199572 2019-07-08 05:02:55 -0700 REGRESSION(r246671): [WPE][GTK] Crash in NetworkProcess since the DNS cache landed 2019-07-08 11:06:35 -0700 1 1 1 Unclassified WebKit WebKitGTK WebKit Nightly Build Unspecified Unspecified RESOLVED FIXED https://bugs.webkit.org/show_bug.cgi?id=199579 P2 Normal --- 1 csaavedra csaavedra bugs-noreply cgarcia commit-queue mcatanzaro oldest_to_newest 1550794 0 csaavedra 2019-07-08 05:02:55 -0700 I've noticed that several tests have started crashing the network process sporadically and all they have in common is that the crashing starts right after the DNS cache was landed. Opening this bug to track this potential regression. Some stacktraces to exemplify this: https://build.webkit.org/results/GTK%20Linux%2064-bit%20Release%20(Tests)/r247200%20(10899)/imported/w3c/web-platform-tests/html/dom/interfaces.worker-crash-log.txt https://build.webkit.org/results/GTK%20Linux%2064-bit%20Release%20(Tests)/r247200%20(10899)/imported/w3c/web-platform-tests/cors/redirect-preflight-2-crash-log.txt https://build.webkit.org/results/GTK%20Linux%2064-bit%20Release%20(Tests)/r247200%20(10899)/imported/w3c/web-platform-tests/websockets/interfaces/WebSocket/readyState/005-crash-log.txt https://build.webkit.org/results/GTK%20Linux%2064-bit%20Release%20(Tests)/r247200%20(10899)/imported/w3c/web-platform-tests/websockets/interfaces/WebSocket/url/006-crash-log.txt https://build.webkit.org/results/GTK%20Linux%2064-bit%20Release%20(Tests)/r247200%20(10899)/imported/w3c/web-platform-tests/websockets/keeping-connection-open/001-crash-log.txt https://build.webkit.org/results/GTK%20Linux%2064-bit%20Release%20(Tests)/r247200%20(10899)/imported/w3c/web-platform-tests/websockets/opening-handshake/005-crash-log.txt https://build.webkit.org/results/GTK%20Linux%2064-bit%20Release%20(Tests)/r247204%20(10901)/imported/w3c/web-platform-tests/content-security-policy/connect-src/connect-src-websocket-blocked.sub-crash-log.txt The stacktraces don't tell much at first sight, but since it seems to be happening deep in glib, there might be some memory corruption somewhere, not sure if related to thread-safety issues in the dns cache resolver (or libsoup even?). 1550796 1 csaavedra 2019-07-08 05:06:16 -0700 The historical of the crashes: https://webkit-test-results.webkit.org/dashboards/flakiness_dashboard.html#showAllRuns=true&tests=imported%2Fw3c%2Fweb-platform-tests%2Fcontent-security-policy%2Fconnect-src%2Fconnect-src-websocket-blocked.sub.html%20imported%2Fw3c%2Fweb-platform-tests%2Fwebsockets%2Fopening-handshake%20imported%2Fw3c%2Fweb-platform-tests%2Fwebsockets%2Fkeeping-connection-open%20imported%2Fw3c%2Fweb-platform-tests%2Fwebsockets%2Finterfaces%2FWebSocket%2Furl%2F006%20imported%2Fw3c%2Fweb-platform-tests%2Fwebsockets%2Finterfaces%2FWebSocket%2FreadyState%2F005 1550798 2 mcatanzaro 2019-07-08 06:27:51 -0700 At least most of these (anything involving magazine_chain_pop_head) are definitely memory corruption. Probably the rest are too. The backtraces are only going to lead you on a false chase. What we need are backtraces from asan or valgrind to point to the source of the memory corruption, not these gdb traces that only point to where it eventually blew up. 1550799 3 mcatanzaro 2019-07-08 06:30:06 -0700 BTW if we don't solve it soon, then we should roll it out. The DNS cache is only a small thing, not worth a serious regression. 