GNUTLS_CERT_INVALID_OCSP_STATUS before
gtls_verify_ocsp_status() is reached. As a result, no attack is possible
that is not already trivially achievable without OCSP stapling when peer
verification is off. This is a correctness and consistency fix, not a
security vulnerability.
Reported-by: Joshua Rogers
Closes#21677
The poll-before-write loop used -1 (infinite) as the Curl_poll timeout,
so a peer that stops reading could stall the transfer indefinitely,
bypassing CURLOPT_TIMEOUT. Use Curl_timeleft_ms() instead and return
CURLE_OPERATION_TIMEDOUT when the deadline is reached or exceeded.
Closes#21685
closing a filter chain and reconnecting it again is a complication
that only the HTTP/1.x proxy filter used. Remove it from all filters.
Instead, a filter can return CURLE_AGAIN during the connect phase
and the cf-setup filter will tear down all "sub filters" and restart
over.
With this, a filter never resets to the initial phase but progresses
through connect -> connected -> shutdown -> destroy once.
Closes#21831
- drop more uses of the word "just". (not enforced here)
- drop some uses of the "will" word.
- "then" -> "than".
- tests/http/testenv/curl.py: fix copy-paste typo in error message.
- pytest: replace `shutdownh` with `shutdown` in test names.
Spotted by GitHub Code Quality.
- comment typos.
- whitespace and newlines fixes.
Closes#21830
Reshuffle code a little to detect when the transfer's credentials
actually change. Otherwise, leave the existing creds in place.
This gives the precise location where we may want to reset other
states that become invalid with change credentials. Also, by
keeping a creds instance as long as it is valid, we can associate
meta data with it.
Closes#21755
