Re: gvfsd-smb-browse caused high cpu use

[Ondrej Holy <oholy redhat com>]

Hello,

maybe the address comes from DNS server, see following bug report:
https://bugzilla.redhat.com/show_bug.cgi?id=1303300

File chooser causes network backend automount. Network backend mounts smbbrowse backend. Server address is 
based on d-conf org.gnome.system.smb.workgroup value. Default workgroup from smbc_getWorkgroup is used if the 
d-conf value is "X-GNOME-DEFAULT-WORKGROUP". It seems that server hostname is resolved "properly" by DNS 
server and smbc_opendir hangs consequently, because server is not responding. But I afraid that we can't do 
much from GVfs, it seems to me this is bug in samba.

Maybe smbc_setTimeout might help to decrease the time of hang a bit at least...

Regards

Ondrej

----- Original Message -----
> Thanks for reply.
> One more thing, the 'save as' like dialog must be opened by firefox.
> I run the strace with -f parameter ,the most suspicous item is below :
>
> 1586 connect(12, {sa_family=AF_INET, sin_port=htons(445),
> sin_addr=inet_addr("172.27.0.222")}, 16) = -1 EALREADY (Operation already in
> progress)
> 1586 fcntl(12, F_SETFL, O_RDWR) = 0
> 1586 fcntl(12, F_GETFL) = 0x2 (flags O_RDWR)
> 1586 fcntl(12, F_GETFL) = 0x2 (flags O_RDWR)
> 1586 fcntl(12, F_SETFL, O_RDWR|O_NONBLOCK) = 0
>
> 172.27.0.222 is a LAN address, and I dont know why gvfs-smb-browse is trying
> to access this address.
> I grep this address via below command:
> grep '172.27.0.222' / etc / *
> which returned nothing to me ... so I dont know which configuration file (if
> there is one) contain this string ~
>
> On 02/17/2016 03:36 PM, Ross Lagerwall wrote:
>
>
>
> On Wed, Feb 17, 2016 at 01:15:55PM +0800, Hacksign wrote:
>
>
>
> Archlinux+Systemd
> awesomewm + thunar + gvfs-smb-browser.
> when thunar opened the 'save as' dialog(or any dialog like save
> as),gvfs-smb-browser  causes high cup load.
> And the high cpu only appear in the first time opened 'save as' like
> dialog.This situation wont appear if I killed the process and open the 'save
> as' dialog again .
> anyone meet the same situation and solved it ?]
>
> Below is the strace log when high cpu usage appears,it is stuck on
> restart_syscall...
>
> |[root XSign hacksign]# strace -p 22560 Process 22560 attached
> restart_syscall(<... resuming interrupted call ...>^CProcess 22560 detached
> <detached ...>|
> I haven't seen this before.  If you pass -f to strace, it will allow you
> to trace all threads -- it is likely that some other thread is doing
> something. Another approach if you're willing, is to reinstall gvfs with
> debuginfo, attach to the process with gdb (e.g. gdb -p 22560) and then
> run "thread apply all bt" a few times to get a snapshot of what is
> happening.
>
>
> _______________________________________________
> gvfs-list mailing list
> gvfs-list gnome org
> https://mail.gnome.org/mailman/listinfo/gvfs-list