chromium issue

philo_neo · Aug 13, 2017

Hello,
Question: chromium under FreeBSD 10.3 (i686) presents problems of execution, I read that it was necessary after the compilation of the port cd /usr/ports/www/chromium/ && make install clean
add kern.ipc.shm_allow_removed = 1 to /etc/sysctl.conf and reboot!
Nothing to do it starts and then disappears, what must you do?

Ps: I also run pkg install chromium
The icon appears

Regards
Philippe

Sensucht94 · Aug 16, 2017

philo_neo, in order to allow anyone to help you, you should run chrome from a terminal emulator and post here the output.

Did you check up for missing dependencies?

philo_neo said:
Ps: I also make pkg install chromium
The icon appears

What do you mean by icon appearing? You should have deleted the port before installing the package or it should be telling it's already updated to the latest version.

Running 'make pkg install', while in the port directory, is not the correct way to build the port, use the handbook as reference to learn more: https://www.freebsd.org/doc/handbook/pkgng-intro.html

vall · Sep 3, 2017

Does anyone uses www/chromium ?

If someone do, how do you install or config it??
I tried to use it for a couple times two years along and always it's garbage.

For now I have 11.1-RELEASE, everything is updated, kern.ipc.shm_allow_removed = 1 in /etc/sysctl.conf.
Chromium is installed from pkg, not from ports.
And on three machines I just have

Code:

$ chrome
Segmentation fault
$

macondo · Sep 3, 2017

 sudo pkg install chromium

after installation is complete:

chrome

voilà!

cpm@ · Sep 4, 2017

It has been reported via bugzilla, so I'll take a look.

According to the OP, chromium 64-bit version appears to be unaffected.

See PR 221266 for more details.

vall · Sep 4, 2017

Yes, it's happening on i386. Everything is as in PR 221266.

I've added dmesg output

cpm@ · Sep 4, 2017

Wozzeck.Live

I agree with you, the current problems we have with chromium on FreeBSD would be resolved quickly if we had official support from the chromium engineering dev team.

There is a initial discussion to enable the GPU hardware acceleration support. Chromium uses OpenGL and probably some extensions which are usually provided by the graphic stack from the corresponding vendor. Assuming that the driver provides right version of OpenGL and extensions, there is high chance that the things would work. In reverse, if mesa is too old or our in-kernel driver is too old for mesa to activate recent OpenGL features there is no development on chromium side that could help there.

You will have more details when it is published on the dedicated mailing list.

cpm@ · Sep 4, 2017

vall said:
Yes, it's happening on i386. Everything is as in PR 221266.

I've added dmesg output

Have you tried to run chromium with GPU hardware acceleration disabled? Note that if software renderer is not in place, then the GPU process won't launch.

chrome --disable-gpu

vall · Sep 5, 2017

cpm@ said:
Have you tried to run chromium with GPU hardware acceleration disabled? Note that if software renderer is not in place, then the GPU process won't launch.

chrome --disable-gpu

yep, that's the same

do I can help as a user?
Is chromium stable on amd64 now (as I remember it craches every second tab)?

cpm@ · Sep 5, 2017

vall said:
yep, that's the same

You can give a try to build chromium from ports instead using binary package.

do I can help as a user?

Yes, why not?

Is chromium stable on amd64 now (as I remember it craches every second tab)?

Yes, it works quite well after disabling V8 cache.

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=212812#c24

FYI, I'm using it right now

vall · Sep 5, 2017

cpm@ said:
Yes, why not?

I meaned "how I can help?"

English is a problem for me ))

Ok I'll try some things and tell about achievements. I've already tried to install from ports yesterday, but my machine is slow and my patience was а small

cpm@ said:
Yes, it works quite well after disabling V8 cache.

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=212812#c24

FYI, I'm using it right now

Thanks for showing this solution! I am forcing to switch my friend to FreeBSD, but he is web-developer and chromium is his greatest need.

cpm@ · Sep 5, 2017

vall said:
I meaned "how I can help?" English is a problem for me ))

As a user: it would be enough to submit patches to fix/improve the port and/or report issues.

cpm@ · Sep 13, 2017

philo_neo, vall

Would you mind trying to rebuild chromium after applying the following patch?

vall · Sep 13, 2017

cpm@ said:
philo_neo, vall

Would you mind trying to rebuild chromium after applying the following patch?