1550806 4 csaavedra 2019-07-08 07:55:53 -0700 ==27300== Thread 1: ==27300== Invalid read of size 8 ==27300== at 0xE4B5029: g_error_free (gerror.c:493) ==27300== by 0x6428719: webkitCachedResolverLookupByNameAsync(_GResolver*, char const*, _GCancellable*, void (*)(_GObject*, _GAsyncResult*, void*), void*)::{lambda(_GObject*, _GAsyncResult*, void*)#1}::_FUN(_ GObject*, _GAsyncResult*, void*) (in /home/claudio/git/gnome/WebKit/WebKitBuild/Release/lib/libwebkit2gtk-4.0.so.37.38.1) ==27300== by 0xED40AF8: g_task_return_now (gtask.c:1148) ==27300== by 0xED40B38: complete_in_idle_cb (gtask.c:1162) ==27300== by 0xE4C99B7: g_main_dispatch (gmain.c:3182) ==27300== by 0xE4C99B7: g_main_context_dispatch (gmain.c:3847) ==27300== by 0xE4C9D77: g_main_context_iterate.isra.26 (gmain.c:3920) ==27300== by 0xE4CA061: g_main_loop_run (gmain.c:4116) ==27300== by 0xAAEEA2F: WTF::RunLoop::run() (in /home/claudio/git/gnome/WebKit/WebKitBuild/Release/lib/libjavascriptcoregtk-4.0.so.18.14.1) ==27300== by 0x6430329: int WebKit::AuxiliaryProcessMain<WebKit::NetworkProcess, WebKit::NetworkProcessMain>(int, char**) (in /home/claudio/git/gnome/WebKit/WebKitBuild/Release/lib/libwebkit2gtk-4.0.so.37.38. 1) ==27300== by 0xFC7B09A: (below main) (libc-start.c:308) ==27300== Address 0x5c749bd8 is 8 bytes inside a block of size 16 free'd ==27300== at 0x48369AB: free (vg_replace_malloc.c:530) ==27300== by 0xE205350: lookup_resolved (soup-address.c:762) ==27300== by 0xED40AF8: g_task_return_now (gtask.c:1148) ==27300== by 0xED41575: g_task_return (gtask.c:1206) ==27300== by 0x6428744: webkitCachedResolverLookupByNameAsync(_GResolver*, char const*, _GCancellable*, void (*)(_GObject*, _GAsyncResult*, void*), void*)::{lambda(_GObject*, _GAsyncResult*, void*)#1}::_FUN(_ GObject*, _GAsyncResult*, void*) (in /home/claudio/git/gnome/WebKit/WebKitBuild/Release/lib/libwebkit2gtk-4.0.so.37.38.1) ==27300== by 0xED40AF8: g_task_return_now (gtask.c:1148) ==27300== by 0xED40B38: complete_in_idle_cb (gtask.c:1162) ==27300== by 0xE4C99B7: g_main_dispatch (gmain.c:3182) ==27300== by 0xE4C99B7: g_main_context_dispatch (gmain.c:3847) ==27300== by 0xE4C9D77: g_main_context_iterate.isra.26 (gmain.c:3920) ==27300== by 0xE4CA061: g_main_loop_run (gmain.c:4116) ==27300== by 0xAAEEA2F: WTF::RunLoop::run() (in /home/claudio/git/gnome/WebKit/WebKitBuild/Release/lib/libjavascriptcoregtk-4.0.so.18.14.1) ==27300== by 0x6430329: int WebKit::AuxiliaryProcessMain<WebKit::NetworkProcess, WebKit::NetworkProcessMain>(int, char**) (in /home/claudio/git/gnome/WebKit/WebKitBuild/Release/lib/libwebkit2gtk-4.0.so.37.38. 