176490 2017-09-06 18:31:51 -0700 [GTK] Crash in WebCore::PlatformDisplayWayland::~PlatformDisplayWayland 2018-04-08 18:36:04 -0700 1 1 1 Unclassified WebKit WebKitGTK Other PC Linux RESOLVED FIXED https://bugs.webkit.org/show_bug.cgi?id=157973 https://bugzilla.gnome.org/show_bug.cgi?id=794228 https://bugs.webkit.org/show_bug.cgi?id=184405 P2 Normal --- 1 mcatanzaro mcatanzaro bugs-noreply cgarcia clopez commit-queue magomez mcatanzaro wk zan oldest_to_newest 1346192 0 mcatanzaro 2017-09-06 18:31:51 -0700 This is an Epiphany search provider crash: #0 0x00007f22ec477af7 in wl_list_insert_list (list=0x2524078, other=0x2524068) at /home/mcatanzaro/Projects/GNOME/wayland/src/wayland-util.c:89 No locals. #1 0x00007f22e6e81389 in wl_priv_signal_emit (signal=0x2524068, data=0x2524000) at /home/mcatanzaro/Projects/GNOME/wayland/src/wayland-server.c:1955 l = 0x2524000 pos = 0x7f230640dfb0 <initial+944> #2 0x00007f22e6e80215 in wl_display_destroy (display=0x2524000) at /home/mcatanzaro/Projects/GNOME/wayland/src/wayland-server.c:1077 s = 0x7f2307aa8000 next = 0x0 global = 0x7f22e0ff7000 gnext = 0x252c620 #3 0x00007f22fe89cf3c in WebCore::PlatformDisplayWayland::~PlatformDisplayWayland ( this=0x7f22e0ff7000, __in_chrg=<optimized out>) at ../../Source/WebCore/platform/graphics/wayland/PlatformDisplayWayland.cpp:71 No locals. #4 0x00007f22fe89cf84 in WebCore::PlatformDisplayWayland::~PlatformDisplayWayland ( this=0x7f22e0ff7000, __in_chrg=<optimized out>) at ../../Source/WebCore/platform/graphics/wayland/PlatformDisplayWayland.cpp:72 No locals. #5 0x00007f22feec6bf8 in std::default_delete<WebCore::PlatformDisplay>::operator() ( this=0x7f2304293908 <WebCore::PlatformDisplay::sharedDisplay()::display>, __ptr=0x7f22e0ff7000) at /usr/include/c++/7/bits/unique_ptr.h:78 No locals. #6 0x00007f22feec6869 in std::unique_ptr<WebCore::PlatformDisplay, std::default_delete<WebCore::PlatformDisplay> >::~unique_ptr (this=0x7f2304293908 <WebCore::PlatformDisplay::sharedDisplay()::display>, __in_chrg=<optimized out>) at /usr/include/c++/7/bits/unique_ptr.h:268 __ptr = @0x7f2304293908: 0x7f22e0ff7000 #7 0x00007f230607bc68 in __run_exit_handlers () from /lib64/libc.so.6 No symbol table info available. #8 0x00007f230607bcba in exit () from /lib64/libc.so.6 No symbol table info available. #9 0x00007f2306061511 in __libc_start_main () from /lib64/libc.so.6 No symbol table info available. #10 0x00000000004041ca in _start () No symbol table info available. I'll continue to vote in favor of getting rid of these buggy exit handlers altogether. I doubt we really need to be calling wl_display_destroy(). 1347746 1 mcatanzaro 2017-09-11 17:16:30 -0700 Unless graphics folks have any ideas, we need to remove the PlatformDisplay exit handlers. 1364655 2 mcatanzaro 2017-10-25 20:31:42 -0700 (In reply to Michael Catanzaro from comment #1) > Unless graphics folks have any ideas, we need to remove the PlatformDisplay > exit handlers. The exit handlers are not needed for program correctness as far as I'm aware, and they've been causing many, many crashes over the years. So I think it's time for them to go. At least this particular one. 1364772 3 324987 mcatanzaro 2017-10-26 04:32:54 -0700 Created attachment 324987 Patch 1364782 4 cgarcia 2017-10-26 04:57:11 -0700 I think we need to understand why it's crashing and only if it's not possible to fix it we can remove the exit handlers. I don't remember why, but I think we shut down the egl displays for a reason. IIRC mesa has its own exit handlers to clean up stuff too. 1365132 5 mcatanzaro 2017-10-26 17:14:07 -0700 (In reply to Carlos Garcia Campos from comment #4) > I think we need to understand why it's crashing and only if it's not > possible to fix it we can remove the exit handlers. I don't remember why, > but I think we shut down the egl displays for a reason. If nobody knows why, we should stop doing it and see if anyone notices a problem. 1365994 6 wk 2017-10-30 05:52:32 -0700 I am seeing this crash with liferea when running liferea headless to get GIR symbols, see https://github.com/lwindolf/liferea/issues/552 for details. 1401773 7 324987 mcatanzaro 2018-02-23 11:10:37 -0800 Comment on attachment 324987 Patch Carlos, I don't understand why it's crashing, and the graphics people don't seem to have been able to debug it. Strongly suggest pushing this in now. If the exit handlers are needed to work around OS or driver bugs, then they should be reintroduced once they can be done without crashing. 