-
Notifications
You must be signed in to change notification settings - Fork 4.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[release/9.0-staging] fix TCP FastOpen compilation #111142
base: release/9.0-staging
Are you sure you want to change the base?
[release/9.0-staging] fix TCP FastOpen compilation #111142
Conversation
Tagging subscribers to this area: @dotnet/ncl |
#if defined(__linux__) && !defined(TCP_FASTOPEN_CONNECT) | ||
// fixup if compiled against old Kernel headers. | ||
// Can be removed once we have at least 4.11 | ||
#define TCP_FASTOPEN_CONNECT 30 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We're relying on getsockopt below failing when this is defined but not understood?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
yes. If the call fails e.g. not supported by given Kernel we would fall-back to traditional connect and data would be sent after handshake.
Impacts new 9.0 feature functionality for important customer, we should backport it. |
Backport of #110561 to release/9.0-staging
Fixes #110560
/cc @wfurt
Customer Impact
Requested by Oracle for ODP.NET. TCP FastOpen does not work on some Linux distros, causing functionality problem in their routing environments.
TCP FastOpen is a new feature in 9.0 and it does not work on all supported platforms.
The problem is caused by inconsistency of used kernel headers .NET builds against - some Docker images have old ones, causing the problem.
Regression
Not a regression. New 9.0 feature does not work on all platforms.
Testing
Tested on manual repro case.
There is no good way to automate tests of the functionality, because the negotiation depends on Kernel TCP cookie cache as well as there is no good way how to determine on receiving side if the data were sent with SYN packet or later.
Risk
Small. this is build problem and it gets glibc binaries on par with musl.