1) ==27300== by 0xFC7B09A: (below main) (libc-start.c:308) ==27300== Block was alloc'd at ==27300== at 0x483577F: malloc (vg_replace_malloc.c:299) ==27300== by 0xE4CF2E0: g_malloc (gmem.c:99) ==27300== by 0xE4E6782: g_slice_alloc (gslice.c:1024) ==27300== by 0xE4B4E2C: g_error_new_valist (gerror.c:410) ==27300== by 0xED42077: g_task_return_new_error (gtask.c:1811) ==27300== by 0xED44C75: do_lookup_by_name (gthreadedresolver.c:123) ==27300== by 0xED416A2: g_task_thread_pool_thread (gtask.c:1331) ==27300== by 0xE4F19B2: g_thread_pool_thread_proxy (gthreadpool.c:307) ==27300== by 0xE4F1054: g_thread_proxy (gthread.c:784) ==27300== by 0xDD5BFA2: start_thread (pthread_create.c:486) ==27300== by 0xFD504CE: clone (clone.S:95) 1550807 5 373625 csaavedra 2019-07-08 08:01:12 -0700 Created attachment 373625 Patch 1550809 6 373625 mcatanzaro 2019-07-08 08:22:57 -0700 Comment on attachment 373625 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=373625&action=review Score +1 for valgrind > Source/WebKit/NetworkProcess/glib/WebKitCachedResolver.cpp:111 > - g_task_return_error(task.get(), error.get()); > + g_task_return_error(task.get(), error); My preference is to keep using GUniqueOutPtr -- so that we never hold ownership in a raw pointer -- and use error.release() here instead. But this is fine too. > Source/WebKit/NetworkProcess/glib/WebKitCachedResolver.cpp:173 > - g_task_return_error(task.get(), error.get()); > + g_task_return_error(task.get(), error); Ditto. 1550812 7 373625 cgarcia 2019-07-08 08:36:13 -0700 Comment on attachment 373625 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=373625&action=review >> Source/WebKit/NetworkProcess/glib/WebKitCachedResolver.cpp:111 >> + g_task_return_error(task.get(), error); > > My preference is to keep using GUniqueOutPtr -- so that we never hold ownership in a raw pointer -- and use error.release() here instead. But this is fine too. Good catch, thanks Claudio! I would also keep the smart pointer leaking it only when transferred. > Source/WebKit/NetworkProcess/glib/WebKitCachedResolver.cpp:166 > + GError* error; This is missing the nullptr initialization. 1550819 8 373625 mcatanzaro 2019-07-08 08:59:39 -0700 Comment on attachment 373625 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=373625&action=review >>> Source/WebKit/NetworkProcess/glib/WebKitCachedResolver.cpp:111 >>> + g_task_return_error(task.get(), error); >> >> My preference is to keep using GUniqueOutPtr -- so that we never hold ownership in a raw pointer -- and use error.release() here instead. But this is fine too. > > Good catch, thanks Claudio! I would also keep the smart pointer leaking it only when transferred. Claudio noticed that GUniqueOutPtr::release returns a GUniquePtr, so it would have to be error.release().release(), which is silly. So I think the smart pointer is undesirable here. I wonder if we have much code that really benefits from that behavior, or if we could change GUniqueOutPtr::release to return the raw pointer directly, as I would have expected. >> Source/WebKit/NetworkProcess/glib/WebKitCachedResolver.cpp:166 >> + GError* error; > > This is missing the nullptr initialization. Oh no, very good catch. 