729604 - Test incremental decompression

Status: RESOLVED WONTFIX

(Firefox for Android Graveyard :: General, defect, P1)

Description

[Mike Hommey [:glandium]]

I landed incremental decompression on mozilla-inbound today (bug 686805). It's still disabled by default there, because it requires something that is not yet possible on tinderbox builds.

Anyways, I built two different APKs off the same changeset, with the same compiled code, but one linked as on tinderbox (which is basically equivalent to any nightly build), and the other one with the best current state of incremental decompression.

It would be helpful to have these two builds tested and compared on different scenarios, on different devices. My testing on a Nexus S seems to show startup improvement.

The two builds in question are available on:
http://people.mozilla.org/~mhommey/incremental/

The -szip variant is the one with incremental decompression enabled. It is about 400K bigger due to the required change in compression format.

Comment 1

[cmtalbert]

Running the Autophone system (using blassey's instrumented startup numbers) on a nexus_one/droid_pro/nexus_s/samsung_gsII. Should have some numbers for you by the end of today.

Comment 2

[Tony Chung [:tchung]]

(In reply to Clint Talbert ( :ctalbert ) from comment #1)
> Running the Autophone system (using blassey's instrumented startup numbers)
> on a nexus_one/droid_pro/nexus_s/samsung_gsII. Should have some numbers for
> you by the end of today.

Hey clint, can your instrumented tests run against both of the suggested builds that mike put out, and show comparable results?   

Mike, If so, would clint's tests suffice for the test coverage you initially requested from QA to run manually?   if yes, then i'll defer our test resources to clint's results for now.   Of course if you want something more specific that requires manual test coverage, i'm happy to get our QA team to help.

Let us know what's the best approach.

Thanks,
Tony

Comment 3

[Mike Hommey [:glandium]]

(In reply to Tony Chung [:tchung] from comment #2)
> Mike, If so, would clint's tests suffice for the test coverage you initially
> requested from QA to run manually? 

That will be enough as an assessment of the benefits of the current implementation. It would be useful to have more extensive coverage in terms of devices afterwards. But then more in terms of stability than performance.

Comment 4

[cmtalbert]

The Non-incremental results.

The tests do 4 tests:
* local-blank - load a blank page from the file system on the phone
* local-twitter - load a twitter page from the file system on the phone
* remote-blank - load the same blank page using wifi
* remote-twitter - load the same twitter page using wifi

Other testing notes:
* The web server serving the pages is always the same and is on the same switch as the phones, so we're talking some best-case networking even in the non-local case.
* Each page is loaded 20 times, however, as noted in the enddrawing test, some of those generate invalid data and are culled out of the results.
* The nexus one has big issues running the incremental build. It usually complains that the build is not installed, but if you do pm -l, you can see it is installed. I only managed to get one successful run on one test with the nexus one.
* The results in milliseconds.

