Author Topic: Firefox - 45.9.0 for OS/2 GA1.1...anyone...anyone???  (Read 8611 times)

Dave Yeo

  • Hero Member
  • *****
  • Thank You
  • -Given: 9
  • -Receive: 110
  • Posts: 1614
    • View Profile
Re: Firefox - 45.9.0 for OS/2 GA1.1...anyone...anyone???
« Reply #45 on: May 01, 2018, 02:08:13 am »
Perhaps comment on the Github issue as dmik doesn't monitor here much.
Here most of my SM builds play video and sound fine, the latest Firefox also works with the right optimizations most of the time, it seems to be a timing issue.
Currently dmik is investigating the high CPU usage when Firefox is minimized.

roberto

  • Sr. Member
  • ****
  • Thank You
  • -Given: 16
  • -Receive: 7
  • Posts: 416
    • View Profile
Re: Firefox - 45.9.0 for OS/2 GA1.1...anyone...anyone???
« Reply #46 on: May 04, 2018, 04:44:37 pm »
Well, in my case I'm watching videos with this program that I attached to qbasic, and with the seamonkey 2.35 almost perfect.
I execute it in a os / 2 window, with qbasic / run c: \ pk02.bas, before opening the seamonkey, I leave the window without moving it, nor minimizing it, I open the seamonkey and it is already.
saludos

Dave Yeo

  • Hero Member
  • *****
  • Thank You
  • -Given: 9
  • -Receive: 110
  • Posts: 1614
    • View Profile
Re: Firefox - 45.9.0 for OS/2 GA1.1...anyone...anyone???
« Reply #47 on: May 04, 2018, 05:52:53 pm »
And what are all those pokes supposed to do?

roberto

  • Sr. Member
  • ****
  • Thank You
  • -Given: 16
  • -Receive: 7
  • Posts: 416
    • View Profile
Re: Firefox - 45.9.0 for OS/2 GA1.1...anyone...anyone???
« Reply #48 on: May 04, 2018, 06:50:49 pm »
Ha, ha, ha I do not know exactly. But it works for me. I guess someone who knows what those memory positions are will understand. I just had a problem with the mahjonng, and looking with theseus I saw that those memory positions, were constantly changing. The question is why. And if I fix them, the mahjong worked, and the videos with some tweaks improved.
saludos

roberto

  • Sr. Member
  • ****
  • Thank You
  • -Given: 16
  • -Receive: 7
  • Posts: 416
    • View Profile
Re: Firefox - 45.9.0 for OS/2 GA1.1...anyone...anyone???
« Reply #49 on: May 05, 2018, 11:47:12 pm »
If I can tell you that there is a consequence, that it occurs in the manjong and in the videos, if I run the pokes, the first pmshell gets most of the work done, and everything is pretty good. But if I do not execute it, the highest workload is seen in the second pmshell.
I understand that the first pmshell is the one with the lowest pid number.

Andreas Schnellbacher

  • Sr. Member
  • ****
  • Thank You
  • -Given: 1
  • -Receive: 16
  • Posts: 333
    • View Profile
Re: Firefox - 45.9.0 for OS/2 GA1.1...anyone...anyone???
« Reply #50 on: May 09, 2018, 05:37:41 pm »
I'd reported that the new Firefox 45.9.0 traps at xul.dll. Today I tried a recent test version after removing old profiles and that has worked. Also PMDll doesn't trap on loading xul.dll anymore. Maybe that's caused by the test version, I haven't tried anything else. The test version is ff45_9_0_t5.7z.

Unfortunately that one is not usable on my older single-core hardware. Trying to move the browser window with the mouse, blocks the system until the focus is changed once by the CAD handler. The window is repainted, but the move isn't processed. CPU usage is 100 % all the time. That's not really new compared to the postings on bitwiseworks/mozilla-os2.

Holger Schuett

  • Jr. Member
  • **
  • Thank You
  • -Given: 0
  • -Receive: 1
  • Posts: 59
    • View Profile
Re: Firefox - 45.9.0 for OS/2 GA1.1...anyone...anyone???
« Reply #51 on: May 13, 2018, 12:23:30 pm »
Hi there,
I have been busy trying the different Firefox versions the last weeks and here is one observation I made.
For the records: I run ArcaOs 5.01 on a Thinkpad X61. The firefox supplied with the system (38.8.0) runs fine and plays youtube videos well when I am connected to my wlan at home.
The 45.9 Version I have been testing concerning the youtube issues is firefox-45.9.0-3.t2.pentium4. I changed one setting in the about:config after searching the web:

Quote
Firefox stores the current session status (opened urls, cookies, history and form data) to the disk on a regular basis. It is used to recover a previous session in case of crash. The default setting is to save the session every 15 seconds, resulting in frequent disk access.

To increase the save interval to 10 minutes for example, change in about:config the setting of browser.sessionstore.interval to 600000

Either I am lucky or this changed the behaviour when starting youtube. The youtube page loads much faster than it did before and now I am able to play the most videos without severe problems.
Perhaps someone would like to try this? As mentioned, perhaps I am only lucky?

Holger

André Heldoorn

  • Sr. Member
  • ****
  • Thank You
  • -Given: 38
  • -Receive: 9
  • Posts: 371
    • View Profile
Re: Firefox - 45.9.0 for OS/2 GA1.1...anyone...anyone???
« Reply #52 on: June 07, 2018, 03:11:15 pm »
Presumably the 100% CPU load of single core CPUs or the ~95% CPU load of a ThinkPad T42's fast Pentium 4 CPU isn't solved by the latest release for "OS/2"?