1550820 9 373629 csaavedra 2019-07-08 09:00:28 -0700 Created attachment 373629 Patch 1550824 10 373629 mcatanzaro 2019-07-08 09:07:57 -0700 Comment on attachment 373629 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=373629&action=review cq+ is going to fail because you don't have the Reviewed by > Source/WebKit/ChangeLog:10 > + but passed onto the caller. > + * NetworkProcess/glib/WebKitCachedResolver.cpp: Missing blank line here. > Source/WebKit/NetworkProcess/glib/WebKitCachedResolver.cpp:111 > - g_task_return_error(task.get(), error.get()); > + g_task_return_error(task.get(), error.release().release()); Carlos, what do you think? 1550826 11 373631 csaavedra 2019-07-08 09:11:35 -0700 Created attachment 373631 Patch for landing 1550827 12 cgarcia 2019-07-08 09:12:18 -0700 (In reply to Michael Catanzaro from comment #8) > Comment on attachment 373625 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=373625&action=review > > >>> Source/WebKit/NetworkProcess/glib/WebKitCachedResolver.cpp:111 > >>> + g_task_return_error(task.get(), error); > >> > >> My preference is to keep using GUniqueOutPtr -- so that we never hold ownership in a raw pointer -- and use error.release() here instead. But this is fine too. > > > > Good catch, thanks Claudio! I would also keep the smart pointer leaking it only when transferred. > > Claudio noticed that GUniqueOutPtr::release returns a GUniquePtr, so it > would have to be error.release().release(), which is silly. So I think the > smart pointer is undesirable here. > > I wonder if we have much code that really benefits from that behavior, or if > we could change GUniqueOutPtr::release to return the raw pointer directly, > as I would have expected. I think the idea here was to convert an OutPtr (that exists only because outPtr is not possible with std::unique_ptr) into a Ptr which is actually a std::unique_ptr. I don't know if we are actually using it that way. We could probably add leakPtr() for the release().release() case. > >> Source/WebKit/NetworkProcess/glib/WebKitCachedResolver.cpp:166 > >> + GError* error; > > > > This is missing the nullptr initialization. > > Oh no, very good catch. 1550841 13 csaavedra 2019-07-08 09:40:09 -0700 Committed r247209: <https://trac.webkit.org/changeset/247209> 1550878 14 mcatanzaro 2019-07-08 11:06:24 -0700 (In reply to Carlos Garcia Campos from comment #12) > I think the idea here was to convert an OutPtr (that exists only because > outPtr is not possible with std::unique_ptr) into a Ptr which is actually a > std::unique_ptr. I don't know if we are actually using it that way. We could > probably add leakPtr() for the release().release() case. There were only three places using release before now. Two of them also want the raw pointer (they're doing release().release()) and the other is test code. Making a GUniquePtr from the raw pointer in that case is trivial: diff --git a/Tools/TestWebKitAPI/Tests/WTF/glib/GUniquePtr.cpp b/Tools/TestWebKitAPI/Tests/WTF/glib/GUniquePtr.cpp index 11f02c7b0f7..288591c8e3f 100644 --- a/Tools/TestWebKitAPI/Tests/WTF/glib/GUniquePtr.cpp +++ b/Tools/TestWebKitAPI/Tests/WTF/glib/GUniquePtr.cpp @@ -184,7 +184,7 @@ TEST(WTF_GUniquePtr, OutPtr) { GUniqueOutPtr<char> a; returnOutChar(&a.outPtr()); - GUniquePtr<char> b = a.release(); + GUniquePtr<char> b(a.release()); ASSERT_STREQ(actual.str().c_str(), takeLogStr().c_str()); actual << "g_free(" << b.get() << ");"; } So my vote is to just change release(). 