Throbber Start - time from starting application to instrumentation reporting the throbber has started
+-------------------------------+----------------+------------------------------+
| phoneid                       | testname       | avg(throbberstart-starttime) |
+-------------------------------+----------------+------------------------------+
| 04_46_65_fd_2f_e1_samsung_gs2 | local-blank    |                    1654.0000 |
| 78_d6_f0_cf_8d_11_nexus_s     | local-blank    |                    2455.0500 |
| 78_d6_f0_cf_d2_17_nexus_s     | local-blank    |                    2440.5500 |
| 98_0c_82_33_ec_8d_samsung_gs2 | local-blank    |                    1673.4000 |
| 98_4b_4a_2c_e4_89_droid_pro   | local-blank    |                    2695.0000 |
---------------------------------------------------------------------------------
| 00_23_76_96_cc_6f_nexus_one   | local-twitter  |                    5728.0000 |
| 04_46_65_fd_2f_e1_samsung_gs2 | local-twitter  |                    1763.0500 |
| 78_d6_f0_cf_8d_11_nexus_s     | local-twitter  |                    2429.4250 |
| 78_d6_f0_cf_d2_17_nexus_s     | local-twitter  |                    2551.2500 |
| 98_0c_82_33_ec_8d_samsung_gs2 | local-twitter  |                    1718.4000 |
| 98_4b_4a_2c_e4_89_droid_pro   | local-twitter  |                    2809.1500 |
---------------------------------------------------------------------------------
| 04_46_65_fd_2f_e1_samsung_gs2 | remote-blank   |                    1694.8500 |
| 78_d6_f0_cf_8d_11_nexus_s     | remote-blank   |                    2396.4000 |
| 78_d6_f0_cf_d2_17_nexus_s     | remote-blank   |                    2516.6500 |
| 98_0c_82_33_ec_8d_samsung_gs2 | remote-blank   |                    1656.1000 |
| 98_4b_4a_2c_e4_89_droid_pro   | remote-blank   |                    2699.9000 |
---------------------------------------------------------------------------------
| 04_46_65_fd_2f_e1_samsung_gs2 | remote-twitter |                    2141.6000 |
| 78_d6_f0_cf_8d_11_nexus_s     | remote-twitter |                    2412.9032 |
| 78_d6_f0_cf_d2_17_nexus_s     | remote-twitter |                    2543.0000 |
| 98_0c_82_33_ec_8d_samsung_gs2 | remote-twitter |                    1939.7000 |
| 98_4b_4a_2c_e4_89_droid_pro   | remote-twitter |                    2698.5000 |
+-------------------------------+----------------+------------------------------+

Throbber stop - time from starting application to instrumentation reporting the throbber has stopped.
+-------------------------------+----------------+-----------------------------+
| phoneid                       | testname       | avg(throbberstop-starttime) |
+-------------------------------+----------------+-----------------------------+
| 04_46_65_fd_2f_e1_samsung_gs2 | local-blank    |                   1764.0000 |
| 78_d6_f0_cf_8d_11_nexus_s     | local-blank    |                   2737.9000 |
| 78_d6_f0_cf_d2_17_nexus_s     | local-blank    |                   2715.8500 |
| 98_0c_82_33_ec_8d_samsung_gs2 | local-blank    |                   1782.5500 |
| 98_4b_4a_2c_e4_89_droid_pro   | local-blank    |                   2984.7000 |
--------------------------------------------------------------------------------
| 00_23_76_96_cc_6f_nexus_one   | local-twitter  |                   5692.0000 |
| 04_46_65_fd_2f_e1_samsung_gs2 | local-twitter  |                   2682.0500 |
| 78_d6_f0_cf_8d_11_nexus_s     | local-twitter  |                   5213.3750 |
| 78_d6_f0_cf_d2_17_nexus_s     | local-twitter  |                   5369.8000 |
| 98_0c_82_33_ec_8d_samsung_gs2 | local-twitter  |                   2676.7000 |
| 98_4b_4a_2c_e4_89_droid_pro   | local-twitter  |                   6307.1500 |
--------------------------------------------------------------------------------
| 04_46_65_fd_2f_e1_samsung_gs2 | remote-blank   |                   1963.2000 |
| 78_d6_f0_cf_8d_11_nexus_s     | remote-blank   |                   2832.3500 |
| 78_d6_f0_cf_d2_17_nexus_s     | remote-blank   |                   2957.1500 |
| 98_0c_82_33_ec_8d_samsung_gs2 | remote-blank   |                   1918.4000 |
| 98_4b_4a_2c_e4_89_droid_pro   | remote-blank   |                   3152.8000 |
--------------------------------------------------------------------------------
| 04_46_65_fd_2f_e1_samsung_gs2 | remote-twitter |                   3781.7000 |
| 78_d6_f0_cf_8d_11_nexus_s     | remote-twitter |                   5323.8710 |
| 78_d6_f0_cf_d2_17_nexus_s     | remote-twitter |                   5827.2500 |
| 98_0c_82_33_ec_8d_samsung_gs2 | remote-twitter |                   3660.9500 |
| 98_4b_4a_2c_e4_89_droid_pro   | remote-twitter |                   6287.0000 |
+-------------------------------+----------------+-----------------------------+

