835592 - Get sample markers working on ~beta era code (b2g18)

Status: RESOLVED WONTFIX

(Core :: General, defect)

Description

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

I tried to backport this in bug 834622, but still couldn't get them to work.

Backport bug 773428 part 1: attachment 707329 [details] [diff] [review]
Backport part 2: attachment 707288 [details] [diff] [review]
Add some IPC markers: attachment 707298 [details] [diff] [review]

The markers I added aren't showing up in the raw JSON dumps with those three patches applied.

Comment 1

[Benoit Girard (:BenWa)]

Cool. I've been wanting to add markers into the ipdl codegen but had not done it yet since Desktop/Fennec doesn't use a ton of ipdl.

Can your back port patch?

Comment 2

[Benoit Girard (:BenWa)]

You may also be interested in looking into bug 785440 for adding them to DOM Bindings.

Comment 3

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

Oh, I think there's a bit of confusion: I added sample *labels* to IPDL-generated code (attachment 706602 [details] [diff] [review]), a long time ago.  I haven't landed it because I don't know the perf impact yet.

This bug is about getting arbitrary sample *markers* working in this era of the code.  With the patches in comment 0, sample *markers* are still completely hosed.

The third patch in comment 0 adds new markers for some dom/ipc clients that are separate from IPDL.

Comment 4

[Benoit Girard (:BenWa)]

Right, I was just remarking that it was neat.

I wrote comment 1 in a hurry, can you post your backport patch as you're trying to apply it to b2g18? There's likely something missing since it works fine on central. Note that bits from bug 773428 is only required to add a marker from JS. SAMPLE_MARKER doesn't need bug 773428.

Comment 5

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

Do you mean

Backport bug 773428 part 1: attachment 707329 [details] [diff] [review]
Backport part 2: attachment 707288 [details] [diff] [review]

?

Comment 6

[Benoit Girard (:BenWa)]

Well I pointed out that you we're missing lines introduced from bug 799638 attachment 671154 [details] [diff] [review] so I wanted to see what you had so far.

AFAIK attachment 671154 [details] [diff] [review] is all you need unless you want to add labels from JS.

Comment 7

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

I do want to add labels from JS :).

Comment 8

[(no longer active)]

(In reply to comment #3)
> Oh, I think there's a bit of confusion: I added sample *labels* to
> IPDL-generated code (attachment 706602 [details] [diff] [review]), a long time ago.  I haven't landed it
> because I don't know the perf impact yet.

For b2g we could try to disable the profiler unless --enable-profiling is used to build (by adding an extra case here: <http://mxr.mozilla.org/mozilla-central/source/configure.in#1697>).  That would let developers take full advantage of this but also make sure other builds are not negatively effected.

Comment 9

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

Attachment: Rebased (non-functional) patches on my branch

I don't think I have attachment 671154 [details] [diff] [review] .  Looks like maybe the one I need.

Comment 10

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

Well that's odd ... with attachment 671154 [details] [diff] [review] applied, I get a marker from the JS caller, but not from the two C++ callers :(.  One of the C++ callers runs in the same process as the JS caller, so that's not the issue.

Comment 11

[Benoit Girard (:BenWa)]

Comment on attachment 707466 [details] [diff] [review]
Rebased (non-functional) patches on my branch

This patch seems good. I can't eye ball what the remaining problem is however.

Comment 12

[Dietrich Ayala (:dietrich)]

We're not pushing anything b2g18 anymore, WONTFIXING. Reopen if this bug is still useful.