1550879 15 mcatanzaro 2019-07-08 11:06:35 -0700 Bug #199579. 373625 2019-07-08 08:01:12 -0700 2019-07-08 09:00:24 -0700 Patch bug-199572-20190708180111.patch text/plain 3906 csaavedra U3VidmVyc2lvbiBSZXZpc2lvbjogMjQ3MjAzCmRpZmYgLS1naXQgYS9Tb3VyY2UvV2ViS2l0L0No YW5nZUxvZyBiL1NvdXJjZS9XZWJLaXQvQ2hhbmdlTG9nCmluZGV4IDdkNjU2ZTA0NTkyZTE3ZjQ1 YWQxY2QwYmE1YTViZGRkZTE5ZmM0MzkuLjNlOTY0ZTNkNjJlNmM0MWIyN2YyNTM4NGJlZjM4YWUw ZTM3MWE1OTEgMTAwNjQ0Ci0tLSBhL1NvdXJjZS9XZWJLaXQvQ2hhbmdlTG9nCisrKyBiL1NvdXJj ZS9XZWJLaXQvQ2hhbmdlTG9nCkBAIC0xLDMgKzEsMTYgQEAKKzIwMTktMDctMDggIENsYXVkaW8g U2FhdmVkcmEgIDxjc2FhdmVkcmFAaWdhbGlhLmNvbT4KKworICAgICAgICBSRUdSRVNTSU9OKHIy NDY2NzEpOiBbV1BFXVtHVEtdIENyYXNoIGluIE5ldHdvcmtQcm9jZXNzIHNpbmNlIHRoZSBETlMg Y2FjaGUgbGFuZGVkCisgICAgICAgIGh0dHBzOi8vYnVncy53ZWJraXQub3JnL3Nob3dfYnVnLmNn aT9pZD0xOTk1NzIKKworICAgICAgICBSZXZpZXdlZCBieSBOT0JPRFkgKE9PUFMhKS4KKworICAg ICAgICBUaGUgR0Vycm9yIGNvbWluZyBmcm9tIHRoZSB3cmFwcGVkIHJlc29sdmVyIHNob3VsZG4n dCBiZSBmcmVlZCwKKyAgICAgICAgYnV0IHBhc3NlZCBvbnRvIHRoZSBjYWxsZXIuCisgICAgICAg ICogTmV0d29ya1Byb2Nlc3MvZ2xpYi9XZWJLaXRDYWNoZWRSZXNvbHZlci5jcHA6CisgICAgICAg ICh3ZWJraXRDYWNoZWRSZXNvbHZlckxvb2t1cEJ5TmFtZUFzeW5jKToKKyAgICAgICAgKHdlYmtp dENhY2hlZFJlc29sdmVyTG9va3VwQnlOYW1lV2l0aEZsYWdzQXN5bmMpOgorCiAyMDE5LTA3LTA4 ICBNaWNoYWVsIENhdGFuemFybyAgPG1jYXRhbnphcm9AaWdhbGlhLmNvbT4KIAogICAgICAgICBb R1RLXSBWaWV3R2VzdHVyZUNvbnRyb2xsZXIgY29uc3RydWN0b3IgbGVhdmVzIG1fZGlyZWN0aW9u IHVuaW5pdGlhbGl6ZWQKZGlmZiAtLWdpdCBhL1NvdXJjZS9XZWJLaXQvTmV0d29ya1Byb2Nlc3Mv Z2xpYi9XZWJLaXRDYWNoZWRSZXNvbHZlci5jcHAgYi9Tb3VyY2UvV2ViS2l0L05ldHdvcmtQcm9j ZXNzL2dsaWIvV2ViS2l0Q2FjaGVkUmVzb2x2ZXIuY3BwCmluZGV4IGM1NDRiNjU5MDYyNTJhZmJl ZmE3MGJlMGQ3ZmQ2YmNiNTVhMDkyZTguLjdjZTgzMjM2ODliNzMwYjY0ZGI4M2RlNGRjZTBmZTZh MDdlMTNjYjIgMTAwNjQ0Ci0tLSBhL1NvdXJjZS9XZWJLaXQvTmV0d29ya1Byb2Nlc3MvZ2xpYi9X ZWJLaXRDYWNoZWRSZXNvbHZlci5jcHAKKysrIGIvU291cmNlL1dlYktpdC9OZXR3b3JrUHJvY2Vz cy9nbGliL1dlYktpdENhY2hlZFJlc29sdmVyLmNwcApAQCAtMTAxLDE0ICsxMDEsMTQgQEAgc3Rh dGljIHZvaWQgd2Via2l0Q2FjaGVkUmVzb2x2ZXJMb29rdXBCeU5hbWVBc3luYyhHUmVzb2x2ZXIq IHJlc29sdmVyLCBjb25zdCBjaGEKICAgICBnX3Rhc2tfc2V0X3Rhc2tfZGF0YSh0YXNrLmdldCgp LCBhc3luY0RhdGEsIHJlaW50ZXJwcmV0X2Nhc3Q8R0Rlc3Ryb3lOb3RpZnk+KGRlc3Ryb3lMb29r dXBBc3luY0RhdGEpKTsKICAgICBnX3Jlc29sdmVyX2xvb2t1cF9ieV9uYW1lX2FzeW5jKHByaXYt PndyYXBwZWRSZXNvbHZlci5nZXQoKSwgaG9zdG5hbWUsIGNhbmNlbGxhYmxlLCBbXShHT2JqZWN0 KiByZXNvbHZlciwgR0FzeW5jUmVzdWx0KiByZXN1bHQsIGdwb2ludGVyIHVzZXJEYXRhKSB7CiAg ICAgICAgIEdSZWZQdHI8R1Rhc2s+IHRhc2sgPSBhZG9wdEdSZWYoR19UQVNLKHVzZXJEYXRhKSk7 Ci0gICAgICAgIEdVbmlxdWVPdXRQdHI8R0Vycm9yPiBlcnJvcjsKLSAgICAgICAgaWYgKGF1dG8q IGFkZHJlc3NMaXN0ID0gZ19yZXNvbHZlcl9sb29rdXBfYnlfbmFtZV9maW5pc2goR19SRVNPTFZF UihyZXNvbHZlciksIHJlc3VsdCwgJmVycm9yLm91dFB0cigpKSkgeworICAgICAgICBHRXJyb3Iq