End Drawing - time from starting application to reporting that the drawing of the loaded page is complete.
* Note: the local-blank is based on a smaller set of data because of bug 724708, which causes the test to report lots of 0's for enddrawing (it never gets an end-drawing notice from the instrumentation). The other tests do not have this issue.
+-------------------------------+----------------+---------------------------+
| phoneid                       | testname       | avg(enddrawing-starttime) |
+-------------------------------+----------------+---------------------------+
| 04_46_65_fd_2f_e1_samsung_gs2 | local-blank    |                 1888.8571 |
| 78_d6_f0_cf_8d_11_nexus_s     | local-blank    |                 3064.7500 |
| 78_d6_f0_cf_d2_17_nexus_s     | local-blank    |                 2981.2000 |
| 98_0c_82_33_ec_8d_samsung_gs2 | local-blank    |                 1910.4286 |
| 98_4b_4a_2c_e4_89_droid_pro   | local-blank    |                 3252.8500 |
------------------------------------------------------------------------------
| 00_23_76_96_cc_6f_nexus_one   | local-twitter  |                10555.0000 |
| 04_46_65_fd_2f_e1_samsung_gs2 | local-twitter  |                 3108.7500 |
| 78_d6_f0_cf_8d_11_nexus_s     | local-twitter  |                 5632.2250 |
| 78_d6_f0_cf_d2_17_nexus_s     | local-twitter  |                 5789.9500 |
| 98_0c_82_33_ec_8d_samsung_gs2 | local-twitter  |                 3097.9500 |
| 98_4b_4a_2c_e4_89_droid_pro   | local-twitter  |                 6533.7500 |
------------------------------------------------------------------------------
| 04_46_65_fd_2f_e1_samsung_gs2 | remote-blank   |                 2193.5500 |
| 78_d6_f0_cf_8d_11_nexus_s     | remote-blank   |                 3218.5000 |
| 78_d6_f0_cf_d2_17_nexus_s     | remote-blank   |                 3331.8500 |
| 98_0c_82_33_ec_8d_samsung_gs2 | remote-blank   |                 2146.7500 |
| 98_4b_4a_2c_e4_89_droid_pro   | remote-blank   |                 3489.4000 |
------------------------------------------------------------------------------
| 04_46_65_fd_2f_e1_samsung_gs2 | remote-twitter |                 4103.7368 |
| 78_d6_f0_cf_8d_11_nexus_s     | remote-twitter |                 5524.9032 |
| 78_d6_f0_cf_d2_17_nexus_s     | remote-twitter |                 6032.5000 |
| 98_0c_82_33_ec_8d_samsung_gs2 | remote-twitter |                 3943.8947 |
| 98_4b_4a_2c_e4_89_droid_pro   | remote-twitter |                 6399.9000 |
+-------------------------------+----------------+---------------------------+

Comment 5

[cmtalbert]

Somehow my comment unset the priority and tracking-fennec flags, trying to set them back. mfinkle had set tracking-fennec to 13+ and Priority to P1.

Comment 6

[cmtalbert]

Ok, and now the incremental results.

This is the -szip build from comment 0. And once again the nexus one crapped out. So I just eliminated it from the results.

