Welcome to GOFASTMOTORSPORTS.com ... ... ... Green Flag - Green Flag - Go Go Go ... ... Brought to you by Ricks Satellite - Where the Big Dish Rules!

Ricks Satellite Wildfeed and Backhaul Forum
Register Latest Topics
 
 
 


Reply
  Author   Comment   Page 117 of 121     «   Prev   114   115   116   117   118   119   120   Next   »
dem0nlord

Avatar / Picture

Registered:
Posts: 75
Reply with quote  #1741 
Back to Jessie on this Xeon too. It's an older engineering sample, 2.3ghz, 10 cores with hyperthreading, 25M cache. It's got a second socket and could take two 12 core CPUs, I wonder how fast it would be then. [eek]

Code:

real    1m16.493s
user    21m41.027s
sys     1m31.244s

DarkSky

Registered:
Posts: 2,262
Reply with quote  #1742 
I compiled the latest from torvalds git to day (4.11.0-rc6) and it works good with media_build other than that nagging problem I have run across on a few kernels the last few days with it erroring out with the frame_vector.c issue. I can not seem to get a handle on why it is doing it so far. I have to rem it out (Makefile.mm)  in the v4l/Makefile and resume compiling.

Code:

ifeq ($(makefile-media),1)
-include $(obj)/Makefile.media
-include $(obj)/Makefile.sound
endif
#ifeq ($(makefile-mm),1)
#-include $(obj)/Makefile.mm
#endif
-include $(obj)/.kconfig.de

Here is what I think I know:

There is a patch for kernel 4.2 for frame_vector.c.
It appears that the later kernels has frame_vector compiled in the kernel.
The linux/Makefile deletes frame_vector.c in the mm/ directory but I have never seen frame_vector.c there though.
It seems to run a check to see if it needs to be compiled in a script and some kernels there are no problems compiling and some error out and they all seem to have it compiled in the kernel.

By the way majortom, the skinny signals on 101W have cranked up their power again and having to view with the $10 el_cheapo usb device. Thank you so much. I can see my star trek marathons every night.

Attached Images
Click image for larger version - Name: st.png, Views: 14, Size: 514.83 KB 

DarkSky

Registered:
Posts: 2,262
Reply with quote  #1743 
I finally got to the bottom of the intermitent frame_vector.c error issue here.

For a very long time when using a deb based systrem I have not done a "make deb-pkg" when compiling a kernel simply because the building of the debs doubled the compile time. I have been making a build link in /lib/modules/'uname -r'/ pointing to the source tree I had compiled. This has not been and issue in the past when I had compiled media_build_udl because frame_vector was compiled as a module with the older kernels and now with the newer kernels it is compiled in the kernel.

media_build runs a check in the in the build directory in the files to see if frame_vector is needed so if you use the kernel tree for the build directory it picks up that frame_vector is needed brcause of the src files being present as opposed to just header/Makefile/Kconfig files being in a kernel header package install.

Since I moved on to arch the compiling process is different and no kernel header packages are created automatically when compiling a kernel (and is true with some other distros also). Arch will supply a kernel header package if you are using their own kernel from their repo.

So what I did was take arch's kernel PKGBUILD script and adapted the creating a kernel header section into a bash script and made my own kernel headers and pointed /lib/modules/'uname -r'/ build to it and media_build_udl built just fine then.

I attached the script in case some one else is in the same boat as I was when building a kernel that is not in your distro's repo and compiling media_build_ud. Drop it in the root of the kernel source tree and run it as root. It will create a directory called build in the root of the kernel source tree and then you make a link in  /lib/modules/'uname -r'/ pointing to the root-of-the-new-kernel/build.

Remove the .txt from the attached file and unpack the tarball.

This is for x86; for use with an arm device it would have to be modified.

 
Attached Files
txt generate-kernel-headers.sh.tar.txt (10.00 KB, 6 views)

dem0nlord

Avatar / Picture

Registered:
Posts: 75
Reply with quote  #1744 
Quote:
Originally Posted by dem0nlord

Raspbian with gcc6.3 and it's few dependencies from the stretch repo:18m30.055s   temp=74.1


Remember this temperature was with a heatsink on and the lid(which is ventilated) of the One Nine Designs case removed since it was going over 80 degrees with the lid installed and using the same heatsink. Today I put my Pi3 board into an aluminum Flirc case which has a protrusion that touches the SOC, eliminating the heatsink. Doing exactly the same build but fully closed up in this case the max temp was 65.5. At least a 15 degree improvement is huge! So there are passive cooling solutions that work, it just isn't those little heatsinks.
majortom

Avatar / Picture

Registered:
Posts: 5,002
Reply with quote  #1745 
Like that idea... Might get a couple myself....
Even though mine doesn't seem to be slowing down from heat. At least I haven't noticed..