IGVycm9yID0gbnVsbHB0cjsKKyAgICAgICAgaWYgKGF1dG8qIGFkZHJlc3NMaXN0ID0gZ19yZXNv bHZlcl9sb29rdXBfYnlfbmFtZV9maW5pc2goR19SRVNPTFZFUihyZXNvbHZlciksIHJlc3VsdCwg JmVycm9yKSkgewogICAgICAgICAgICAgYXV0byogcHJpdiA9IFdFQktJVF9DQUNIRURfUkVTT0xW RVIoZ190YXNrX2dldF9zb3VyY2Vfb2JqZWN0KHRhc2suZ2V0KCkpKS0+cHJpdjsKICAgICAgICAg ICAgIGF1dG8qIGFzeW5jRGF0YSA9IHN0YXRpY19jYXN0PExvb2t1cEFzeW5jRGF0YSo+KGdfdGFz a19nZXRfdGFza19kYXRhKHRhc2suZ2V0KCkpKTsKICAgICAgICAgICAgIHByaXYtPmNhY2hlLnVw ZGF0ZShhc3luY0RhdGEtPmhvc3RuYW1lLCBhZGRyZXNzTGlzdEdMaXN0VG9WZWN0b3IoYWRkcmVz c0xpc3QpKTsKICAgICAgICAgICAgIGdfdGFza19yZXR1cm5fcG9pbnRlcih0YXNrLmdldCgpLCBh ZGRyZXNzTGlzdCwgcmVpbnRlcnByZXRfY2FzdDxHRGVzdHJveU5vdGlmeT4oZ19yZXNvbHZlcl9m cmVlX2FkZHJlc3NlcykpOwogICAgICAgICB9IGVsc2UKLSAgICAgICAgICAgIGdfdGFza19yZXR1 cm5fZXJyb3IodGFzay5nZXQoKSwgZXJyb3IuZ2V0KCkpOworICAgICAgICAgICAgZ190YXNrX3Jl dHVybl9lcnJvcih0YXNrLmdldCgpLCBlcnJvcik7CiAgICAgfSwgdGFzay5sZWFrUmVmKCkpOwog fQogCkBAIC0xNjMsMTQgKzE2MywxNCBAQCBzdGF0aWMgdm9pZCB3ZWJraXRDYWNoZWRSZXNvbHZl ckxvb2t1cEJ5TmFtZVdpdGhGbGFnc0FzeW5jKEdSZXNvbHZlciogcmVzb2x2ZXIsCiAgICAgZ190 YXNrX3NldF90YXNrX2RhdGEodGFzay5nZXQoKSwgYXN5bmNEYXRhLCByZWludGVycHJldF9jYXN0 PEdEZXN0cm95Tm90aWZ5PihkZXN0cm95TG9va3VwQXN5bmNEYXRhKSk7CiAgICAgZ19yZXNvbHZl cl9sb29rdXBfYnlfbmFtZV93aXRoX2ZsYWdzX2FzeW5jKHByaXYtPndyYXBwZWRSZXNvbHZlci5n ZXQoKSwgaG9zdG5hbWUsIGZsYWdzLCBjYW5jZWxsYWJsZSwgW10oR09iamVjdCogcmVzb2x2ZXIs IEdBc3luY1Jlc3VsdCogcmVzdWx0LCBncG9pbnRlciB1c2VyRGF0YSkgewogICAgICAgICBHUmVm UHRyPEdUYXNrPiB0YXNrID0gYWRvcHRHUmVmKEdfVEFTSyh1c2VyRGF0YSkpOwotICAgICAgICBH VW5pcXVlT3V0UHRyPEdFcnJvcj4gZXJyb3I7Ci0gICAgICAgIGlmIChhdXRvKiBhZGRyZXNzTGlz dCA9IGdfcmVzb2x2ZXJfbG9va3VwX2J5X25hbWVfd2l0aF9mbGFnc19maW5pc2goR19SRVNPTFZF UihyZXNvbHZlciksIHJlc3VsdCwgJmVycm9yLm91dFB0cigpKSkgeworICAgICAgICBHRXJyb3Iq IGVycm9yOworICAgICAgICBpZiAoYXV0byogYWRkcmVzc0xpc3QgPSBnX3Jlc29sdmVyX2xvb2t1 cF9ieV9uYW1lX3dpdGhfZmxhZ3NfZmluaXNoKEdfUkVTT0xWRVIocmVzb2x2ZXIpLCByZXN1bHQs ICZlcnJvcikpIHsKICAgICAgICAgICAgIGF1dG8qIHByaXYgPSBXRUJLSVRfQ0FDSEVEX1JFU09M VkVSKGdfdGFza19nZXRfc291cmNlX29iamVjdCh0YXNrLmdldCgpKSktPnByaXY7CiAgICAgICAg ICAgICBhdXRvKiBhc3luY0RhdGEgPSBzdGF0aWNfY2FzdDxMb29rdXBBc3luY0RhdGEqPihnX3Rh c2tfZ2V0X3Rhc2tfZGF0YSh0YXNrLmdldCgpKSk7CiAgICAgICAgICAgICBwcml2LT5jYWNoZS51 cGRhdGUoYXN5bmNEYXRhLT5ob3N0bmFtZSwgYWRkcmVzc0xpc3RHTGlzdFRvVmVjdG9yKGFkZHJl c3NMaXN0KSwgYXN5bmNEYXRhLT5kbnNDYWNoZVR5cGUpOwogICAgICAgICAgICAgZ190YXNrX3Jl dHVybl9wb2ludGVyKHRhc2suZ2V0KCksIGFkZHJlc3NMaXN0LCByZWludGVycHJldF9jYXN0PEdE ZXN0cm95Tm90aWZ5PihnX3Jlc29sdmVyX2ZyZWVfYWRkcmVzc2VzKSk7CiAgICAgICAgIH0gZWxz ZQotICAgICAgICAgICAgZ190YXNrX3JldHVybl9lcnJvcih0YXNrLmdldCgpLCBlcnJvci5nZXQo KSk7CisgICAgICAgICAgICBnX3Rhc2tfcmV0dXJuX2Vycm9yKHRhc2suZ2V0KCksIGVycm9yKTsK ICAgICB9LCB0YXNrLmxlYWtSZWYoKSk7CiB9CiAK 373629 2019-07-08 09:00:28 -0700 2019-07-08 09:11:31 -0700 Patch bug-199572-20190708190027.patch text/plain 2298 csaavedra U3VidmVyc2lvbiBSZXZpc2lvbjogMjQ3MjAzCmRpZmYgLS1naXQgYS9Tb3VyY2UvV2ViS2l0L0No YW5nZUxvZyBiL1NvdXJjZS9XZWJLaXQvQ2hhbmdlTG9nCmluZGV4IDdkNjU2ZTA0NTkyZTE3ZjQ1 YWQxY2QwYmE1YTViZGRkZTE5ZmM0MzkuLjNlOTY0ZTNkNjJlNmM0MWIyN2YyNTM4NGJlZjM4YWUw