rust 1.50 on aarch64

mortonP · Tux's lil' helper Joined: 22 Dec 2015 Posts: 88

I'm trying to build Rust 1.50 on aarch64.
Rust 1.49 worked fine (afair Rust 1.48 also) but with this latest release build is stuck at

Lebkoungcity · Posted: Sat Mar 13, 2021 7:21 am Post subject:

Hi mortonP,

I don't have a solution for your problem but I had something similar on my RPi4 32bit-install. After hours of compiling it stopped at some point and stood there motionless until I killed it. Maybe at the same point as yours? I don't know I don't have the build-log anymore...
As a workaround I emerged dev-lang/rust-bin (and eselected it afterwards with 'eselect rust'). Maybe I will try dev-lang/rust sometime in the future but for the time being I could go on with emerging stuff.

I keep my fingers crossed there'll be a solution for your issue!
Andy
_________________
"The most dangerous world view is the world view of the ones who haven't viewed the world."
Alexander von Humboldt (1769-1859)

NeddySeagoon · Posted: Sat Mar 13, 2021 9:49 am Post subject:

mortonP,

It does build. I've done it on an X-Gene 1 and a Cavium Thunder

Poke about in `/var/tmp/portage/dev-lang/rust-1.50.0/work/ and find the build log.
What is the timestamp on the build log?
That's the last time the log was updated.

Pastebin the last 100 lines or so from the build log.

There is a very long single threaded part at the end of the build, so even on the 96 core Cavium Thunder its slow as only one core can work on that part of the build.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

mortonP · Tux's lil' helper Joined: 22 Dec 2015 Posts: 88

I do not doubt that it built properly, however I also know that in rare cases when sun and moon are in a certain alignment, stuff does not work, and later magically again. Maybe me and Lebkoungcity have rare luck to live through such a moon and sun alignment? :-)

The end of terminal output is https://pastebin.com/raw/hAg0ZLCQ

Searching for this output in a build log

mortonP · Tux's lil' helper Joined: 22 Dec 2015 Posts: 88

As an update, I can report Rust 1.51 builds fine!
(although it takes almost 2 days ;-)

Maybe 1.50 really had a bug, or the stars are in a different alignment now and a rare race does not show up...

Morton

mortonP · Tux's lil' helper Joined: 22 Dec 2015 Posts: 88

hangs with Rust 1.56 again - bad stars alignment again? :-(

with MAKEOPTS="-j1" it took a loooong time, but it ran through...

pingtoo · Posted: Mon Oct 25, 2021 8:20 pm Post subject:

Wild guess, have you try to use pstree -g and strace?

Use pstree to find if there some unexpected process branch that is waiting for something.

Use strace on leaf of each pstree branch to see the leaf process if they were waiting for something.

mortonP · Tux's lil' helper Joined: 22 Dec 2015 Posts: 88

I do remember, I first saw in htop one core doing seemingly nothing - so e.g. 4 core parallelization but only 3 cores working, and soon later all cores doing nothing. Probably one core/process gets stuck and then everyone else waits for something and this never finishes. I saw with ps process stuck in "S" state and an additional modifier, but don't remember the details.
Next time I'll try to take a closer look :-)