nihilist/tor
1
0
Fork 0
mirror of https://gitlab.torproject.org/tpo/core/tor.git synced 2024-11-27 22:03:31 +01:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'cell_comment' into 'main'

Browse Source 
Correct a comment on channel_should_write_to_kernel.

See merge request tpo/core/tor!757
This commit is contained in:
David Goulet 2023-09-12 14:31:08 +00:00
commit 9f2d3e7750
1 changed files with 10 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
src/core/or/scheduler_kist.c
View File

@ -447,10 +447,16 @@ update_socket_written(socket_table_t *table, channel_t *chan, size_t bytes)
* by only writing a channel's outbuf to the kernel if it has 8 cells or more * by only writing a channel's outbuf to the kernel if it has 8 cells or more
* in it. * in it.
* *
* Note: The number 8 has been picked for no particular reasons except that it * Note: The number 8 was picked so that, when using 512-byte cells, it
* is 4096 bytes which is a common number for buffering. A TLS record can hold * would produce 4096 bytes: a common number for buffering. A TLS
* up to 16KiB thus using 8 cells means that a relay will at most send a TLS * record can hold up to 16KiB; thus, using 8 512-byte cells means that
* record of 4KiB or 1/4 of the maximum capacity of a TLS record. * a relay will at most send a TLS record of 4KiB or 1/4 of the maximum
* capacity of a TLS record.
*
* Of course, the above calculation became incorrect when we moved to
* 514-byte cells in order to accommodate a 4-byte circuit ID; we may
* want to consider profiling with '7' to see if it produces better
* results. (TODO)
*/ */
MOCK_IMPL(int, channel_should_write_to_kernel, MOCK_IMPL(int, channel_should_write_to_kernel,
(outbuf_table_t *table, channel_t *chan)) (outbuf_table_t *table, channel_t *chan))