/compat/linux Web Browser

[Ted Mittelstaedt]

contact the app vendor there is certainly some kind of bug where it is 
not cleaning up properly on exit, perhaps

not calling free on all the memory it allocates, assuming that when the 
program closes it will automatically

free all ram.  This is sloppy programming on the part of the app vendor.

Ted

On 10/25/2020 1:05 PM, JB via freebsd-emulation wrote:
> Hi Emulators,
> Can someone suggest the next troubleshooting steps for an application that doesn't work well from the Linux compatibility layer?
>
> Application: pale moon web browser
> Version: 28.14.2
> Host: FreeBSD host1 12.1-RELEASE-p10 FreeBSD 12.1-RELEASE-p10 GENERIC amd64
> Linux version: linux-c7-7.8.2003
> Installed packages: [see attachments]
> Installation details:
> 1. Downloaded the application's tarball (link below) and extracted it into /compat/linux/opt/.
> 2. Confirmed all libraries listed by `ldd' are installed inside /compat/linux/ or in memory.
> 3. Confirmed the binary is supported via `brandelf' ('SVR4' (0))
> User profile: newly created, with all defaults (no extensions)
> Command used: /compat/linux/opt/palemoon.dir/palemoon -P linux_c7 -new-window about:blank
>
> Problem description:
> The application runs increasingly longer with each successive launch until it hangs indefinitely, pegging two CPU cores at 100%, with the only recourse being to kill the application with ^C.
>
> Here's an example:
> Launch the browser for the first time (since log-in) and it hangs without displaying the browser window. Kill it. Launch it again, this time the window appears but hangs while loading the homepage (remote). Kill it. Launch again with '-new-window about:blank' to sidestep the hanging. It works and is completely responsive until you load a website (freebsd.org and palemoon.org). Kill and relaunch again. The websites load if done one at a time. Next, try accessing a third-party link from palemoon.org (to install extensions), and it hangs again. This is as far as this dance gets us; no further progress appears possible. If launched now, the window appears, the webpages load (done one at a time), and the browser will happily sit there idling at 100% CPU usage on one core, but fully responsive otherwise ... until logout, and the cycle repeats itself.
>
> What I've tried:
> The usual gambit. `truss' output shows the browser being very chatty, but mostly in line with pale moon being launched from ports (now-discontinued). `vmstat' looks clean, and aside from `top' showing pale moon's RAM load (SIZE) at 25G, the thread names appear to be switching normally (assuming it's normal) between SSL Cert #1, IndexedDB, StreamTrans #2, etc. The browser's error console is empty.
>
> Attachments:
> `procstat -r'
> `procstat -kk'
> `truss -c'
> `top -Hd 10'
> palemoon core dumps list
> installed packages
> pale moon memory report
>
> Download link: http://linux.palemoon.org/download/mainline/
>
> Thanks.
>
> Sent with [ProtonMail](https://protonmail.com) Secure Email.
>
> _______________________________________________
> freebsd-emulation at freebsd.org mailing list
> https://lists.freebsd.org/mailman/listinfo/freebsd-emulation
> To unsubscribe, send any mail to "freebsd-emulation-unsubscribe at freebsd.org"