gvfsd-http eats 100% CPU

[Matteo Dell'Amico]

Every now and then, I find that gvfsd-http is eating 100% CPU; I have to kill it manually. By having a look with strace I see an endless loop of

poll([{fd=3, events=POLLIN}], 1, 0) = 0 (Timeout)
recvfrom(18, "", 4, 0, NULL, NULL)

I wonder whether this is happening because I set up a SOCKS proxy, and something similar to what is described in https://bugs.launchpad.net/ubuntu/+source/telepathy-gabble/+bug/1094032 is happening.

ProblemType: Bug
DistroRelease: Ubuntu 13.04
Package: gvfs-backends 1.16.1-0ubuntu1
ProcVersionSignature: Ubuntu 3.8.0-19.29-generic 3.8.8
Uname: Linux 3.8.0-19-generic x86_64
NonfreeKernelModules: wl
ApportVersion: 2.9.2-0ubuntu8
Architecture: amd64
Date: Tue Apr 30 10:53:24 2013
EcryptfsInUse: Yes
InstallationDate: Installed on 2012-07-23 (280 days ago)
InstallationMedia: Ubuntu 12.04 LTS "Precise Pangolin" - Release amd64 (20120425)
MarkForUpload: True
SourcePackage: gvfs
UpgradeStatus: Upgraded to raring on 2013-04-29 (0 days ago)

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Christian Kirbach (christian-kirbach-e)]

I can see this as well. netstat shows lots of connections kept in CLOSE_WAIT state.

adding a backtrace at some random point I hooked into the process.

I hooked in a couple of times and it seems to always or most often work in g_main_context_find_source_by_user_data()

My strace on gvfsd-http looks like

futex(0x166aa70, FUTEX_WAKE_PRIVATE, 1) = 1
futex(0x166aa70, FUTEX_WAKE_PRIVATE, 1) = 1
brk(0x36b3000) = 0x36b3000
futex(0x166aa70, FUTEX_WAKE_PRIVATE, 1) = 1
futex(0x166aa70, FUTEX_WAKE_PRIVATE, 1) = 1
futex(0x166aa70, FUTEX_WAKE_PRIVATE, 1) = 1
futex(0x166aa70, FUTEX_WAKE_PRIVATE, 1) = 1
futex(0x166aa70, FUTEX_WAKE_PRIVATE, 1) = 1
futex(0x166aa70, FUTEX_WAKE_PRIVATE, 1) = 1
brk(0x36d4000) = 0x36d4000
futex(0x166aa70, FUTEX_WAKE_PRIVATE, 1) = 1
futex(0x166aa70, FUTEX_WAKE_PRIVATE, 1) = 1
futex(0x166aa70, FUTEX_WAIT_PRIVATE, 2, NULL) = -1 EAGAIN (Resource temporarily unavailable)
futex(0x166aa70, FUTEX_WAKE_PRIVATE, 1) = 0
futex(0x166aa70, FUTEX_WAKE_PRIVATE, 1) = 1
futex(0x166aa70, FUTEX_WAKE_PRIVATE, 1) = 1
brk(0x36f5000) = 0x36f5000
brk(0x36f4000) = 0x36f4000
futex(0x166aa70, FUTEX_WAKE_PRIVATE, 1) = 1
futex(0x166aa70, FUTEX_WAKE_PRIVATE, 1) = 1
futex(0x166aa70, FUTEX_WAKE_PRIVATE, 1) = 1
brk(0x3715000) = 0x3715000
futex(0x166aa70, FUTEX_WAKE_PRIVATE, 1) = 1

[Mehdi (mehdi-ti)]

I use socks proxy now and then, and I have encounter the exact problem several times. I can confirm the bug, yet I'm not sure if it's occurring only when using socks proxy.

I'm getting 100% cpu usage from the following process:

/usr/lib/gvfs/gvfsd-http --spawner :1.7 /org/gtk/gvfs/exec_spaw/2

the trace for the process outputs the following in an "endless loop":

[pid 7705] poll([{fd=3, events=POLLIN}], 1, 0) = 0 (Timeout)
[pid 7705] recvfrom(9, "", 4, 0, NULL, NULL) = 0

[Pierre van Male (vmalep)]

Dear all,

Recently, I noticed that I was experiencing the same problem (Ubuntu 13.04).

I use to create an SSH tunnel to a personal server (ssh -D ...) and configure a system-wide proxy via this tunnel.

I will remove the proxy for the coming days and see if the problem persists.

Best regards,
Pierre