[core,gateway] use TcpConnectTimeout for read as well #10288
Conversation
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
|
Refer to this link for build results (access rights to CI server needed): |
akallabeth
force-pushed
the
http_read_timeout
branch
from
e64195b
to
fc9f41f
Compare
|
clang-tidy review says "All clean, LGTM! 👍" |
|
Refer to this link for build results (access rights to CI server needed): |
akallabeth
force-pushed
the
http_read_timeout
branch
from
fc9f41f
to
44705b9
Compare
|
clang-tidy review says "All clean, LGTM! 👍" |
|
Refer to this link for build results (access rights to CI server needed): |
|
@freerdp-bot test please |
|
Refer to this link for build results (access rights to CI server needed): |
|
@freerdp-bot test please |
|
Refer to this link for build results (access rights to CI server needed): |
akallabeth
force-pushed
the
http_read_timeout
branch
from
44705b9
to
13aa14f
Compare
|
clang-tidy review says "All clean, LGTM! 👍" |
|
Refer to this link for build results (access rights to CI server needed): |
|
@freerdp-bot test please |
|
Refer to this link for build results (access rights to CI server needed): |
|
@freerdp-bot test please |
|
Refer to this link for build results (access rights to CI server needed): |
|
Hello @akallabeth. Thank you for the pull request. Unfortunately, the changes refuse to let me connect. FreeRDP asks for a redirect URL, and after I pass it on, I get Details |
|
@progxaker ok, so it works as intended ;) |
|
But the AVD instance was off, and I wanted to wait for the auto-reconnect mechanism to start it up automatically. Instead, FreeRDP drops the connection after 15 seconds, i.e., it doesn't wait for the instance is up. |
|
@progxaker hmm, for that there should be a response from the remote end, see your previous logs with entries like |
|
Hello @akallabeth. I don't think the timeout parameter helps here. I set it to 20000ms ( LogsUPD: yes, it looks like. DetailsI added the diff --git a/libfreerdp/core/gateway/http.c b/libfreerdp/core/gateway/http.c
index dce527804..f4ecbed17 100644
--- a/libfreerdp/core/gateway/http.c
+++ b/libfreerdp/core/gateway/http.c
@@ -1240,6 +1240,7 @@ HttpResponse* http_response_recv(rdpTls* tls, BOOL readContentLength)
/* Read until we encounter \r\n\r\n */
ERR_clear_error();
const int wstatus = BIO_wait_read(tls->bio, timeout);
+ WLog_ERR(TAG, "[BIO_wait_read] wstatus: %i", wstatus);
if (wstatus < 0)
{
if (!BIO_should_retry(tls->bio))and got P.S. This is also not related to the reconnection mechanism, because in the second attempt (with the log function) AVD was started (by first FreeRDP run). |
http_response_recv might never receive an answer, so do wait for input and fail if the timeout is exceeded.
akallabeth
force-pushed
the
http_read_timeout
branch
from
13aa14f
to
4425772
Compare
|
@progxaker ok, there was indeed a bug with the timeout handling in the custom |
|
clang-tidy review says "All clean, LGTM! 👍" |
|
Refer to this link for build results (access rights to CI server needed): |
BIO_C_WAIT_READ and BIO_C_WAIT_WRITE now set errno = ETIMEDOUT if a timeout occurred.
akallabeth
force-pushed
the
http_read_timeout
branch
from
4425772
to
c4fe6e7
Compare
|
clang-tidy review says "All clean, LGTM! 👍" |
|
Refer to this link for build results (access rights to CI server needed): |
http_response_recv might never receive an answer, so do wait for input and fail if the timeout is exceeded.
@progxaker you might test this branch and it should now at least fail with a timeout instead of idling forever