+------ Chuck Swiger wrote (Wed, 28-May-2008, 15:27 -0700):
| On May 28, 2008, at 7:08 AM, Nigel Horne wrote:
| > We welcome any feedback and bugs on this RC prior to the release
| > of 0.93.1, which is currently scheduled for 6th June. It doesn't
| > matter
| > if you don't have a test environment, you can still help us for
| > example by
| > downloading the release candidate and checking it compiles on your
| > system even if you don't install it; we particularly welcome reports
| > on platform compatibility.
|
| 0.91.1RC1 build and seems to run fine under FreeBSD 5.5, 6.3, and
| Solaris 8. <...>
|
| On FreeBSD 4.11, the compile fails with:
Using the patch:
https://wwws.clamav.net/bugzilla/attachment.cgi?id=730
and the fix committed for r3852 as mentioned in this bug report:
https://wwws.clamav.net/bugzilla/show_bug.cgi?id=1039
0.93.1rc1 builds on FreeBSD 4.11 (i386) using the system gcc 2.95.4,
but clamd appears to be broken. The clamd process seems to start OK,
creating a socket, and pid and log files. But, communication with
clamdscan over the socket doesn't work. Runs of clamscan succeed,
as expected, so the libraries likely are OK.
The 0.92.1 clamd works on this system, but neither 0.93 nor 0.93.1rc1 do.
Running strace on the 0.92.1 and 0.93.1rc1 shows a difference in
system calls: 0.92.1 uses poll(2) while 0.93.1rc1 uses accept(2).
clamdscan works fine with 0.92.1 clamd, but just hangs with 0.93.1rc1.
The strace of the 0.93.1rc1 clamd looks like this when clamdscan
is started:
accept(6, 0, NULL) = 7
fcntl(7, F_GETFL) = 0x2 (flags O_RDWR)
fcntl(7, F_SETFL, O_RDWR|O_NONBLOCK) = 0
gettimeofday({1212382127, 673446}, NULL) = 0
break(0xb1c4000) = 0
setitimer(ITIMER_PROF, {it_interval={0, 10000}, it_value={0, 10000}},
NULL) = 0
mmap(0xbfaef000, 65536, PROT_READ|PROT_WRITE, MAP_STACK, -1, 0) =
0xbfaef000
gettimeofday({1212382127, 675265}, NULL) = 0
fstat(6, {st_mode=S_IFSOCK|0666, st_size=0, ...}) = 0
accept(6,
The log file generated during clamd starupt looks OK. FWIW, this
version of clamd does not shutdown on a TERM signal. It does accept
QUIT and ABRT sigs, though without cleaning up its socket and pid file.
The same behavior is evident when the 0.92.1 clamdscan is used with
the 0.93.1rc1 clamd.
Likely I'll be better served by spending time working on the replacement
for this old, old, inherited server, rather than on sorting this out.
Still, if anyone can provide guidance for further diagnosis, feel free...
Best Regards,
Chuck
_______________________________________________
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://lurker.clamav.net/list/clamav-users.html
