Summary: | Fixes for ftp/wu-ftpd 2.6.2 | ||
---|---|---|---|
Product: | Ports & Packages | Reporter: | Rudolf Čejka <cejkar> |
Component: | Individual Port(s) | Assignee: | Andrey A. Chernov <ache> |
Status: | Closed FIXED | ||
Severity: | Affects Only Me | ||
Priority: | Normal | ||
Version: | Latest | ||
Hardware: | Any | ||
OS: | Any |
Description
Rudolf Čejka
2002-10-13 18:50:01 UTC
Responsible Changed From-To: freebsd-ports->ache over to maintainer On Sun, Oct 13, 2002 at 19:43:18 +0200, Cejka Rudolf wrote: > > patch-ra: This is not very good patch and there should be a better fix, > but I use it and it works for me with full satisfaction - problems with > SIGPIPE on sockets: If somebody downloads a file and connection is broken > (you can use kill -9 ftp_client), partial download is not logged into > xferlog and information about partially downloaded file is lost. Authors > of wu-ftpd said to me, that on Linux it works, so maybe broken connection > on Linux does not generate SIGPIPE and maybe we can use "Implement > SO_NOSIGPIPE option for sockets." since June 2002 instead of my old patch. We can't use some hack without full understanding of what really happens there. Do you try NEED_SIGFIX define for SIGPIPE? > patch-rb: NCARGS (= ARG_MAX = 65536) is very small on FreeBSD and it > is used in wu-ftpd in place, where it is not needed. If you use file > completion in ftp client (NLIST command on server) in some big directory, > for example in FreeBSD/distfiles with > 100 KB of filenames, wu-ftpd > daemon in this listing fails. This patch increases NCARGS by using > another macro from 65536 to 524288 - I hope it is sufficient for most cases. From where you get this number? > patch-rd - patch-rk: It is very easy to download more than MAX_INT > bytes now... So extend session statistics to long long and when client > is exiting session, goodbye message with session statistics is not > overflowed. %qd is depricated. Use %lld instead to print long long. Printing (unsigned)off_t as off_t could mislead, it should be printed as (unsighed)off_t too under %u (I mean malloc failed stuff). -- Andrey A. Chernov http://ache.pp.ru/ State Changed From-To: open->feedback ache@ requested feedback. Give submitter last chance to respond before timing out or suspending? State Changed From-To: feedback->closed [From private discussion which didn't make it into GNATS:] Won't add local features to FreeBSD port. Sorry it didn't work out. |