Ok, I'll try

tingo · Sep 13, 2017

cpm@ said:
philo_neo, vall

Would you mind trying to rebuild chromium after applying the following patch?

FWIW, I tried this patch. I patched like this:

Code:

root@kg-core1# pwd
/usr/ports/www/chromium
root@kg-core1# patch  -p 2 < /home/tingo//work/chromium/chromium.patch
[...]

and that looked ok, it only complained about a few missing newlines.
but when I try to upgrade chromium, it fails during patching:

Code:

===>  Patching for chromium-60.0.3112.113_2
===>  Applying extra patch /usr/ports/www/chromium/files/extra-patch-clang
===>  Applying extra patch /usr/ports/www/chromium/files/extra-patch-libc++-old
===>  Applying FreeBSD patches for chromium-60.0.3112.113_2
  I can't seem to find a patch in there anywhere.
=> FreeBSD patch patch-base_task_scheduler_delayed_task_manager.cc failed to apply cleanly.
=> Patch(es)  patch-BUILD.gn patch-apps_ui_views_app__window__frame__view.cc patch-ash_display_mirror__window__controller.cc patch-base_BUILD.gn patch-base__trace_event__process_memory_dump.h patch-base_allocator_allocator__shim.cc patch-base_allocator_allocator__shim__default__dispatch__to__glibc.cc patch-base_allocator_allocator__shim__override__libc__symbols.h patch-base_allocator_allocator__shim__unittest.cc patch-base_debug_debugger__posix.cc patch-base_debug_proc__maps__linux.cc patch-base_debug_stack__trace.h patch-base_debug_stack__trace__posix.cc patch-base_debug_thread__heap__usage__tracker.cc patch-base_files_file__path__unittest.cc patch-base_files_file__path__watcher__kqueue.h patch-base_files_file__util.h patch-base_files_memory_mapped_file_posix.cc patch-base_i18n_icu__util.cc patch-base_linux__util.cc patch-base_logging__unittest.cc patch-base_native__library__posix.cc patch-base_posix_unix__domain__socket__linux.cc patch-base_posix_unix__domain__socket__linux__unittest.cc patch-base_process_internal__linux.cc patch-base_process_internal__linux.h patch-base_process_launch.cc patch-base_process_launch.h patch-base_process_launch__posix.cc patch-base_process_memory.cc patch-base_process_memory.h patch-base_process_memory__unittest.cc patch-base_process_process__handle__freebsd.cc patch-base_process_process__info__linux.cc patch-base_process_process__iterator__freebsd.cc patch-base_process_process__metrics.h patch-base_process_process__metrics__freebsd.cc patch-base_process_process__posix.cc patch-base_profiler_stack__sampling__profiler__unittest.cc patch-base_security__unittest.cc patch-base_sys__info__freebsd.cc patch-base_sys__info__posix.cc applied cleanly.
*** Error code 1

Stop.
make[1]: stopped in /usr/ports/www/chromium

Do I need more arguments to patch, or is something broken?
This on

Code:

tingo@kg-core1$ uname -a
FreeBSD kg-core1.kg4.no 10.3-STABLE FreeBSD 10.3-STABLE #0 r310083: Wed Dec 14 21:00:13 CET 2016     root@kg-core1.kg4.no:/usr/obj/usr/src/sys/GENERIC  amd64

HTH

cpm@ · Sep 13, 2017

It works fine for me...