Throbber start
+-------------------------------+----------------+------------------------------+
| phoneid                       | testname       | avg(throbberstart-starttime) |
+-------------------------------+----------------+------------------------------+
| 04_46_65_fd_2f_e1_samsung_gs2 | local-blank    |                    1309.1500 |
| 78_d6_f0_cf_8d_11_nexus_s     | local-blank    |                    2002.9500 |
| 78_d6_f0_cf_d2_17_nexus_s     | local-blank    |                    2040.4500 |
| 98_0c_82_33_ec_8d_samsung_gs2 | local-blank    |                    1303.2500 |
| 98_4b_4a_2c_e4_89_droid_pro   | local-blank    |                    2311.8500 |
---------------------------------------------------------------------------------
| 04_46_65_fd_2f_e1_samsung_gs2 | local-twitter  |                    1460.8000 |
| 78_d6_f0_cf_8d_11_nexus_s     | local-twitter  |                    2045.8000 |
| 78_d6_f0_cf_d2_17_nexus_s     | local-twitter  |                    2098.1000 |
| 98_0c_82_33_ec_8d_samsung_gs2 | local-twitter  |                    1411.4000 |
| 98_4b_4a_2c_e4_89_droid_pro   | local-twitter  |                    2426.3000 |
---------------------------------------------------------------------------------
| 04_46_65_fd_2f_e1_samsung_gs2 | remote-blank   |                    1323.7000 |
| 78_d6_f0_cf_8d_11_nexus_s     | remote-blank   |                    2099.8500 |
| 78_d6_f0_cf_d2_17_nexus_s     | remote-blank   |                    2156.9000 |
| 98_0c_82_33_ec_8d_samsung_gs2 | remote-blank   |                    1312.4500 |
| 98_4b_4a_2c_e4_89_droid_pro   | remote-blank   |                    2320.2000 |
---------------------------------------------------------------------------------
| 04_46_65_fd_2f_e1_samsung_gs2 | remote-twitter |                    1322.3000 |
| 78_d6_f0_cf_8d_11_nexus_s     | remote-twitter |                    2031.1000 |
| 78_d6_f0_cf_d2_17_nexus_s     | remote-twitter |                    2077.8500 |
| 98_0c_82_33_ec_8d_samsung_gs2 | remote-twitter |                    1319.5500 |
| 98_4b_4a_2c_e4_89_droid_pro   | remote-twitter |                    2325.3684 |
+-------------------------------+----------------+------------------------------+

Throbber Stop
+-------------------------------+----------------+-----------------------------+
| phoneid                       | testname       | avg(throbberstop-starttime) |
+-------------------------------+----------------+-----------------------------+
| 04_46_65_fd_2f_e1_samsung_gs2 | local-blank    |                   1455.4000 |
| 78_d6_f0_cf_8d_11_nexus_s     | local-blank    |                   2355.3000 |
| 78_d6_f0_cf_d2_17_nexus_s     | local-blank    |                   2377.7500 |
| 98_0c_82_33_ec_8d_samsung_gs2 | local-blank    |                   1445.5500 |
| 98_4b_4a_2c_e4_89_droid_pro   | local-blank    |                   2631.1500 |
| 04_46_65_fd_2f_e1_samsung_gs2 | local-twitter  |                   2571.1000 |
| 78_d6_f0_cf_8d_11_nexus_s     | local-twitter  |                   4473.9000 |
| 78_d6_f0_cf_d2_17_nexus_s     | local-twitter  |                   4628.2000 |
| 98_0c_82_33_ec_8d_samsung_gs2 | local-twitter  |                   2536.0000 |
| 98_4b_4a_2c_e4_89_droid_pro   | local-twitter  |                   5533.1000 |
| 04_46_65_fd_2f_e1_samsung_gs2 | remote-blank   |                   1646.4000 |
| 78_d6_f0_cf_8d_11_nexus_s     | remote-blank   |                   2635.5000 |
| 78_d6_f0_cf_d2_17_nexus_s     | remote-blank   |                   2715.3500 |
| 98_0c_82_33_ec_8d_samsung_gs2 | remote-blank   |                   1608.2000 |
| 98_4b_4a_2c_e4_89_droid_pro   | remote-blank   |                   2806.6000 |
| 04_46_65_fd_2f_e1_samsung_gs2 | remote-twitter |                   3341.4000 |
| 78_d6_f0_cf_8d_11_nexus_s     | remote-twitter |                   4994.2500 |
| 78_d6_f0_cf_d2_17_nexus_s     | remote-twitter |                   5093.6500 |
| 98_0c_82_33_ec_8d_samsung_gs2 | remote-twitter |                   3155.0000 |
| 98_4b_4a_2c_e4_89_droid_pro   | remote-twitter |                   5947.6316 |
+-------------------------------+----------------+-----------------------------+

