Closed Bug 983548 Opened 10 years ago Closed 8 years ago

pdf.js regression takes 6x longer to load HP-Vulnerability_Infographic_v3-03.04.2014.pdf

Categories

(Core :: Graphics: Layers, defect)

30 Branch
All
macOS
defect
Not set
normal

Tracking

()

RESOLVED INCOMPLETE
Tracking Status
firefox29 --- unaffected
firefox30 --- affected

People

(Reporter: cpeterson, Unassigned)

References

Details

(Keywords: perf, regression)

Nightly 30 build 2014-02-12 could load the following PDF in about 5 seconds on my MacBook Pro, but build 2014-02-13 takes over 30 seconds.

http://www.pwn2own.com/wp-content/uploads/2014/03/HP-Vulnerability_Infographic_v3-03.04.2014.pdf

I bisected the performance regression to this pushlog:

https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=41b068af7b8d&tochange=4e9b713f435a

This bug is presumably caused by bug 948765. This bug might be a duplicate of bug 973976, but has different STR and a much bigger performance regression.
At first, I couldn't believe that the PDF stuff had anything to do with my canvas patches, by my god, you're *so* right! :-/

With my patches from Bug 948765, it really is dog slow!

The *real* (long time) fix for this has to be done by someone who understands CoreGraphics better than me. My immediate workaround are the patches on Bug 973976. They bypass the Moz2D code paths, so that it works as before. I tested applying those, and the performance is indeed "back to normal" then.

I'll ask someone to tag that bug with checkin-needed.
Not using CoreGraphics anymore, is this still a problem?
Flags: needinfo?(cpeterson)
Version: unspecified → 30 Branch
Unfortunately, the PDF URL in comment 0 is now 404 so I can verify whether this is still a problem. I should have attached a copy of the PDF to this bug two years ago. :(
Status: NEW → RESOLVED
Closed: 8 years ago
Flags: needinfo?(cpeterson)
Resolution: --- → INCOMPLETE
You need to log in before you can comment on or make changes to this bug.