Code:

# cd /usr/ports/www/chromium
# patch -p2 < /home/cpm/Desktop/chromium-libc++.diff
Hmm...  Looks like a unified diff to me...
The text leading up to this was:
--------------------------
|Index: www/chromium/Makefile
|===================================================================
|--- www/chromium/Makefile    (revision 449598)
|+++ www/chromium/Makefile    (working copy)
--------------------------
Patching file Makefile using Plan A...
Hunk #1 succeeded at 3.
Hunk #2 succeeded at 201.
Hmm...  The next patch looks like a unified diff to me...
The text leading up to this was:
--------------------------
|Index: www/chromium/files/extra-patch-libc++-new
|===================================================================
|--- www/chromium/files/extra-patch-libc++-new    (nonexistent)
|+++ www/chromium/files/extra-patch-libc++-new    (working copy)
--------------------------
(Creating file files/extra-patch-libc++-new...)
Patching file files/extra-patch-libc++-new using Plan A...
Empty context always matches.
Hunk #1 succeeded at 1.
Hmm...  The next patch looks like a unified diff to me...
The text leading up to this was:
--------------------------
|
|Property changes on: www/chromium/files/extra-patch-libc++-new
|___________________________________________________________________
|Added: fbsd:nokeywords
|## -0,0 +1 ##
|+yes
|\ No newline at end of property
|Added: svn:eol-style
|## -0,0 +1 ##
|+native
|\ No newline at end of property
|Added: svn:mime-type
|## -0,0 +1 ##
|+text/plain
|\ No newline at end of property
|Index: www/chromium/files/extra-patch-libc++-old
|===================================================================
|--- www/chromium/files/extra-patch-libc++-old    (nonexistent)
|+++ www/chromium/files/extra-patch-libc++-old    (working copy)
--------------------------
(Creating file files/extra-patch-libc++-old...)
Patching file files/extra-patch-libc++-old using Plan A...
Empty context always matches.
Hunk #1 succeeded at 1.
Hmm...  The next patch looks like a unified diff to me...
The text leading up to this was:
--------------------------
|
|Property changes on: www/chromium/files/extra-patch-libc++-old
|___________________________________________________________________
|Added: fbsd:nokeywords
|## -0,0 +1 ##
|+yes
|\ No newline at end of property
|Added: svn:eol-style
|## -0,0 +1 ##
|+native
|\ No newline at end of property
|Added: svn:mime-type
|## -0,0 +1 ##
|+text/plain
|\ No newline at end of property
|Index: www/chromium/files/patch-base_task_scheduler_delayed_task_manager.cc
|===================================================================
|--- www/chromium/files/patch-base_task_scheduler_delayed_task_manager.cc    (revision 449598)
|+++ www/chromium/files/patch-base_task_scheduler_delayed_task_manager.cc    (nonexistent)
--------------------------
Patching file files/patch-base_task_scheduler_delayed_task_manager.cc using Plan A...
Hunk #1 succeeded at 0.
Hmm...  The next patch looks like a unified diff to me...
The text leading up to this was:
--------------------------
|
|Property changes on: www/chromium/files/patch-base_task_scheduler_delayed_task_manager.cc
|___________________________________________________________________
|Deleted: fbsd:nokeywords
|## -1 +0,0 ##
|-yes
|\ No newline at end of property
|Deleted: svn:eol-style
|## -1 +0,0 ##
|-native
|\ No newline at end of property
|Deleted: svn:mime-type
|## -1 +0,0 ##
|-text/plain
|\ No newline at end of property
|Index: www/chromium/files/patch-tools_gn_bootstrap_bootstrap.py
|===================================================================
|--- www/chromium/files/patch-tools_gn_bootstrap_bootstrap.py    (revision 449598)
|+++ www/chromium/files/patch-tools_gn_bootstrap_bootstrap.py    (nonexistent)
--------------------------
Patching file files/patch-tools_gn_bootstrap_bootstrap.py using Plan A...
Hunk #1 succeeded at 0.
Hmm...  Ignoring the trailing garbage.
done

P.S. You should remove the leftover patches:

patch-base_task_scheduler_delayed_task_manager.cc
patch-tools_gn_bootstrap_bootstrap.py

tingo · Sep 15, 2017

Aha, so I should have used '-E' with the patch command then. Removed the two zero-length patch files, and now chromium is building.
Thanks for the hint.