End Drawing
* Once again, the same caveat applies to the local-blank numbers as above (however there were less 0's in this batch than in the other, not sure if that is just random chance or what).
+-------------------------------+----------------+---------------------------+
| phoneid                       | testname       | avg(enddrawing-starttime) |
+-------------------------------+----------------+---------------------------+
| 04_46_65_fd_2f_e1_samsung_gs2 | local-blank    |                 1573.7500 |
| 78_d6_f0_cf_8d_11_nexus_s     | local-blank    |                 2639.9091 |
| 78_d6_f0_cf_d2_17_nexus_s     | local-blank    |                 2665.2727 |
| 98_0c_82_33_ec_8d_samsung_gs2 | local-blank    |                 1556.4286 |
| 98_4b_4a_2c_e4_89_droid_pro   | local-blank    |                 2923.9000 |
------------------------------------------------------------------------------
| 04_46_65_fd_2f_e1_samsung_gs2 | local-twitter  |                 3006.9500 |
| 78_d6_f0_cf_8d_11_nexus_s     | local-twitter  |                 4928.7000 |
| 78_d6_f0_cf_d2_17_nexus_s     | local-twitter  |                 5078.3500 |
| 98_0c_82_33_ec_8d_samsung_gs2 | local-twitter  |                 2957.6000 |
| 98_4b_4a_2c_e4_89_droid_pro   | local-twitter  |                 6156.7000 |
------------------------------------------------------------------------------
| 04_46_65_fd_2f_e1_samsung_gs2 | remote-blank   |                 1880.1000 |
| 78_d6_f0_cf_8d_11_nexus_s     | remote-blank   |                 2844.3500 |
| 78_d6_f0_cf_d2_17_nexus_s     | remote-blank   |                 2966.5000 |
| 98_0c_82_33_ec_8d_samsung_gs2 | remote-blank   |                 1829.2000 |
| 98_4b_4a_2c_e4_89_droid_pro   | remote-blank   |                 3202.6000 |
------------------------------------------------------------------------------
| 04_46_65_fd_2f_e1_samsung_gs2 | remote-twitter |                 3564.2000 |
| 78_d6_f0_cf_8d_11_nexus_s     | remote-twitter |                 5216.6000 |
| 78_d6_f0_cf_d2_17_nexus_s     | remote-twitter |                 5314.7000 |
| 98_0c_82_33_ec_8d_samsung_gs2 | remote-twitter |                 3383.1500 |
| 98_4b_4a_2c_e4_89_droid_pro   | remote-twitter |                 6103.5263 |
+-------------------------------+----------------+---------------------------+

Comment 7

[cmtalbert]

Ok, after trying a bunch of complicated mysql, I gave up and used google docs.

But you can see that across the board, the numbers look good. Incremental unzipping is a win:
https://docs.google.com/spreadsheet/ccc?key=0AkFMXPiC6OptdDdQSTBYYUNqMFNlTzhHT2ZqTGtYSEE

I dropped all the nexus one data so that the numbers would line up.  We should figure out why it doesn't work at all on the nexus one.

Comment 8

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Perhaps we can move the nexus one to a different bug?  Mike, is this good enough for what you need?  or do you need some more comparisons?

Comment 9

[Mike Hommey [:glandium]]

(In reply to Naoki Hirata :nhirata from comment #8)
> Perhaps we can move the nexus one to a different bug?  Mike, is this good
> enough for what you need?  or do you need some more comparisons?

