nihilist/tor
1
0
Fork 0
mirror of https://gitlab.torproject.org/tpo/core/tor.git synced 2024-12-12 13:43:36 +01:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'maint-0.3.2' into release-0.3.2

Browse Source
This commit is contained in:
Nick Mathewson 2017-12-21 10:20:35 -05:00
commit 2a7cad9572
2 changed files with 17 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
changes/bug24665 Normal file
View File

@ -0,0 +1,6 @@
o Major bugfixes (KIST, scheduler):
- The KIST scheduler did not correctly account for data already enqueued
in each connection's send socket buffer, particularly in cases when the
TCP/IP congestion window was reduced between scheduler calls. This
situation lead to excessive per-connection buffering in the kernel, and
a potential memory DoS. Fixes bug 24665; bugfix on 0.3.2.1-alpha.

22
src/or/scheduler_kist.c
View File

@ -266,8 +266,7 @@ update_socket_info_impl, (socket_table_ent_t *ent))
/* These values from the kernel are uint32_t, they will always fit into a
* int64_t tcp_space variable but if the congestion window cwnd is smaller
* than the unacked packets, the remaining TCP space is set to 0 so we don't
* write more on this channel. */
* than the unacked packets, the remaining TCP space is set to 0. */
if (ent->cwnd >= ent->unacked) {
tcp_space = (ent->cwnd - ent->unacked) * (int64_t)(ent->mss);
} else {
@ -276,20 +275,21 @@ update_socket_info_impl, (socket_table_ent_t *ent))
/* The clamp_double_to_int64 makes sure the first part fits into an int64_t.
* In fact, if sock_buf_size_factor is still forced to be >= 0 in config.c,
* then it will be positive for sure. Then we subtract a uint32_t. At worst
* we end up negative, but then we just set extra_space to 0 in the sanity
* check.*/
* then it will be positive for sure. Then we subtract a uint32_t. Getting a
* negative value is OK, see after how it is being handled. */
extra_space =
clamp_double_to_int64(
(ent->cwnd * (int64_t)ent->mss) * sock_buf_size_factor) -
ent->notsent;
if (extra_space < 0) {
extra_space = 0;
}
/* Finally we set the limit. Adding two positive int64_t together will always
* fit in an uint64_t. */
if ((tcp_space + extra_space) < 0) {
/* This means that the "notsent" queue is just too big so we shouldn't put
* more in the kernel for now. */
ent->limit = 0;
} else {
/* Adding two positive int64_t together will always fit in an uint64_t.
* And we know this will always be positive. */
ent->limit = (uint64_t)tcp_space + (uint64_t)extra_space;
}
return;
#else /* !(defined(HAVE_KIST_SUPPORT)) */