769761 - Heap unclassified is quite high on Android on planet.mozilla.org

Status: RESOLVED INVALID

(Firefox for Android Graveyard :: General, defect)

Description

[Jeff Muizelaar [:jrmuizel]]

I'm 59MB of explicit and 24MB of heap-unclassified. The resident size is also quite a bit bigger than the explicit heap size (119MB).

Comment 1

[Nicholas Nethercote [inactive]]

The resident/explicit mismatch might be due to Java.

As for heap-unclassified, I'm not even sure if DMD will run on Android, so it's unlikely for there to be any progress here.

Comment 2

[Nicholas Nethercote [inactive]]

> As for heap-unclassified, I'm not even sure if DMD will run on Android

The new version of DMD should run on Android, though I'm not sure of the exact steps.  It'll be similar to the steps for B2G...

Comment 3

[Kartikaya Gupta (email:[e-mail])]

(In reply to Nicholas Nethercote [:njn] from comment #2)
> The new version of DMD should run on Android, though I'm not sure of the
> exact steps.

I've added steps to run DMD on Android to https://wiki.mozilla.org/Performance/MemShrink/DMD

Comment 4

[Nicholas Nethercote [inactive]]

Thanks, kats!  A couple of questions about the docs:

- Don't you need to define $DMD?  What if you want to define an option via it (such as DMD='--sample-below=1')?

- Do you need to run a stack-fixing script after dumping?

Comment 5

[Kartikaya Gupta (email:[e-mail])]

(In reply to Nicholas Nethercote [:njn] from comment #4)
> - Don't you need to define $DMD?  What if you want to define an option via
> it (such as DMD='--sample-below=1')?

Yeah, I set DMD=1 on the command line to start fennec. (The line was fairly long so perhaps it was truncated out of sight; I've since line-wrapped the command to avoid that). You can set DMD to whatever you need.

> 
> - Do you need to run a stack-fixing script after dumping?

I hadn't actually looked closely at the log I got, but it turns out that there are no stacks in it :( See http://people.mozilla.com/~kgupta/bug/769761-dmd.txt for an example.

Comment 6

[Nicholas Nethercote [inactive]]

> I hadn't actually looked closely at the log I got, but it turns out that
> there are no stacks in it :( See
> http://people.mozilla.com/~kgupta/bug/769761-dmd.txt for an example.

That's bad!  Can you file a bug and CC jlebar and glandium?  Stacks are working on B2G so hopefully there's something simple going wrong.

Comment 7

[Kartikaya Gupta (email:[e-mail])]

Filed bug 823354 for that.

Comment 8

[Aaron Train [:aaronmt]]

Is there any reason why this is still unconfirmed?

Comment 9

[Nicholas Nethercote [inactive]]

> Is there any reason why this is still unconfirmed?

Because the difference between UNCONFIRMED and NEW and ASSIGNED isn't that meaningful in practice -- many bugs don't have them set appropriately.  General activity within a bug (i.e. comments and patches) is a much better indicator of whether it's getting attention.

Comment 10

[miralobontiu]

Hi,

Does this issue still reproduce?

Thanks!

Comment 11

[Jeff Muizelaar [:jrmuizel]]

I'll just close it as invalid