SerializeObject failed with error code 0x80000004 for object

jengelh

that’s because readelf for some stupid reason truncates the names by default, even if they’re in the last column (one explicitly needs -W), so a trivial grep would not find it. nm is better suited for that.

martimcfly

Badly not…

[root@server lib]# nm libkcserver.so
nm: libkcserver.so: no symbols

[root@serverlib lib]# readelf -w libkcserver.so
Contents of the .eh_frame section:

00000000 ZERO terminator


[root@server lib]#

It didn’t matter if I did this…

./bootstrap.sh
CPPFLAGS="-I/usr/include/python2.7" CXXFLAGS="-O0 -ggdb3" ./configure \
  --prefix=/ \
  --localstatedir=/var \
  --sysconfdir=/etc \
  --exec-prefix=/usr \
  --sbindir=/usr/bin \
  --datarootdir=/usr/share \
  --includedir=/usr/include \
  --enable-release \
  --enable-epoll \
  --enable-python \
  --disable-debug \
  --disable-static \
  --with-userscript-prefix=/etc/kopano/userscripts \
  --with-quotatemplate-prefix=/etc/kopano/quotamails \
  --with-searchscripts-prefix=/etc/kopano/searchscripts \
  --with-php=7 \
  PYTHON=/usr/bin/python2

make
make install DESTDIR="$(realpath ${pkgdir})"

or this

./bootstrap.sh
CPPFLAGS="-I/usr/include/python2.7" ./configure \
 CXXFLAGS="-O0 -ggdb3" \
  --prefix=/ \
  --localstatedir=/var \
  --sysconfdir=/etc \
  --exec-prefix=/usr \
  --sbindir=/usr/bin \
  --datarootdir=/usr/share \
  --includedir=/usr/include \
  --enable-release \
  --enable-epoll \
  --enable-python \
  --disable-debug \
  --disable-static \
  --with-userscript-prefix=/etc/kopano/userscripts \
  --with-quotatemplate-prefix=/etc/kopano/quotamails \
  --with-searchscripts-prefix=/etc/kopano/searchscripts \
  --with-php=7 \
  PYTHON=/usr/bin/python2

make
make install DESTDIR="$(realpath ${pkgdir})"

Here is the log for the last build command…

https://gist.githubusercontent.com/mpietruschka/79ef730d2fa2defeac80a0301d59e3ec/raw/a911d8dfe68fa8a477de54dd681d1b39804a5f9e/build_kopano-core.log

This is the packaging script…

https://gist.githubusercontent.com/mpietruschka/8c8293b1898e9de53bbe77f0f50607f1/raw/ee256e7a9d26c7d70a0e39d1483dcb024e1df46c/PKGBUILD

Any ideas?

jengelh

==> Tidying install...
  -> Removing libtool files...
  -> Purging unwanted files...
  -> Removing static library files...
  -> Stripping unneeded symbols from binaries and libraries...

This strip may be it, and it’s not coming from KC. But worry not, kopano-server can be started right from the build directory, it does not need make install or extra packaging for a debug session.

nm .libs/libkcserver.so

martimcfly

You’re awesome! This has been it!

I had to change the packager defaults to not strip symbols or remove libtool .la files and debug informations. I disabled it specifically for this package.

/etc/makepkg.conf

#########################################################################
# GLOBAL PACKAGE OPTIONS
#   These are default values for the options=() settings
#########################################################################
#
# Default: OPTIONS=(strip docs !libtool !staticlibs emptydirs zipman purge !optipng !upx !debug)
#  A negated option will do the opposite of the comments below.
#
#-- strip:      Strip symbols from binaries/libraries
#-- docs:       Save doc directories specified by DOC_DIRS
#-- libtool:    Leave libtool (.la) files in packages
#-- staticlibs: Leave static library (.a) files in packages
#-- emptydirs:  Leave empty directories in packages
#-- zipman:     Compress manual (man and info) pages in MAN_DIRS with gzip
#-- purge:      Remove files specified by PURGE_TARGETS
#-- upx:        Compress binary executable files using UPX
#-- optipng:    Optimize PNG images with optipng
#-- debug:      Add debugging flags as specified in DEBUG_* variables
#
OPTIONS=(!strip docs libtool staticlibs emptydirs zipman purge !optipng !upx debug)

The method finally got visible!

[root@server alarm]# nm /usr/lib/libkcserver.so | grep SerializeM
0018da7c t _ZN2KC16SerializeMessageEPNS_9ECSessionEPNS_10ECDatabaseEPNS_19ECAttachmentStorageEPKNS_10StreamCapsEjjjP4GUIDjPNS_12ECSerializerEb

Now we can go on with gdb :D

Marti

martimcfly

@jengelh I couldn’t start kopano from build directory because its build in a chroot environment on a different device :)

jengelh

Let me put up an obligatory questioning of the motive here ;-)

thought you wanted a native build because “I build this for Armv7 and it takes a while to finish”
if you have a fast machine to compile with, the DB could be transferred

I built KC in reasonable time even with an AMD AXP from 2003, which is probably a reasonable assumption for what a today’s raspberry pi ARM would be capable of.

martimcfly

I build for odroid on another ordroid :) both are armv7 and way faster as a RPI. The setup for the chroot environment and compilation take about 30 minutes together. A transfer of the database takes for ever. I don‘t think this is an option ;)

I could speed up compilation with distributed compilation. But I had no time to update my workers with gcc 7.

jengelh

It is because libtidy crashes, and that kills kopano-backup’s worker subprocess, which the main process also completely fails to act on.

The server side then reports 80000004 because the client disappeared in the middle of a send(2) call.

martimcfly

Hey jengelh,

I’d like to say thank you for all the tremendous help you gave going through the code finding the problem in the last place one would expect!

Marti

jengelh

https://github.com/htacg/tidy-html5/issues/739