Dave Yeo

  • Hero Member
  • *****
  • Thank You
  • -Given: 9
  • -Receive: 110
  • Posts: 1614
    • View Profile
Re: Firefox - 45.9.0 for OS/2 GA1.1...anyone...anyone???
« Reply #53 on: June 07, 2018, 05:23:03 pm »
Presumably the 100% CPU load of single core CPUs or the ~95% CPU load of a ThinkPad T42's fast Pentium 4 CPU isn't solved by the latest release for "OS/2"?

Testing here, on my T42 with a 1.6Ghz Pentium M, displaying the start page and waiting a few minutes for it to calm down, it is using about 2% (0.8-4.4) CPU with the very odd peak to about 33%.

André Heldoorn

  • Sr. Member
  • ****
  • Thank You
  • -Given: 38
  • -Receive: 9
  • Posts: 371
    • View Profile
Re: Firefox - 45.9.0 for OS/2 GA1.1...anyone...anyone???
« Reply #54 on: June 07, 2018, 08:52:29 pm »
Interesting. Is there already a SM release based on the same code? Sounds like its worth another full test now, based on the T42 "benchmark" with a former CPU load of nearly 100% (same here).

Dave Yeo

  • Hero Member
  • *****
  • Thank You
  • -Given: 9
  • -Receive: 110
  • Posts: 1614
    • View Profile
Re: Firefox - 45.9.0 for OS/2 GA1.1...anyone...anyone???
« Reply #55 on: June 08, 2018, 02:06:27 am »
Interesting. Is there already a SM release based on the same code?

Almost, Firefox currently is optimized with the JavaScript code for speed, -O3 and the rest for size, -Os.
I built https://bitbucket.org/dryeo/dry-comm-esr31/downloads/seamonkey-2.42.9esr_r3.en-US.os2.zip with straight -O3 and targeting the i686. There's also https://bitbucket.org/dryeo/dry-comm-esr31/downloads/thunderbird-45.8.0_r3.en-US.os2.zip with the same optimizations.
I'll probably do a build targeting the Pentium M with the JS code -O3 and the rest -Os. Our compiler is broken for building it all targeting -O3 and any CPU using SSE or SSE2, doesn't align the instructions and static variables correctly.

Rich Walsh

  • Jr. Member
  • **
  • Thank You
  • -Given: 1
  • -Receive: 6
  • Posts: 66
  • ONU! (OS/2 is NOT Unix!)
    • View Profile
Re: Firefox - 45.9.0 for OS/2 GA1.1...anyone...anyone???
« Reply #56 on: June 08, 2018, 03:39:57 am »
I'll probably do a build targeting the Pentium M with the JS code -O3 and the rest -Os. Our compiler is broken for building it all targeting -O3 and any CPU using SSE or SSE2, doesn't align the instructions and static variables correctly.

Why, oh why, oh why do you keep using -O3?  If it's incompetent when it comes to SSE, why do you think it's capable of any extreme optimization? While the bugs it introduces for SSE may be painfully obvious, the bugs it causes elsewhere are likely subtle and hard to track down - but equally as obvious when you try to use the thing.

Dial it back, pleeeeze...

Dave Yeo

  • Hero Member
  • *****
  • Thank You
  • -Given: 9
  • -Receive: 110
  • Posts: 1614
    • View Profile
Re: Firefox - 45.9.0 for OS/2 GA1.1...anyone...anyone???
« Reply #57 on: June 08, 2018, 05:21:23 am »
I'll probably do a build targeting the Pentium M with the JS code -O3 and the rest -Os. Our compiler is broken for building it all targeting -O3 and any CPU using SSE or SSE2, doesn't align the instructions and static variables correctly.

Why, oh why, oh why do you keep using -O3?  If it's incompetent when it comes to SSE, why do you think it's capable of any extreme optimization? While the bugs it introduces for SSE may be painfully obvious, the bugs it causes elsewhere are likely subtle and hard to track down - but equally as obvious when you try to use the thing.

Dial it back, pleeeeze...

With -O2, lots of people were getting failures to play Youtube.

Joop

  • Sr. Member
  • ****
  • Thank You
  • -Given: 0
  • -Receive: 38
  • Posts: 450
    • View Profile
Re: Firefox - 45.9.0 for OS/2 GA1.1...anyone...anyone???
« Reply #58 on: July 22, 2018, 01:52:38 pm »
Don't get it to work. There is trouble with LIBC.

Exception C0000005 - Access Violation

Process:  FIREFOX.EXE (05/21/2018 17:17:07 51,129)
PID:      48 (72)
TID:      01 (1)
Priority: 200

Filename: LIBC066.DLL (10/26/2014 20:16:29 1,361,666)
Address:  005B:1DCBE6C0 (0001:0005E6C0)
Cause:    Attempted to read from 00000001
          (not a valid address)

After that a listing of about 200Kb.
As far as I know its the last version of LIBC (libc-0.6.6-csd6.zip) and the last version of FF (firefox-45.9.0-4.oc00.pentium4.7z) (why not just zip?)
Is there a way to run the program?
« Last Edit: July 22, 2018, 01:55:07 pm by Joop »

ivan

  • Hero Member
  • *****
  • Thank You
  • -Given: 1
  • -Receive: 27
  • Posts: 821
    • View Profile
Re: Firefox - 45.9.0 for OS/2 GA1.1...anyone...anyone???
« Reply #59 on: July 22, 2018, 02:12:25 pm »
Why oh why don't those porting firefox and other programs supply a zip file containing ALL the various DLLs that they use to have it working?  That way there shouldn't be the dll hell we appear to have with versions of RPM files from odd experimental repositories or even those not in any repository yet.