That's good enough for comparisons. Thanks a lot. Next I'd like the -szip build to be tested on plenty various devices, but that's lower priority.

Comment 10

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Thanks Mike.  Just to get clarity from your comment 9, what are we testing for?  Just bugs in general?  or are you looking for performance numbers?

Comment 11

[Mike Hommey [:glandium]]

(In reply to Naoki Hirata :nhirata from comment #10)
> Thanks Mike.  Just to get clarity from your comment 9, what are we testing
> for?  Just bugs in general?  or are you looking for performance numbers?

cf. comment 3, stability.

Comment 12

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Ah ok.  will do.

Comment 13

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

I didn't get any crashes on ICS as well as gingerbread more so than before; I did notice that the web search for google did not work:

02-28 14:30:32.759: I/GeckoApp(18712): title - webgl - Google Search
02-28 14:30:32.970: E/GeckoConsole(18712): [JavaScript Error: "no element found" {file: "https://www.google.com/search?q=webgl&ie=utf-8&oe=utf-8&aq=t&rls=org.mozilla:en-US:unofficial&sky=ee1" line: 2 column: 503 source: "<!DOCTYPE html PUBLIC "-//WAPFORUM//DTD XHTML Mobile 1.0//EN" "http://www.wapforum.org/DTD/xhtml-mobile10.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><title>webgl - Google Search</title><meta http-equiv="Content-Type" content="application/xhtml+xml; charset=UTF-8"/></head><H1>Server Error</H1>The server encountered an error and could not complete your request.<p>If the problem persists, please mail <tt>[e-mail]</tt> and mention this error message and the query that caused it.</p>"}]


I also got some other Gecko issues:
02-28 14:31:25.978: I/GeckoApp(18712): Got message: Menu:Remove
02-28 14:31:25.978: E/GeckoConsole(18712): [JavaScript Warning: "Duplicate resource declaration for 'gre-resources' ignored." {file: "jar:jar:file:///data/app/org.mozilla.fennec_mh-1.apk!/omni.ja!/chrome/nonlocalized.manifest" line: 6}]
02-28 14:31:25.985: E/GeckoConsole(18712): Could not read chrome manifest 'file:///data/data/org.mozilla.fennec_mh/chrome.manifest'.

Question: would this affect l10n in any fashion?

Comment 14

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Clarification on Question in Comment 13.  "This" is in regards to the incremental decompression change.

Comment 15

[Mike Hommey [:glandium]]

(In reply to Naoki Hirata :nhirata from comment #13)
> I didn't get any crashes on ICS as well as gingerbread more so than before;
> I did notice that the web search for google did not work:
> 
> 02-28 14:30:32.759: I/GeckoApp(18712): title - webgl - Google Search
> 02-28 14:30:32.970: E/GeckoConsole(18712): [JavaScript Error: "no element
> found" {file:
> "https://www.google.com/search?q=webgl&ie=utf-8&oe=utf-8&aq=t&rls=org.
> mozilla:en-US:unofficial&sky=ee1" line: 2 column: 503 source: "<!DOCTYPE
> html PUBLIC "-//WAPFORUM//DTD XHTML Mobile 1.0//EN"
> "http://www.wapforum.org/DTD/xhtml-mobile10.dtd"><html
> xmlns="http://www.w3.org/1999/xhtml"><head><title>webgl - Google
> Search</title><meta http-equiv="Content-Type"
> content="application/xhtml+xml; charset=UTF-8"/></head><H1>Server
> Error</H1>The server encountered an error and could not complete your
> request.<p>If the problem persists, please mail <tt>[e-mail]</tt>
> and mention this error message and the query that caused it.</p>"}]
> 
> 
> I also got some other Gecko issues:
> 02-28 14:31:25.978: I/GeckoApp(18712): Got message: Menu:Remove
> 02-28 14:31:25.978: E/GeckoConsole(18712): [JavaScript Warning: "Duplicate
> resource declaration for 'gre-resources' ignored." {file:
> "jar:jar:file:///data/app/org.mozilla.fennec_mh-1.apk!/omni.ja!/chrome/
> nonlocalized.manifest" line: 6}]
> 02-28 14:31:25.985: E/GeckoConsole(18712): Could not read chrome manifest
> 'file:///data/data/org.mozilla.fennec_mh/chrome.manifest'.

