ports/161737: www/chromium 14.0.835.x stalls (probably javascript-related

[Conrad J. Sabatier]

On Tue, 10 Jan 2012 21:10:17 GMT
rene at freebsd.org wrote:

> Synopsis: www/chromium 14.0.835.x stalls (probably javascript-related
> 
> State-Changed-From-To: open->closed
> State-Changed-By: rene
> State-Changed-When: Tue Jan 10 21:09:22 UTC 2012
> State-Changed-Why: 
> Submitter reports that the original problem is solved.
> 
> http://www.freebsd.org/cgi/query-pr.cgi?pr=161737

Any chance of us seeing just what exactly the solution was?  I'm
still seeing of lot of the same type of behavior described by the
original submitter -- pages stalling interminably or loading only
partially (while presenting the impression of having completed),
unresponsive buttons, etc.

Chrome was fast as greased lightning when I first started using it, but
lately I'm often resorting to using firefox or epiphany due to this
extreme sluggishness, which is very disappointing, since I've really
come to feel "at home" with chrome over time and it has become my
preferred browser.

I'm wondering if there's some other, external factor at play here,
perhaps something in the (10.0) kernel's networking that's contributing
to these extreme delays and failures to properly and completely load
pages.

I build chrome with clang, by the way.

Chromium 16.0.912.77 (Developer Build 0)
OS	FreeBSD
WebKit	535.7 (Unknown URL at 0)
JavaScript	V8 3.6.6.19
Flash	11.1 r102
User Agent	Mozilla/5.0 (X11; FreeBSD amd64) AppleWebKit/535.7
(KHTML, like Gecko) Chrome/16.0.912.77 Safari/535.7 Command
Line	 chrome --flag-switches-begin --downloads-new-ui
--enable-autologin --enable-smooth-scrolling --force-compositing-mode
--enable-accelerated-2d-canvas --ignore-gpu-blacklist
--preload-instant-search --flag-switches-end
https://sourceforge.net/tracker/?func=detail&atid=989750&aid=3480153&group_id=204472
Executable Path	/usr/home/conrads/Mail/inbox/chrome Profile
Path	/usr/home/conrads/.config/chromium/Default

The only non-default settings of any possible relevance I have in
sysctl.conf (and have been using for a very long time, so shouldn't
really be a factor) are:

#
# general kernel options
#

kern.ipc.somaxconn=256
kern.maxfiles=24576
kern.maxfilesperproc=16384

#
# networking options
#

# tcp options

net.inet.tcp.blackhole=2
net.inet.tcp.recvspace=131072
net.inet.tcp.sendspace=65536

# udp options

net.inet.udp.blackhole=1

#
# miscellaneous
#

vfs.usermount=1
vm.pmap.shpgperproc=1024

And loader tunables:

kern.maxswzone="100663296"	# (was seeing frequent "out of swap
				# space" messages until I modified this
				# setting)
net.inet.tcp.syncache.bucketlimit="128"
net.inet.tcp.syncache.cachelimit="32768"

-- 
Conrad J. Sabatier
conrads at cox.net