svn:r1588

the connect attempt fails immediately, we were trying to
double-remove it from the dns pending list.

i think this might have been The Bug.
(thanks weasel!)


svn:r1587

(it has been haunting us since the days of 0.0.2preX)


svn:r1576

svn:r1522

svn:r1365

but leave dirservers just as likely to crash


svn:r1363

(but only enable this once we've found the bug)


svn:r1358

somewhere


svn:r1354

this was causing the seg faults on sparc processors

i wonder if i got them all.


svn:r1314

svn:r1267

svn:r1266

svn:r1260

it's simply not true


svn:r1236

and make them say it at log level info rather than err


svn:r1232

svn:r1231

svn:r1230

svn:r1169

resolve in response to a DNS worker dying) and connection_dns_remove (remove
a pending connection from the resolve structure.)


svn:r1168

svn:r1167

svn:r1166

svn:r1165

typically the resolve will be pending, but it could be failed if it
just failed and we're working through the 'pending' list, or it could
be valid if some other dnsworker raced us to fill in the entry.


svn:r1164

Push responsibility for connection marking down as far as possible; have only a close path; add some missing end cells; change return conventions a little.


svn:r1149

who wants to shut down a connection calls connection_mark_for_close instead
of setting marked_for_close to 1.  This automatically removes the connection
from the DNS cache if needed, sends a RELAY END cell if appropriate, and can
be changed to do whatever else is needed.

Still to do:
  - The same for circuits, maybe.
  - Add some kind of hold_connection_open_until_flushed flag, maybe.
  - Change stuff that closes connections with return -1 to use mark_for_close,
    maybe.


svn:r1145

corruption faster; also, check for corruption in dns.c so we can fail fast
for the bug that's nailing Lucky and moria3.


svn:r1123

svn:r1121

svn:r1118

svn:r1116

we're not entirely migrated to burst yet, for backward compatibility

note some win32 probable-bugs

clean up routerlist.c


svn:r982

We were telling a child to die by closing the parent's file descriptor
to him. But newer children were inheriting the open file descriptor from
the parent, and since they weren't closing them, the socket never closed,
so the child never read eof, so he never knew to exit.

As a side effect to this bug, we were probably failing to properly close
connections to remote hosts, ORs, and OPs, after a dns child was born.

I'm surprised Tor worked at all.


svn:r974

svn:r949

svn:r936

svn:r932

svn:r929

svn:r919

svn:r915

svn:r837

change RecommendedVersions into a config option, so dirservers can hup
  for a new one


svn:r809

make exit dns cache entries last 15 minutes


svn:r807

svn:r778