AFAICT, none of the above should be related to incremental decompression

> Question: would this affect l10n in any fashion?

No.

Comment 16

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Ok, cool.  I'll investigate the other issues more closely.  Is there anything else that needs to be done for these builds?

Comment 17

[Mike Hommey [:glandium]]

I think we're all good, thanks.

Comment 18

[Mike Hommey [:glandium]]

Do we have HTC phones around (HTC Desire HD)? It seems we're going to have trouble with incremental decompression on these :-/
http://www.mail-archive.com/[e-mail]/msg153389.html

Comment 19

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Mike, I found that the build doesn't work well with the LG Revolution.  I tried launching it; fennec does launch, however it took like 30 seconds to launch.  Following that trying to press anything is pretty dog slow  (expect around a 30 second lag time in response).  Android version for this phone is 2.2.2; Kernel version is 2.6.32.9-perf lg-electronics@area88 #1

Seems to work fine on the HTC Desire HD : Kernel version 2.6.35.10-g931a37e htc-kernel@and18-2 #1 Wed Nov 9 14:04:03 CST 2011

I had done a system update.  The system update is dated after the mail post, it's possible that HTC had fixed the issue with this patch?

Also tested out on the HTC Flyer and it seemed fine.

Comment 20

[Mike Hommey [:glandium]]

(In reply to Naoki Hirata :nhirata from comment #19)
> Mike, I found that the build doesn't work well with the LG Revolution.  I
> tried launching it; fennec does launch, however it took like 30 seconds to
> launch.  Following that trying to press anything is pretty dog slow  (expect
> around a 30 second lag time in response).  Android version for this phone is
> 2.2.2; Kernel version is 2.6.32.9-perf lg-electronics@area88 #1

How does the non -szip build perform there?

> Seems to work fine on the HTC Desire HD : Kernel version 2.6.35.10-g931a37e
> htc-kernel@and18-2 #1 Wed Nov 9 14:04:03 CST 2011
> 
> I had done a system update.  The system update is dated after the mail post,
> it's possible that HTC had fixed the issue with this patch?

I checked the kernel source for that kernel (since HTC publishes them), and indeed, the code mentioned in the bug is gone.

Comment 21

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

(In reply to Mike Hommey [:glandium] from comment #20)
> (In reply to Naoki Hirata :nhirata from comment #19)
> > Mike, I found that the build doesn't work well with the LG Revolution.  I
> > tried launching it; fennec does launch, however it took like 30 seconds to
> > launch.  Following that trying to press anything is pretty dog slow  (expect
> > around a 30 second lag time in response).  Android version for this phone is
> > 2.2.2; Kernel version is 2.6.32.9-perf lg-electronics@area88 #1
> 
> How does the non -szip build perform there?

The non-szip build seems to be working correctly.  I am able to browse to www.cnn.com, www.google.com, news.google.com; open a link in a different tab, switch to another tab.  Performance seems fine.  Panning shows checkerboarding (which is about on par with the nightly); graphic issues occur when dismissing the vkb which is the same as the nightly.

Comment 22

[Mike Hommey [:glandium]]

How could we figure what's going wrong on the LG Revolution with the szip build?

Comment 23

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Attachment: logcat

Comment 24

[Mike Hommey [:glandium]]

(In reply to Naoki Hirata :nhirata from comment #23)
> Created attachment 605558 [details]
> logcat

From the perspective of the linker and on-demand decompression, this looks fairly normal :-/

Comment 25

[Jan Henning [:JanH]]

Superseded by bug 1291424, I think.