While browsing some https web sites, some of them are forums, mozilla loads the page incorrectly, placing icon's in strange places, showing the wrong icon in others,, misplacing buttons... It's kind of hard to explain so I took a picture of the same page loaded correctly and incorrectly. One more thing, the error is kind of random, sometimes it loads the page well.. sometimes its a little messed up, and some other times it's simply undreadable (loading just one icon.. and weird things like that) also, I dropped the optimization flags to -0 to see if the problem persisted .. and it did... also, I have moznocomposer and moznoirc. Reproducible: Sometimes Steps to Reproduce: 1.emerge mozilla-1.3-r1 2.go to some https page, (the ones i'm using are php generated) 3. Actual Results: messed up web pages Expected Results: good web pages [69p] quark : ~ $ emerge info Portage 2.0.47-r10 (default-x86-1.4, gcc-3.2.2, glibc-2.3.1-r4) ================================================================= System uname: 2.4.20-ck6 i686 AMD Athlon(TM) MP 1500+ GENTOO_MIRRORS="ftp://ftp.rnl.ist.utl.pt/gentoo http://gentoo.oregonstate.edu/ h ttp://www.ibiblio.org/pub/Linux/distributions/gentoo" CONFIG_PROTECT="/etc /var/qmail/control /usr/share/config /usr/kde/2/share/confi g /usr/kde/3/share/config /usr/X11R6/lib/X11/xkb" CONFIG_PROTECT_MASK="/etc/gconf /etc/env.d" PORTDIR="/usr/portage" DISTDIR="/usr/portage/distfiles" PKGDIR="/usr/portage/packages" PORTAGE_TMPDIR="/var/tmp" PORTDIR_OVERLAY="" USE="x86 gdbm tcltk tcpd esd qt 3dnow acpi cdr gpm mmx sse kerberos ssl ipv6 lda p afs pam crypt threads java junit python perl libwww zlib berkdb doc readline s lang quicktime oggvorbis mpeg avi gif png jpeg tiff sdl oss alsa xmms mozilla mo znoirc moznocompose truetype ncurses nls X xv dga opengl svga motif pdflib gtk g tk2 imlib" COMPILER="gcc3" CHOST="i686-pc-linux-gnu" CFLAGS="-march=athlon-mp -O2 -pipe -fomit-frame-pointer" CXXFLAGS="-march=athlon-mp -O2 -pipe -fomit-frame-pointer" ACCEPT_KEYWORDS="x86" MAKEOPTS="-j5" AUTOCLEAN="yes" SYNC="rsync://rsync.gentoo.org/gentoo-portage" FEATURES="sandbox ccache userpriv usersandbox"
here are the "bad" and "good" pages <bad> http://mega.ist.utl.pt/~mmsf/mozilla-bug.jpeg <good> http://mega.ist.utl.pt/~mmsf/mozilla-nobug.jpeg
I can confirm that there is some strange https problem with mozilla 1.3: My internet banking access worked perfectly with mozilla 1.2 and fails with 1.3: * It reproducibly fails to open the application in a new window: The window opens as before, but it remains blank (shows "about:blank") without any error messages and without any further activity: Mozilla just "forgets" to load the javascript application into the newly opened window. In this case, reload does not help! * It randomly drops actions in the main page: Clicks to buttons or links are silently ignored in about half of all attempts: The displayed page does not change, nothing is loaded, no error message, "idle" indication in the upper right corner and the cursor after a very short indication of activity (hourglass just for a barely noticeable moment). * In some (but rare) random cases, it also shows incomplete pages. Moreover, my firewall clearly indicates that there is a problem with https, each failed action causes one or more of the following entries: May 27 16:18:27 kklaptop kernel: IN=ppp0 OUT= MAC= SRC=193.110.28.65 DST=81.223. 1.206 LEN=63 TOS=0x00 PREC=0x00 TTL=56 ID=39963 PROTO=TCP SPT=443 DPT=57221 WIND OW=24616 RES=0x00 ACK PSH FIN URGP=0 I verified that SRC is the banking system, DST is me. It looks like the banking system sends a packet to me when my system already considers the https connection closed.
Downgraded to 1.2.x again, everything fine. This is definitely a 1.3 problem.
Still a (the same?) problem with mozilla-1.4-r3.
I found two reasons for my problems. See 151443 in mozilla's bugzilla, my comments at the end.
shoudn't this be considered nonvalid or something, were in mozilla 1.4 now... right?
Both problems are also valid for 1.5.
Has any of this been investigated upstream?
It has turned out that these are two problems: * Connections are mysteriously broken with HTTP/1.1 and pipelining - obviously, the other side can't handle that. * The other half of the problem is in Mozilla bug nr 219990. With pipelining turned off and the profile variable described in 219990 reset, everything works fine (however, that variable is still a mystery to me).
Good to know.
