54299 2011-02-11 11:28:06 -0800 need makeContextCurrent() called in prepareForSoftwareDraw() 2011-02-11 15:22:46 -0800 1 1 1 Unclassified WebKit Canvas 528+ (Nightly build) All All RESOLVED FIXED P2 Normal --- 1 reed webkit-unassigned commit-queue jamesr kbr mdelaney7 reed vangelis oldest_to_newest 349868 0 reed 2011-02-11 11:28:06 -0800 when skia is backed by a gpu, we need to ensure that the current gl context is always set before we render. 349873 1 82154 reed 2011-02-11 11:33:12 -0800 Created attachment 82154 fix for 54299 -- call makeContextCurrent() in prepareForSoftwareDraw() when Skia is backed by the GPU 349881 2 82157 reed 2011-02-11 11:39:24 -0800 Created attachment 82157 fix weird loss of comments in ChangeLog 349887 3 jamesr 2011-02-11 11:44:54 -0800 I don't totally understand this patch - why doesn't the code issuing the GL calls (in this case skia) take care of setting the current context? 349937 4 reed 2011-02-11 12:50:08 -0800 The library explicitly relies on its client to take care of that, since creation of management of clients is platform dependent, and clients of ganesh also can use the gpu directly (ganesh is just there to translate 2D into gpu calls). 349949 5 reed 2011-02-11 13:01:21 -0800 As a 2nd note, the existing GraphicsContext3DOpenGL.cpp issues this call before every gl... call. This patch follows the same logical pattern, but happens to be a bit more efficient, in that we only issue the call before each high-level 2D primitive, which may turn into may gl... calls. 350028 6 82157 jamesr 2011-02-11 14:19:30 -0800 Comment on attachment 82157 fix weird loss of comments in ChangeLog Makes sense. I think the confusing part is that prepareForSoftwareDraw() is misnamed - what it really means in this context is prepareForSkiaDraw(), as opposed to the h/w drawing backend. 350046 7 82157 vangelis 2011-02-11 14:36:13 -0800 Comment on attachment 82157 fix weird loss of comments in ChangeLog I think we need to make the same change to ::prepareForHardwareDraw() as well. In addition we need to replace: #if ENABLE(SKIA_GPU) by #if ENABLE(ACCELERATED_2D_CANVAS) && ENABLE(SKIA_GPU) 350061 8 jamesr 2011-02-11 14:50:57 -0800 (In reply to comment #7) > (From update of attachment 82157 [details]) > I think we need to make the same change to ::prepareForHardwareDraw() as well. In addition we need to replace: > #if ENABLE(SKIA_GPU) by #if ENABLE(ACCELERATED_2D_CANVAS) && ENABLE(SKIA_GPU) prepareForHardwareDraw() is never reached in the skia GPU path 350074 9 commit-queue 2011-02-11 15:04:12 -0800 The commit-queue encountered the following flaky tests while processing attachment 82157: http/tests/websocket/tests/alert-in-event-handler.html bug 54316 (author: abarth@webkit.org) The commit-queue is continuing to process your patch. 