vall · Sep 22, 2017

Oh, I've built it.
Twice. First one to make sure that problem is.
Then clean ports/www/chromium catalogue, reexport chromium port, then apply patch, delete patch-base_task_scheduler_delayed_task_manager.cc and patch-tools_gn_bootstrap_bootstrap.py and build again.
Maybe I'm doing something wrong and somebody should check too, but still Segmentation fault.

upd:
And while I was building this port (yep I can't run computer all day long) new version of port with fix was released ))
should rebuild again

upd2:
Tried to install version 61.0.3163.79 from pkg. Segmentation fault.
Maybe we just should upgrade to amd64 ))

cpm@ · Sep 22, 2017

vall said:
upd2:
Tried to install version 61.0.3163.79 from pkg. Segmentation fault.
Maybe we just should upgrade to amd64 ))

Please, show me the following output

chrome --disable-gpu --enable-logging=stderr --v=1

end222 · Sep 22, 2017

cpm@ said:
Please, show me the following output

chrome --disable-gpu --enable-logging=stderr --v=1

I get this output too:

Code:

$ chrome --disable-gpu --enable-logging=stderr --v=1
Segmentation fault

I am using the latest version installed with pkg in a i386 machine

jrm@ · Sep 23, 2017

Off topic aside: cpm@, you are a kind soul to maintain and debug a port that takes 15.5 hours to build.

cpm@ · Sep 23, 2017

end222 said:
I get this output too:

Code:

$ chrome --disable-gpu --enable-logging=stderr --v=1 Segmentation fault

I am using the latest version installed with pkg in a i386 machine

I'm not sure if it would make any difference, but it will probably work.

Add ld(1) option --no-keep-memory for IA32 machine.

Code:

Index: Makefile
===================================================================
--- Makefile    (revision 450418)
+++ Makefile    (working copy)
@@ -209,6 +209,12 @@
 GN_ARGS+=    use_vulcanize=false
 .endif

+.if ${ARCH} == i386
+GN_ARGS+=    extra_ldflags="-L${LOCALBASE}/lib -Wl,--no-keep-memory"
+.else
+GN_ARGS+=    extra_ldflags="-L${LOCALBASE}/lib"
+.endif
+
 .if !defined(GN_ONLY)
 pre-everything::
    @${ECHO_MSG}

vall · Sep 23, 2017

cpm@ said:

I'm not sure if it would make any difference, but it will probably work.

Add ld(1) option --no-keep-memory for IA32 machine.

Code:

Index: Makefile
===================================================================
--- Makefile    (revision 450418)
+++ Makefile    (working copy)
@@ -209,6 +209,12 @@
 GN_ARGS+=    use_vulcanize=false
 .endif

+.if ${ARCH} == i386
+GN_ARGS+=    extra_ldflags="-L${LOCALBASE}/lib -Wl,--no-keep-memory"
+.else
+GN_ARGS+=    extra_ldflags="-L${LOCALBASE}/lib"
+.endif
+
 .if !defined(GN_ONLY)
 pre-everything::
    @${ECHO_MSG}

On my mom's SandyBridge in VirtualBox compilation is 6 time faster then on Pentium4

and near object-file number 4500 I have this:

ld.lld : unknown argument: --no-keep-memory

bookwormep · Sep 23, 2017

I'll give another try to re-build www/chromium; after refreshing ports using portsnap.

EDIT: Previous thread re: www/chromium :
https://forums.freebsd.org/threads/61667/

cpm@ · Sep 24, 2017

vall said:
ld.lld : unknown argument: --no-keep-memory

Yes, it should only work with GNU ld(1).

Change the following build argument

Code:

GN_ARGS+=use_lld=false

in chromium/Makefile.

chromium issue

philo_neo

Sensucht94

Guest

vall

macondo

Attachments

cpm@

Moderator

vall

Attachments

cpm@

Moderator

cpm@

Moderator

vall

cpm@

Moderator

vall

cpm@

Moderator

cpm@

Moderator

vall

tingo

cpm@

Moderator

tingo

vall

cpm@

Moderator

end222

jrm@

cpm@

Moderator

vall

bookwormep

cpm@

Moderator