diff options
author | Eric S. Raymond <esr@thyrsus.com> | 2005-07-05 13:14:55 +0000 |
---|---|---|
committer | Eric S. Raymond <esr@thyrsus.com> | 2005-07-05 13:14:55 +0000 |
commit | be8dbaedd20cf289d47ca6f2b5295c97a611489f (patch) | |
tree | a5656f13bc6dd1bd562c30b7b09406a1d5da7df2 | |
parent | 3ce4cf8f1526bc468e3d2b4aaf0765feb41e10ae (diff) | |
download | gpsd-be8dbaedd20cf289d47ca6f2b5295c97a611489f.tar.gz |
FIONREAD can't spin anymore, because FIONREAD is gone.
-rw-r--r-- | TODO | 17 |
1 files changed, 0 insertions, 17 deletions
@@ -67,23 +67,6 @@ But Rob's xgps memory leak doesn't reproduce on a stock Fedora Core 3 system. ESR tested for this in the simplest possible way, by doing system("free t") at the end of each handle_input() call. -*** Under Reiserfs gpsd may conspire with I/O buffering traffic to eat the CPU. - -Rob Janssen writes: -> When gpsd is running and I start a backup of my system, which starts and -> mounts a normally idle disk and does a tar cvzf to it, the load of the -> system is quite high and gpsd seems to get out of sync with the received -> data from the SiRF receiver. -> It then gets stuck in a tight loop. As gpsd is running at nice -> --10, this further increases the load and slows down the backup to a crawl. -> Killing gpsd makes the backup finish and then it can be started again. - -He further observers that the FIONREAD ioctl() we use to check for -input waiting spins returning 0 when this happens. We suspect that -memory resource crises do something nasty to the buffering in the -Linux serial layer. The bug seems not to occur during ext3 backups, -only Reiserfs ones. - ** To do: *** Hotplug interface problems |