If there is not any objection I will commit this patch to trunk. >From what I can see this patch solves the bug 2976 too. Am I correct? Should I close the bug 2976 too? (and probably commit the patch with --fixes flag)
On 06/14/2012 02:47 AM, Amos Jeffries wrote: > On 14.06.2012 09:01, Tsantilas Christos wrote: >> Here is a patch which implements Alex proposal. >> >> Regards, >> Christos >> >> On 06/08/2012 05:52 PM, Alex Rousskov wrote: >>> On 06/08/2012 06:00 AM, Amos Jeffries wrote: >>>> On 8/06/2012 9:23 p.m., Alexander Komyagin wrote: >>>>> Hello Alex! >>>>> >>> >>> If they are cbdata, let's just lock/release them properly, without ever >>> freeing/deleting them explicitly. My understanding is that the following >>> is necessary to fix the underlying problem: >>> >>> 1. add_http_port() needs to lock the new port pointer before linking it. >>> 2. parsePortCfg() needs to lock the new port pointer before linking it. >>> 3. free_PortCfg() needs to unlock the old port pointer before unlinking >>> it. It should not delete the old pointer. >>> >>> Anything else? >>> >>> >>> Thank you, >>> >>> Alex. >>> > > +1. looks okay. if it runs please apply. > > NP: Committer please also remove the comment "BUG 2976: Squid only > accepts intercepted HTTP." in prepareTransparentURL() > > > Amos >