350076 10 82157 commit-queue 2011-02-11 15:05:42 -0800 Comment on attachment 82157 fix weird loss of comments in ChangeLog Clearing flags on attachment: 82157 Committed r78377: <http://trac.webkit.org/changeset/78377> 350077 11 commit-queue 2011-02-11 15:05:47 -0800 All reviewed patches have been landed. Closing bug. 350089 12 vangelis 2011-02-11 15:22:46 -0800 (In reply to comment #8) > (In reply to comment #7) > > (From update of attachment 82157 [details] [details]) > > I think we need to make the same change to ::prepareForHardwareDraw() as well. In addition we need to replace: > > #if ENABLE(SKIA_GPU) by #if ENABLE(ACCELERATED_2D_CANVAS) && ENABLE(SKIA_GPU) > > prepareForHardwareDraw() is never reached in the skia GPU path Good point! But then we have another problem... prepareForSoftwareDraw() isn't called before every call that renders via platformContext()->canvas() which means that there are calls (e.g. GraphicsContext::clearRect()) which won't set the current GL context correctly. 82154 2011-02-11 11:33:12 -0800 2011-02-11 11:39:24 -0800 fix for 54299 -- call makeContextCurrent() in prepareForSoftwareDraw() when Skia is backed by the GPU prepare.txt text/plain 1290 reed SW5kZXg6IFdlYkNvcmUvQ2hhbmdlTG9nCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIFdlYkNvcmUvQ2hhbmdlTG9n CShyZXZpc2lvbiA3ODM1MSkKKysrIFdlYkNvcmUvQ2hhbmdlTG9nCSh3b3JraW5nIGNvcHkpCkBA IC0xLDMgKzEsMTQgQEAKKzIwMTEtMDItMTEgIE1pa2UgUmVlZCAgPHJlZWRAZ29vZ2xlLmNvbT4K KworICAgICAgICBSZXZpZXdlZCBieSBOT0JPRFkgKE9PUFMhKS4KKworICAgICAgICBOZWVkIGEg c2hvcnQgZGVzY3JpcHRpb24gYW5kIGJ1ZyBVUkwgKE9PUFMhKQorCisgICAgICAgIE5vIG5ldyB0 ZXN0cy4gKE9PUFMhKQorCisgICAgICAgICogcGxhdGZvcm0vZ3JhcGhpY3Mvc2tpYS9QbGF0Zm9y bUNvbnRleHRTa2lhLmNwcDoKKyAgICAgICAgKFdlYkNvcmU6OlBsYXRmb3JtQ29udGV4dFNraWE6 OnByZXBhcmVGb3JTb2Z0d2FyZURyYXcpOgorCiAyMDExLTAyLTExICBTaGVyaWZmIEJvdCAgPHdl YmtpdC5yZXZpZXcuYm90QGdtYWlsLmNvbT4KIAogICAgICAgICBVbnJldmlld2VkLCByb2xsaW5n IG91dCByNzgzMzEuCkluZGV4OiBXZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL3NraWEvUGxhdGZv cm1Db250ZXh0U2tpYS5jcHAKPT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gV2ViQ29yZS9wbGF0Zm9ybS9ncmFwaGlj cy9za2lhL1BsYXRmb3JtQ29udGV4dFNraWEuY3BwCShyZXZpc2lvbiA3ODMwOSkKKysrIFdlYkNv cmUvcGxhdGZvcm0vZ3JhcGhpY3Mvc2tpYS9QbGF0Zm9ybUNvbnRleHRTa2lhLmNwcAkod29ya2lu ZyBjb3B5KQpAQCAtNzU5LDggKzc1OSwxMyBAQCB2b2lkIFBsYXRmb3JtQ29udGV4dFNraWE6OnNl