__________________
Perfect 10 7.5' mesh Chapparal C/Ku Co-Rotor feed, Norsat 8115 CBAND, Norsat 4106A Ku Thomson Saginaw Actuator, Vbox positioner,
Home Brew Polarotor683 Servo circuit. Openbox S9, Prof7500, v4l-updatelee linux drivers installed.
ricardo's geo-orbit archive: https://web.archive.org/web/20071202191450/http://www.geo-orbit.org:80/sizepgs/tuningp2.html
majortom

Avatar / Picture

Registered:
Posts: 5,002
Reply with quote  #1746 
Quote:
Originally Posted by DarkSky
I finally got to the bottom of the intermitent frame_vector.c error issue here....

Makes sense to me darksky. Was wondering what you guys were talkin about... Having never had a problem, was like huh
what are they talkin about??? LOL[biggrin]


__________________
Perfect 10 7.5' mesh Chapparal C/Ku Co-Rotor feed, Norsat 8115 CBAND, Norsat 4106A Ku Thomson Saginaw Actuator, Vbox positioner,
Home Brew Polarotor683 Servo circuit. Openbox S9, Prof7500, v4l-updatelee linux drivers installed.
ricardo's geo-orbit archive: https://web.archive.org/web/20071202191450/http://www.geo-orbit.org:80/sizepgs/tuningp2.html
DarkSky

Registered:
Posts: 2,262
Reply with quote  #1747 
dem0nlord:

I like the idea of using a case for a heatsink also. It should increase the life of the pi in the long run. Some times extreme heat messes with the solder joints over a period of time.

majortom:

I have thought about how media_build searches for certain strings for frame_vector and they would be strings used in the source code and not being a programer it does not make any sense that I can comprehend in my little pea brain. If it was me seems like I would run a check in the .config file looking for CONFIG_FRAME_VECTOR=y or CONFIG_FRAME_VECTOR=m but what do I know?

I guess there is a good reason why they switched to building frame_vector in the kernel and some times when they do switch you do not have an option to build it as a module. There is a patch in backports/ for v4.2_frame_vector.patch so I am guessing that was the last time it was built as a module. The patch is a complete frame_vector.c file and it is put in linux/mm/ directory.
majortom

Avatar / Picture

Registered:
Posts: 5,002
Reply with quote  #1748 
I have no idea what it is for either. So I am not doing anything with it one way or the other.
Doesn't get in my way... In fact until you guys mentioned it, had no idea there was ever a file by that name.

If you can test for the same issue, using the stock kernels, and the stock media_build, on the same system,
I would submit a post to the linux-media mailing list about it. In theory the maintainers could fix if they could repeat the issue
and sooner or later we'd pull that in to here...

You can find some more history of it by searching the tree, looking for related comments in the commits.
https://bitbucket.org/majortom/media_build_udl/commits/all?search=frame_vector


__________________
Perfect 10 7.5' mesh Chapparal C/Ku Co-Rotor feed, Norsat 8115 CBAND, Norsat 4106A Ku Thomson Saginaw Actuator, Vbox positioner,
Home Brew Polarotor683 Servo circuit. Openbox S9, Prof7500, v4l-updatelee linux drivers installed.
ricardo's geo-orbit archive: https://web.archive.org/web/20071202191450/http://www.geo-orbit.org:80/sizepgs/tuningp2.html
DarkSky

Registered:
Posts: 2,262
Reply with quote  #1749 
A FYI:

UDL updated his kernel a couple days ago if you want to upgrade to his current version. I upgraded my system here and all is working well as usual:

Code:

VERSION = 4
PATCHLEVEL = 11
SUBLEVEL = 0
EXTRAVERSION = -rc5
NAME = Fearless Coyote
majortom

Avatar / Picture

Registered:
Posts: 5,002
Reply with quote  #1750 
Darksky noticed that the most recent merge in v4l-updatelee had broke the el_cheapo hdstar device again, and I noticed it has this dma buffer fix in it.
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/drivers/media/usb/dvb-usb/dw2102.c?id=606142af57dad981b78707234cfbd15f9f7b7125

somewhere in the course of merging, it undoes what we did to get the el_cheapo / geniatech hdstar device working.
He went through the original December commit to dw2102.c and re-implemented it, resulting in this patch. Both he and I have tested it, confirming no longer needing to disable CONFIG_VMAP_STACK
in .config and the geniatech hdstar device is working fine again. I have committed that patch to local copy of v4l-updatelee tree, but haven't pushed it to bitbucket yet.
Are there any other devices that used the dvb_usb_dw2102 that also required disabling that since kernel 4.9?