ZTM3MWE1OTEgMTAwNjQ0Ci0tLSBhL1NvdXJjZS9XZWJLaXQvQ2hhbmdlTG9nCisrKyBiL1NvdXJj ZS9XZWJLaXQvQ2hhbmdlTG9nCkBAIC0xLDMgKzEsMTYgQEAKKzIwMTktMDctMDggIENsYXVkaW8g U2FhdmVkcmEgIDxjc2FhdmVkcmFAaWdhbGlhLmNvbT4KKworICAgICAgICBSRUdSRVNTSU9OKHIy NDY2NzEpOiBbV1BFXVtHVEtdIENyYXNoIGluIE5ldHdvcmtQcm9jZXNzIHNpbmNlIHRoZSBETlMg Y2FjaGUgbGFuZGVkCisgICAgICAgIGh0dHBzOi8vYnVncy53ZWJraXQub3JnL3Nob3dfYnVnLmNn aT9pZD0xOTk1NzIKKworICAgICAgICBSZXZpZXdlZCBieSBOT0JPRFkgKE9PUFMhKS4KKworICAg ICAgICBUaGUgR0Vycm9yIGNvbWluZyBmcm9tIHRoZSB3cmFwcGVkIHJlc29sdmVyIHNob3VsZG4n dCBiZSBmcmVlZCwKKyAgICAgICAgYnV0IHBhc3NlZCBvbnRvIHRoZSBjYWxsZXIuCisgICAgICAg ICogTmV0d29ya1Byb2Nlc3MvZ2xpYi9XZWJLaXRDYWNoZWRSZXNvbHZlci5jcHA6CisgICAgICAg ICh3ZWJraXRDYWNoZWRSZXNvbHZlckxvb2t1cEJ5TmFtZUFzeW5jKToKKyAgICAgICAgKHdlYmtp dENhY2hlZFJlc29sdmVyTG9va3VwQnlOYW1lV2l0aEZsYWdzQXN5bmMpOgorCiAyMDE5LTA3LTA4 ICBNaWNoYWVsIENhdGFuemFybyAgPG1jYXRhbnphcm9AaWdhbGlhLmNvbT4KIAogICAgICAgICBb R1RLXSBWaWV3R2VzdHVyZUNvbnRyb2xsZXIgY29uc3RydWN0b3IgbGVhdmVzIG1fZGlyZWN0aW9u IHVuaW5pdGlhbGl6ZWQKZGlmZiAtLWdpdCBhL1NvdXJjZS9XZWJLaXQvTmV0d29ya1Byb2Nlc3Mv Z2xpYi9XZWJLaXRDYWNoZWRSZXNvbHZlci5jcHAgYi9Tb3VyY2UvV2ViS2l0L05ldHdvcmtQcm9j ZXNzL2dsaWIvV2ViS2l0Q2FjaGVkUmVzb2x2ZXIuY3BwCmluZGV4IGM1NDRiNjU5MDYyNTJhZmJl ZmE3MGJlMGQ3ZmQ2YmNiNTVhMDkyZTguLjNkZWFiOWU0NmVkMWZmZDg1ZDFhZjIwMjJiNTFhZGIx ZmJiMjFkZmUgMTAwNjQ0Ci0tLSBhL1NvdXJjZS9XZWJLaXQvTmV0d29ya1Byb2Nlc3MvZ2xpYi9X ZWJLaXRDYWNoZWRSZXNvbHZlci5jcHAKKysrIGIvU291cmNlL1dlYktpdC9OZXR3b3JrUHJvY2Vz cy9nbGliL1dlYktpdENhY2hlZFJlc29sdmVyLmNwcApAQCAtMTA4LDcgKzEwOCw3IEBAIHN0YXRp YyB2b2lkIHdlYmtpdENhY2hlZFJlc29sdmVyTG9va3VwQnlOYW1lQXN5bmMoR1Jlc29sdmVyKiBy ZXNvbHZlciwgY29uc3QgY2hhCiAgICAgICAgICAgICBwcml2LT5jYWNoZS51cGRhdGUoYXN5bmNE YXRhLT5ob3N0bmFtZSwgYWRkcmVzc0xpc3RHTGlzdFRvVmVjdG9yKGFkZHJlc3NMaXN0KSk7CiAg ICAgICAgICAgICBnX3Rhc2tfcmV0dXJuX3BvaW50ZXIodGFzay5nZXQoKSwgYWRkcmVzc0xpc3Qs IHJlaW50ZXJwcmV0X2Nhc3Q8R0Rlc3Ryb3lOb3RpZnk+KGdfcmVzb2x2ZXJfZnJlZV9hZGRyZXNz ZXMpKTsKICAgICAgICAgfSBlbHNlCi0gICAgICAgICAgICBnX3Rhc2tfcmV0dXJuX2Vycm9yKHRh c2suZ2V0KCksIGVycm9yLmdldCgpKTsKKyAgICAgICAgICAgIGdfdGFza19yZXR1cm5fZXJyb3Io dGFzay5nZXQoKSwgZXJyb3IucmVsZWFzZSgpLnJlbGVhc2UoKSk7CiAgICAgfSwgdGFzay5sZWFr UmVmKCkpOwogfQogCkBAIC0xNzAsNyArMTcwLDcgQEAgc3RhdGljIHZvaWQgd2Via2l0Q2FjaGVk UmVzb2x2ZXJMb29rdXBCeU5hbWVXaXRoRmxhZ3NBc3luYyhHUmVzb2x2ZXIqIHJlc29sdmVyLAog ICAgICAgICAgICAgcHJpdi0+Y2FjaGUudXBkYXRlKGFzeW5jRGF0YS0+aG9zdG5hbWUsIGFkZHJl c3NMaXN0R0xpc3RUb1ZlY3RvcihhZGRyZXNzTGlzdCksIGFzeW5jRGF0YS0+ZG5zQ2FjaGVUeXBl KTsKICAgICAgICAgICAgIGdfdGFza19yZXR1cm5fcG9pbnRlcih0YXNrLmdldCgpLCBhZGRyZXNz TGlzdCwgcmVpbnRlcnByZXRfY2FzdDxHRGVzdHJveU5vdGlmeT4oZ19yZXNvbHZlcl9mcmVlX2Fk ZHJlc3NlcykpOwogICAgICAgICB9IGVsc2UKLSAgICAgICAgICAgIGdfdGFza19yZXR1cm5fZXJy b3IodGFzay5nZXQoKSwgZXJyb3IuZ2V0KCkpOworICAgICAgICAgICAgZ190YXNrX3JldHVybl9l cnJvcih0YXNrLmdldCgpLCBlcnJvci5yZWxlYXNlKCkucmVsZWFzZSgpKTsKICAgICB9LCB0YXNr LmxlYWtSZWYoKSk7CiB9CiAK 373631 2019-07-08 09:11:35 -0700 2019-07-08 09:36:19 -0700 Patch for landing bug-199572-20190708191134.patch text/plain 2300 csaavedra