682201 - Android reftests are disabled for BMP, ICO decoder and encoder

Status: RESOLVED DUPLICATE of bug 753639

(Core :: Graphics: ImageLib, defect)

Description

[Brian R. Bondy [:bbondy]]

The reftests for BMP and ICO encoders and decoders (all recently added) are not working on Android so skip Android was added.  This task is to re-enable them and fix anything that doesn't work on Android for these reftests.

Comment 1

[Phil Ringnalda (:philor)]

Whenever I see a test sticking in a Tegra's craw, I think to myself "I should cc jmaher on this bug!"

Comment 2

[Joel Maher ( :jmaher ) (UTC -8)]

oh, thanks for disabling them and filing a bug.

Comment 3

[Brian R. Bondy [:bbondy]]

Attachment: Patch that re-enables android reftests

Running patch which removes the if-android-check through try now in case the problem was just Bug 682568's decodeondraw = true.

http://tbpl.allizom.org/?tree=Try&usebuildbot=1&rev=c79ed4256163

Comment 4

[Brian R. Bondy [:bbondy]]

The problem is with one of the corrupted BMP bitmaps.  The corrupted image should show the background color but instead it shows nothing.
I think I'll have to setup an Android dev environment to fix.

The strange thing is the reftest for this corrupted bitmap passes:

// have valid BPP values before adding the frame
if (mBIH.compression == BI_RLE8 && mBIH.bpp != 8) {
  PR_LOG(gBMPLog, PR_LOG_DEBUG, 
 ("BMP RLE8 compression only supports 8 bits per pixel\n"));
  PostDataError();
  return;
}


But the reftest for this corrupted bitmap fails (which follows directly in the code):

if (mBIH.compression == BI_RLE4 && mBIH.bpp != 4) {
  PR_LOG(gBMPLog, PR_LOG_DEBUG, 
 ("BMP RLE4 compression only supports 4 bits per pixel\n"));
  PostDataError();
  return;
}

Comment 5

[Brian R. Bondy [:bbondy]]

Comment on attachment 556616 [details] [diff] [review]
Patch that re-enables android reftests

Review of attachment 556616 [details] [diff] [review]:
-----------------------------------------------------------------

So I'm not sure which changeset fixed these tests but they are passing now as demonstrated on this recent push:
https://tbpl.mozilla.org/?tree=Try&rev=016a845a547d

Comment 6

[Brian R. Bondy [:bbondy]]

http://hg.mozilla.org/integration/mozilla-inbound/rev/4db54f0cede5

Comment 7

[Brian R. Bondy [:bbondy]]

Now that we know it's actually an intermittent problem, I'll close this as a dupe of the new one and I'll backout the changeset from inbound.

Comment 8

[Brian R. Bondy [:bbondy]]

Backed out here:
http://hg.mozilla.org/integration/mozilla-inbound/rev/2dd6926fcab1

Work for this will continue in bug 753639

Comment 9

[Ed Morley [:emorley]]

(In reply to Brian R. Bondy [:bbondy] from comment #6)
> http://hg.mozilla.org/integration/mozilla-inbound/rev/4db54f0cede5

https://hg.mozilla.org/mozilla-central/rev/4db54f0cede5

Backout to merge shortly.

Comment 10

[Joe Drew (not getting mail)]

backout merged https://hg.mozilla.org/mozilla-central/rev/2dd6926fcab1