On another note, this patch sitting in patchwork, should fix the hauppauge aero-m, which also had that same bug, requiring disabling CONFIG_VMAP_STACK to workaround it. So I think the aero-m should get fixed also in due time..

https://patchwork.kernel.org/patch/9693445/




Attached Images
Click image for larger version - Name: all_set_vmap_stack.png, Views: 12, Size: 380.09 KB 

 
Attached Files
txt dw2102.c.patch.txt (6.33 KB, 7 views)


__________________
Perfect 10 7.5' mesh Chapparal C/Ku Co-Rotor feed, Norsat 8115 CBAND, Norsat 4106A Ku Thomson Saginaw Actuator, Vbox positioner,
Home Brew Polarotor683 Servo circuit. Openbox S9, Prof7500, v4l-updatelee linux drivers installed.
ricardo's geo-orbit archive: https://web.archive.org/web/20071202191450/http://www.geo-orbit.org:80/sizepgs/tuningp2.html

updatelee

Avatar / Picture

Registered:
Posts: 3,117
Reply with quote  #1751 
Patch applied and pushed upstream, thanks [smile]

UDL

__________________
TBS6925/5980, Prof 7301/7500/8000, Genpix Skywalker-1, Skystar 2 Express HD
Hauppauge 950Q/Aero/Aero-m, Kworld 330U/435v3/445v3

I use Linux and support open source projects
Download my opensrc projects at http://updatelee.blogspot.com
andyinyakima

Avatar / Picture

Registered:
Posts: 1,520
Reply with quote  #1752 
Ok, If I understand this correctly, I can clone v4l-updatelee --depth 1 and be bug free?
__________________
andyinyakima
-----------------------------------
Recs: TBS6983, TBS5925 __ Ant: 90cm on SG6100__OSes: Linux with v4l-updatelee drivers __ RPi2 with v4l-updatelee drivers

Open Book, Open Source, You have to Open it to Know what's in it!
DarkSky

Registered:
Posts: 2,262
Reply with quote  #1753 
Andy;

I just compiled it and it is bug free for me with my devices. El_Cheapo (Geniatech HD Star) doing great with VMAP_STACK=y in .config compiled in.

Pic is my 7301 and el_cheapo tuned and el_cheapo playing video in vlc.

Attached Images
Click image for larger version - Name: bug-free.png, Views: 9, Size: 413.24 KB 

andyinyakima

Avatar / Picture

Registered:
Posts: 1,520
Reply with quote  #1754 
I have v4l-updatelee and updateDVB installed.

My 5925 which is USB won't come up at first

sudo cp -Rv firmware/* /lib/firmware
sudo reboot

called from the v4l-updatelee folder cured that. Did I overlook something? Probably!

Was nice to install qwt from apt-get and not a mercurial (svn) tarball went I compiled updateDVB.
Actually I compiled it first and got a qwt error. Went back to look for the qwt down load and saw
it setting in the sudo apt-get install line.[smile]

Anyhow looking good.


Screenshot_2017-04-27_13-49-49.png 


__________________
andyinyakima
-----------------------------------
Recs: TBS6983, TBS5925 __ Ant: 90cm on SG6100__OSes: Linux with v4l-updatelee drivers __ RPi2 with v4l-updatelee drivers

Open Book, Open Source, You have to Open it to Know what's in it!
updatelee

Avatar / Picture

Registered:
Posts: 3,117
Reply with quote  #1755 
updated README to reflect, thanks [smile]

UDL

__________________
TBS6925/5980, Prof 7301/7500/8000, Genpix Skywalker-1, Skystar 2 Express HD
Hauppauge 950Q/Aero/Aero-m, Kworld 330U/435v3/445v3

I use Linux and support open source projects
Download my opensrc projects at http://updatelee.blogspot.com
Previous Topic | Next Topic
Print
Reply

Quick Navigation:

Easily create a Forum Website with Website Toolbox.

Fellow Members, your posts are welcome here! Do not worry about posting everything perfect. Different receivers and LNB's will give you different Frequencies and Symbol Rates. Some set top boxes, PCI cards and USB receivers, Do Not Require all of the same information that others may need. It is not Required to post everything that others may need to tune in a feed. It is just most important to share the find. We can always adjust the Frequency and Symbol Rates and try the various Modulations and FEC's on our own receivers until we get a lock and then give a polite reply with what works for your receiver, as that information might help others as well. We all appreciate the efforts and energy of the Posters!

Thank You for Visiting GOFASTMOTORSPORTS.com - Keep Your Eyes on the Sky and the Track!