[Bug 260116] [sctp] POLLOUT/EVFILT_WRITE is always true for poll/kqueue

2022-02-08 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=260116 Michael Tuexen changed: What|Removed |Added Assignee|b...@freebsd.org|tue...@freebsd.org -- You are re

[Bug 260116] [sctp] POLLOUT/EVFILT_WRITE is always true for poll/kqueue

2022-02-08 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=260116 --- Comment #9 from Michael Tuexen --- (In reply to Björn Svensson from comment #8) I think for 1-to-1 style sockets it should work and it worked (I think) until the socket layer in FreeBSD was changed. I'll look into the patch soon. -- Y

[Bug 260116] [sctp] POLLOUT/EVFILT_WRITE is always true for poll/kqueue

2022-02-08 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=260116 --- Comment #8 from Björn Svensson --- I have noted that the non-blocking section (https://datatracker.ietf.org/doc/html/rfc6458#section-3.2) mention that an application should use caution when writing to a 1-to-many style socket and using

[Bug 260116] [sctp] POLLOUT/EVFILT_WRITE is always true for poll/kqueue

2022-02-08 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=260116 Björn Svensson changed: What|Removed |Added Attachment #231307|0 |1 is obsolete|

[Bug 260116] [sctp] POLLOUT/EVFILT_WRITE is always true for poll/kqueue

2022-01-25 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=260116 --- Comment #6 from Björn Svensson --- (In reply to Michael Tuexen from comment #5) I'm trying to fix the problem that a sctp socket is not indicating that the sendbuffer is full when polling the socket, or via kernel event notifications.

[Bug 260116] [sctp] POLLOUT/EVFILT_WRITE is always true for poll/kqueue

2022-01-25 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=260116 --- Comment #5 from Michael Tuexen --- (In reply to Björn Svensson from comment #3) Can you state what you are trying to fix? Are you trying to fix sendfile() for SCTP or the kqueue write stuff or something else? -- You are receiving this

[Bug 260116] [sctp] POLLOUT/EVFILT_WRITE is always true for poll/kqueue

2022-01-25 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=260116 --- Comment #4 from Michael Tuexen --- (In reply to Björn Svensson from comment #3) Can you state what you are trying to fix? Are you trying to fix sendfile() for SCTP or the kqueue write stuff or something else? -- You are receiving this

[Bug 260116] [sctp] POLLOUT/EVFILT_WRITE is always true for poll/kqueue

2022-01-25 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=260116 --- Comment #3 from Björn Svensson --- Created attachment 231307 --> https://bugs.freebsd.org/bugzilla/attachment.cgi?id=231307&action=edit Proposed patch This proposed patch adds a macro for unifying the sb_acc and sb_ccc counters after

[Bug 260116] [sctp] POLLOUT/EVFILT_WRITE is always true for poll/kqueue

2022-01-24 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=260116 --- Comment #2 from Björn Svensson --- After some investigations we have seen that the introduction of "sendfile(2) system call' required some changes in the struct sockbuf. The field sb_cc was split into sb_ccc and sb_acc in commit: 0f9d0a

[Bug 260116] [sctp] POLLOUT/EVFILT_WRITE is always true for poll/kqueue

2021-11-29 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=260116 --- Comment #1 from Albin --- Created attachment 229791 --> https://bugs.freebsd.org/bugzilla/attachment.cgi?id=229791&action=edit Test files for reproducing the bug I added some tests if it helps. They can be compiled using: g++ -o clie

[Bug 260116] [sctp] POLLOUT/EVFILT_WRITE is always true for poll/kqueue

2021-11-29 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=260116 Bug ID: 260116 Summary: [sctp] POLLOUT/EVFILT_WRITE is always true for poll/kqueue Product: Base System Version: 13.0-RELEASE Hardware: Any OS: Any