U3VidmVyc2lvbiBSZXZpc2lvbjogMjQ3MjAzCmRpZmYgLS1naXQgYS9Tb3VyY2UvV2ViS2l0L0No YW5nZUxvZyBiL1NvdXJjZS9XZWJLaXQvQ2hhbmdlTG9nCmluZGV4IDdkNjU2ZTA0NTkyZTE3ZjQ1 YWQxY2QwYmE1YTViZGRkZTE5ZmM0MzkuLmIwZmNiY2Q1OTY5ZjM5YmY3ZjBlN2U4Zjg0NDg3NDA0 YTU1Y2U5M2MgMTAwNjQ0Ci0tLSBhL1NvdXJjZS9XZWJLaXQvQ2hhbmdlTG9nCisrKyBiL1NvdXJj ZS9XZWJLaXQvQ2hhbmdlTG9nCkBAIC0xLDMgKzEsMTcgQEAKKzIwMTktMDctMDggIENsYXVkaW8g U2FhdmVkcmEgIDxjc2FhdmVkcmFAaWdhbGlhLmNvbT4KKworICAgICAgICBSRUdSRVNTSU9OKHIy NDY2NzEpOiBbV1BFXVtHVEtdIENyYXNoIGluIE5ldHdvcmtQcm9jZXNzIHNpbmNlIHRoZSBETlMg Y2FjaGUgbGFuZGVkCisgICAgICAgIGh0dHBzOi8vYnVncy53ZWJraXQub3JnL3Nob3dfYnVnLmNn aT9pZD0xOTk1NzIKKworICAgICAgICBSZXZpZXdlZCBieSBOT0JPRFkgKE9PUFMhKS4KKworICAg ICAgICBUaGUgR0Vycm9yIGNvbWluZyBmcm9tIHRoZSB3cmFwcGVkIHJlc29sdmVyIHNob3VsZG4n dCBiZSBmcmVlZCwKKyAgICAgICAgYnV0IHBhc3NlZCBvbnRvIHRoZSBjYWxsZXIuCisKKyAgICAg ICAgKiBOZXR3b3JrUHJvY2Vzcy9nbGliL1dlYktpdENhY2hlZFJlc29sdmVyLmNwcDoKKyAgICAg ICAgKHdlYmtpdENhY2hlZFJlc29sdmVyTG9va3VwQnlOYW1lQXN5bmMpOgorICAgICAgICAod2Vi a2l0Q2FjaGVkUmVzb2x2ZXJMb29rdXBCeU5hbWVXaXRoRmxhZ3NBc3luYyk6CisKIDIwMTktMDct MDggIE1pY2hhZWwgQ2F0YW56YXJvICA8bWNhdGFuemFyb0BpZ2FsaWEuY29tPgogCiAgICAgICAg IFtHVEtdIFZpZXdHZXN0dXJlQ29udHJvbGxlciBjb25zdHJ1Y3RvciBsZWF2ZXMgbV9kaXJlY3Rp b24gdW5pbml0aWFsaXplZApkaWZmIC0tZ2l0IGEvU291cmNlL1dlYktpdC9OZXR3b3JrUHJvY2Vz cy9nbGliL1dlYktpdENhY2hlZFJlc29sdmVyLmNwcCBiL1NvdXJjZS9XZWJLaXQvTmV0d29ya1By b2Nlc3MvZ2xpYi9XZWJLaXRDYWNoZWRSZXNvbHZlci5jcHAKaW5kZXggYzU0NGI2NTkwNjI1MmFm YmVmYTcwYmUwZDdmZDZiY2I1NWEwOTJlOC4uM2RlYWI5ZTQ2ZWQxZmZkODVkMWFmMjAyMmI1MWFk YjFmYmIyMWRmZSAxMDA2NDQKLS0tIGEvU291cmNlL1dlYktpdC9OZXR3b3JrUHJvY2Vzcy9nbGli L1dlYktpdENhY2hlZFJlc29sdmVyLmNwcAorKysgYi9Tb3VyY2UvV2ViS2l0L05ldHdvcmtQcm9j ZXNzL2dsaWIvV2ViS2l0Q2FjaGVkUmVzb2x2ZXIuY3BwCkBAIC0xMDgsNyArMTA4LDcgQEAgc3Rh dGljIHZvaWQgd2Via2l0Q2FjaGVkUmVzb2x2ZXJMb29rdXBCeU5hbWVBc3luYyhHUmVzb2x2ZXIq IHJlc29sdmVyLCBjb25zdCBjaGEKICAgICAgICAgICAgIHByaXYtPmNhY2hlLnVwZGF0ZShhc3lu Y0RhdGEtPmhvc3RuYW1lLCBhZGRyZXNzTGlzdEdMaXN0VG9WZWN0b3IoYWRkcmVzc0xpc3QpKTsK ICAgICAgICAgICAgIGdfdGFza19yZXR1cm5fcG9pbnRlcih0YXNrLmdldCgpLCBhZGRyZXNzTGlz dCwgcmVpbnRlcnByZXRfY2FzdDxHRGVzdHJveU5vdGlmeT4oZ19yZXNvbHZlcl9mcmVlX2FkZHJl c3NlcykpOwogICAgICAgICB9IGVsc2UKLSAgICAgICAgICAgIGdfdGFza19yZXR1cm5fZXJyb3Io dGFzay5nZXQoKSwgZXJyb3IuZ2V0KCkpOworICAgICAgICAgICAgZ190YXNrX3JldHVybl9lcnJv cih0YXNrLmdldCgpLCBlcnJvci5yZWxlYXNlKCkucmVsZWFzZSgpKTsKICAgICB9LCB0YXNrLmxl YWtSZWYoKSk7CiB9CiAKQEAgLTE3MCw3ICsxNzAsNyBAQCBzdGF0aWMgdm9pZCB3ZWJraXRDYWNo ZWRSZXNvbHZlckxvb2t1cEJ5TmFtZVdpdGhGbGFnc0FzeW5jKEdSZXNvbHZlciogcmVzb2x2ZXIs CiAgICAgICAgICAgICBwcml2LT5jYWNoZS51cGRhdGUoYXN5bmNEYXRhLT5ob3N0bmFtZSwgYWRk cmVzc0xpc3RHTGlzdFRvVmVjdG9yKGFkZHJlc3NMaXN0KSwgYXN5bmNEYXRhLT5kbnNDYWNoZVR5 cGUpOwogICAgICAgICAgICAgZ190YXNrX3JldHVybl9wb2ludGVyKHRhc2suZ2V0KCksIGFkZHJl c3NMaXN0LCByZWludGVycHJldF9jYXN0PEdEZXN0cm95Tm90aWZ5PihnX3Jlc29sdmVyX2ZyZWVf YWRkcmVzc2VzKSk7CiAgICAgICAgIH0gZWxzZQotICAgICAgICAgICAgZ190YXNrX3JldHVybl9l cnJvcih0YXNrLmdldCgpLCBlcnJvci5nZXQoKSk7CisgICAgICAgICAgICBnX3Rhc2tfcmV0dXJu X2Vycm9yKHRhc2suZ2V0KCksIGVycm9yLnJlbGVhc2UoKS5yZWxlYXNlKCkpOwogICAgIH0sIHRh c2subGVha1JlZigpKTsKIH0KIAo=