dFNoYXJlZEdyYXBoCiAKIHZvaWQgUGxhdGZvcm1Db250ZXh0U2tpYTo6cHJlcGFyZUZvclNvZnR3 YXJlRHJhdygpIGNvbnN0CiB7Ci0gICAgaWYgKCFtX3VzZUdQVSkKKyAgICBpZiAoIW1fdXNlR1BV KSB7CisjaWYgRU5BQkxFKFNLSUFfR1BVKQorICAgICAgICBpZiAobV9ncHVDYW52YXMpCisgICAg ICAgICAgICBtX2dwdUNhbnZhcy0+Y29udGV4dCgpLT5tYWtlQ29udGV4dEN1cnJlbnQoKTsKKyNl bmRpZgogICAgICAgICByZXR1cm47CisgICAgfQogCiAgICAgaWYgKG1fYmFja2luZ1N0b3JlU3Rh dGUgPT0gSGFyZHdhcmUpIHsKICAgICAgICAgLy8gRGVwZW5kaW5nIG9uIHRoZSBibGVuZCBtb2Rl IHdlIG5lZWQgdG8gZG8gb25lIG9mIGEgZmV3IHRoaW5nczoK 82157 2011-02-11 11:39:24 -0800 2011-02-11 15:05:42 -0800 fix weird loss of comments in ChangeLog prepare.txt text/plain 1464 reed SW5kZXg6IFdlYkNvcmUvQ2hhbmdlTG9nCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIFdlYkNvcmUvQ2hhbmdlTG9n CShyZXZpc2lvbiA3ODM1MSkKKysrIFdlYkNvcmUvQ2hhbmdlTG9nCSh3b3JraW5nIGNvcHkpCkBA IC0xLDMgKzEsMTUgQEAKKzIwMTEtMDItMTEgIE1pa2UgUmVlZCAgPHJlZWRAZ29vZ2xlLmNvbT4K KworICAgICAgICBSZXZpZXdlZCBieSBOT0JPRFkgKE9PUFMhKS4KKworICAgICAgICBOZWVkIG1h a2VDb250ZXh0Q3VycmVudCgpIGNhbGxlZCBpbiBwcmVwYXJlRm9yU29mdHdhcmVEcmF3KCksIGlu IHRoZSBjYXNlIHRoYXQgc2tpYSdzIGJhY2tlbmQKKyAgICAgICAgaXMgdGhlIGdwdS4gVGhpcyBt YXRjaGVzIHRoZSBwYXR0ZXJuIGluIEdyYXBoaWNzQ29udGV4dDNET3BlbkdMLmNwcAorCisgICAg ICAgIE5vIG5ldyB0ZXN0cy4gQWxsIGV4aXN0aW5nIGNhbnZhcyBsYXlvdXR0ZXN0cyBleGVyY2lz ZSB0aGlzIGNvZGUgcGF0aAorCisgICAgICAgICogcGxhdGZvcm0vZ3JhcGhpY3Mvc2tpYS9QbGF0 Zm9ybUNvbnRleHRTa2lhLmNwcDoKKyAgICAgICAgKFdlYkNvcmU6OlBsYXRmb3JtQ29udGV4dFNr aWE6OnByZXBhcmVGb3JTb2Z0d2FyZURyYXcpOgorCiAyMDExLTAyLTExICBTaGVyaWZmIEJvdCAg PHdlYmtpdC5yZXZpZXcuYm90QGdtYWlsLmNvbT4KIAogICAgICAgICBVbnJldmlld2VkLCByb2xs aW5nIG91dCByNzgzMzEuCkluZGV4OiBXZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL3NraWEvUGxh dGZvcm1Db250ZXh0U2tpYS5jcHAKPT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gV2ViQ29yZS9wbGF0Zm9ybS9ncmFw aGljcy9za2lhL1BsYXRmb3JtQ29udGV4dFNraWEuY3BwCShyZXZpc2lvbiA3ODMwOSkKKysrIFdl YkNvcmUvcGxhdGZvcm0vZ3JhcGhpY3Mvc2tpYS9QbGF0Zm9ybUNvbnRleHRTa2lhLmNwcAkod29y a2luZyBjb3B5KQpAQCAtNzU5LDggKzc1OSwxMyBAQCB2b2lkIFBsYXRmb3JtQ29udGV4dFNraWE6 OnNldFNoYXJlZEdyYXBoCiAKIHZvaWQgUGxhdGZvcm1Db250ZXh0U2tpYTo6cHJlcGFyZUZvclNv ZnR3YXJlRHJhdygpIGNvbnN0CiB7Ci0gICAgaWYgKCFtX3VzZUdQVSkKKyAgICBpZiAoIW1fdXNl R1BVKSB7CisjaWYgRU5BQkxFKFNLSUFfR1BVKQorICAgICAgICBpZiAobV9ncHVDYW52YXMpCisg ICAgICAgICAgICBtX2dwdUNhbnZhcy0+Y29udGV4dCgpLT5tYWtlQ29udGV4dEN1cnJlbnQoKTsK KyNlbmRpZgogICAgICAgICByZXR1cm47CisgICAgfQogCiAgICAgaWYgKG1fYmFja2luZ1N0b3Jl U3RhdGUgPT0gSGFyZHdhcmUpIHsKICAgICAgICAgLy8gRGVwZW5kaW5nIG9uIHRoZSBibGVuZCBt b2RlIHdlIG5lZWQgdG8gZG8gb25lIG9mIGEgZmV3IHRoaW5nczoK