On 06/28/2011 10:01 PM, Michael Scheidell wrote: > > > On 6/28/11 2:49 PM, Török Edwin wrote: >> If you still have one of those hung 0.97.1 (or come across in the future) >> can you run >> $ gcore<pid> >> >> This should generate a core file of the hung process that can be >> investigated later >> (as long as you still have the coresponding clamd and libclamav binaries >> too). >> >> Best regards, >> --Edwin > > yes, but darn clients start whining when they can't get their facebook > invites immediately :-) > > trying to reproduce in lab with daily and safebrowsing files, also, trying > with the emails that made it hang. > > left some backup systems still running 0.97.1 in the field.. if a backup > system fails, we can leave it hung till we figure out what happened. > > so, WHAT that might affect threading did you change from 0.97 to 0.97.1? > also, this affected boxes running TCP socket, as well as running unix sock. > > and, only on amd64 boxes.. all the i386/32bix boxes are fine. > >
Maybe this one: http://git.clamav.net/gitweb?p=clamav-devel.git;a=commitdiff;h=bbfe830c935837cfc357541cb307a7b863394abb;hp=d9ff9e65080d7c70de722e174d365d3b2cb312d3 But it survived a full regression test here (Linux/amd64). Will go through the code again tomorrow and see if I can make it deadlock. Best regards, --Edwin _______________________________________________ Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net http://www.clamav.net/support/ml
