| Summary: | fast/images/animated-png.html is crashing / failing on Yosemite | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | Chris Dumez <cdumez> |
| Component: | Platform | Assignee: | Nobody <webkit-unassigned> |
| Status: | NEW --- | ||
| Severity: | Normal | CC: | ap, barraclough, cgarcia, koivisto, simon.fraser, waldyrious |
| Priority: | P2 | ||
| Version: | 528+ (Nightly build) | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
| Bug Depends on: | |||
| Bug Blocks: | 17022 | ||
|
Description
Chris Dumez
2015-03-16 09:27:12 PDT
The crash seems to occur on Yosemite only so I skipped the test for now in: <http://trac.webkit.org/changeset/181557> hmm, mac EWS was green. I wouldn't say this is a regression, r181553 added a new test that crashes in mac, but the code introduced in that revision is not used by mac, right? So whatever causes the crash was already there before r181553, but the new test revealed it. (In reply to comment #2) > hmm, mac EWS was green. I wouldn't say this is a regression, r181553 added a > new test that crashes in mac, but the code introduced in that revision is > not used by mac, right? So whatever causes the crash was already there > before r181553, but the new test revealed it. EWS is running Mavericks I believe and the crash seems to happen on Yosemite only. (In reply to comment #3) > (In reply to comment #2) > > hmm, mac EWS was green. I wouldn't say this is a regression, r181553 added a > > new test that crashes in mac, but the code introduced in that revision is > > not used by mac, right? So whatever causes the crash was already there > > before r181553, but the new test revealed it. > > EWS is running Mavericks I believe and the crash seems to happen on Yosemite > only. aha, that makes sense :-) This test also fails on some bots, as opposed to crashing. Perhaps we should skip it? This may be a bug in ImageIO framework, specific to APNG as it crashes in copyImageBlockSetAPNG(). (In reply to comment #5) > This test also fails on some bots, as opposed to crashing. Perhaps we should > skip it? Alexey is right, it only crashes on Yosemite WK2. On Yosemite WK1, we get the following Image diff: https://build.webkit.org/results/Apple%20Yosemite%20Debug%20WK1%20(Tests)/r181556%20(2993)/fast/images/animated-png-diffs.html I'll update the TestExpectations accordingly. Updated Mac TestExpectations in <http://trac.webkit.org/changeset/181559>. The crash may be rdar://problem/17490843 > The crash may be rdar://problem/17490843
While that's possible, I'm not sure if that explains why crashing is WebKit2 only.
In the past, crashes of this sort were sometimes consequences of bugs in WebKit networking stack.
Also, I'm getting a somewhat different crash log on OS X 10.10.2:
Thread 6 Crashed:
0 com.apple.vImage 0x00007fff881a526a sConvert_Planar1toPlanar8 + 122
1 com.apple.vImage 0x00007fff8836a344 Convert1To8Bit + 212
2 com.apple.vImage 0x00007fff883651e0 AnyToAnyBlock + 1392
3 com.apple.vImage 0x00007fff883649a3 vImageConvert_AnyToAny + 2003
4 com.apple.ImageIO.framework 0x00007fff8d997989 vImageExpandProc + 409
5 com.apple.ImageIO.framework 0x00007fff8d9bbcbd copyImageBlockSetAPNG + 4908
6 com.apple.ImageIO.framework 0x00007fff8d98b9c2 ImageProviderCopyImageBlockSetCallback + 581
7 com.apple.CoreGraphics 0x00007fff857b590d img_blocks_create + 651
8 com.apple.CoreGraphics 0x00007fff85768956 img_data_lock + 2073
9 com.apple.CoreGraphics 0x00007fff857670de CGSImageDataLock + 151
10 libRIP.A.dylib 0x00007fff8968d2d2 ripc_AcquireImage + 906
11 libRIP.A.dylib 0x00007fff8968bdf5 ripc_DrawImage + 1037
12 com.apple.CoreGraphics 0x00007fff85766c97 CGContextDrawImage + 457
13 com.apple.QuartzCore 0x00007fff8d8bac0d CA::CG::ImageDelegate::draw(CGContext*) const + 375
14 com.apple.QuartzCore 0x00007fff8d8b7adb CA::CG::MosaicBitmapDelegate::read(unsigned int, CA::Bounds const&, unsigned char*, long) const + 195
15 com.apple.QuartzCore 0x00007fff8d8ce2ca CA::OGL::Mosaic::draw(CA::OGL::Context&, unsigned int, unsigned int, CA::OGL::Mosaic::Key const*, int, int, bool, CA::OGL::Mosaic::Delegate const&) + 1240
16 com.apple.QuartzCore 0x00007fff8d8bb8e8 CA::CG::fill_image(CA::CG::Renderer&, CGImage*, CA::Rect const&, CA::Mat2<double> const&, bool, bool, CGInterpolationQuality, CA::Bounds const*) + 2287
17 com.apple.QuartzCore 0x00007fff8d8bc18f CA::CG::DrawImage::draw_image(CA::CG::Renderer&, bool) const + 109
18 com.apple.QuartzCore 0x00007fff8d8b8022 CA::CG::DrawOp::render(CA::CG::Renderer&) const + 1172
19 com.apple.QuartzCore 0x00007fff8d8c9e8c CA::CG::Queue::render_callback(void*) + 332
This is not a recent regression, this test crashes in shipping Safari (8.0.3). Further tweaked test expectations in http://trac.webkit.org/r181684. |