1403597 8 mcatanzaro 2018-03-05 05:15:59 -0800 Crash stats: https://retrace.fedoraproject.org/faf/problems/bthash/?bth=7712e893327fdae3c939ee81aa54d082ebbdc717 https://retrace.fedoraproject.org/faf/problems/bthash/?bth=8adf67e9e11bb1765726e37b1f3d7028ca66a8db 1403600 9 clopez 2018-03-05 05:29:00 -0800 (In reply to Carlos Garcia Campos from comment #4) > I think we need to understand why it's crashing and only if it's not > possible to fix it we can remove the exit handlers. I don't remember why, > but I think we shut down the egl displays for a reason. IIRC mesa has its > own exit handlers to clean up stuff too. Does http://trac.webkit.org/r201595 help remembering? 1403601 10 mcatanzaro 2018-03-05 05:39:39 -0800 (In reply to Carlos Alberto Lopez Perez from comment #9) > Does http://trac.webkit.org/r201595 help remembering? Note: it was an exit-time destructor before that patch, too (since the PlatformDisplay was not NeverDestroyed). The goal there was to move it to run earlier at exit time, to avoid a specific crash. But as long as we do work at exit time, there will always be more crashes.... 1405484 11 mcatanzaro 2018-03-10 09:51:53 -0800 (In reply to Michael Catanzaro from comment #8) > Crash stats: > > https://retrace.fedoraproject.org/faf/problems/bthash/ > ?bth=7712e893327fdae3c939ee81aa54d082ebbdc717 > > https://retrace.fedoraproject.org/faf/problems/bthash/ > ?bth=8adf67e9e11bb1765726e37b1f3d7028ca66a8db I really underestimated the total number of crashes here. Those links are currently showing 704 total crashes. I missed this one: https://retrace.fedoraproject.org/faf/reports/1793321/ which shows 2024 reports. That's 2728 total reports of this crash from Fedora alone. 1405486 12 mcatanzaro 2018-03-10 10:16:24 -0800 (In reply to Michael Catanzaro from comment #11) > which shows 2024 reports. That's 2728 total reports of this crash from > Fedora alone. I should add: only 688 unique reports (reports from users who haven't reported the crash before). So about four reports per unique user. 1405513 13 mcatanzaro 2018-03-10 17:24:35 -0800 ephy-search-provider crashes 100% for me... I didn't notice for so long because the search provider was broken for most of the 3.26 cycle, and the update fixing it just hit Fedora 27. Some debugging: Since this is the search provider, we wind up taking this path in PlatformDisplay::createPlatformDisplay: #if PLATFORM(WAYLAND) if (auto platformDisplay = PlatformDisplayWayland::create()) return platformDisplay; #endif Notably, that's the only path where m_nativeDisplayOwned is true, so it's the only path where the native display is destroyed. That's why the crash only occurs for the search provider. So we destroy it here: PlatformDisplayWayland::~PlatformDisplayWayland() { if (m_nativeDisplayOwned == NativeDisplayOwned::Yes) wl_display_destroy(m_display); } That's definitely wrong, because this wl_display was created with wl_display_connect(), which is Wayland client API, but wl_display_destroy() is Wayland server API. So we must never call wl_display_destroy(). The right function to use is wl_display_disconnect(), which parallels the call to wl_display_connect() that we used to create the wl_display in PlatformDisplayWayland::create. It's confusing, though, because the UI process is both a Wayland client of the desktop compositor, and also and a Wayland server for the web process. Now, with that fixed locally, there's no longer any crash inside wl_display_destroy(), and wl_display_disconnect() works fine. But I'm seeing a similar crash immediately after that, when the wl_compositor is destroyed. I suspect this indicates the root of the problem: we're trying to use the same wl_display as both a Wayland client and Wayland server, and I guess that isn't right. It doesn't happen when there's a GtkWindow because then GDK's wl_display is used as the client display instead. Thread 1 "epiphany-search" received signal SIGSEGV, Segmentation fault. 0x00007fffed7fc919 in wl_map_insert_at (map=<optimized out>, flags=<optimized out>, i=<optimized out>, data=<optimized out>) at src/wayland-util.c:247 247 src/wayland-util.c: No such file or directory. (gdb) bt #0 0x00007fffed7fc919 in wl_map_insert_at (map=<optimized out>, flags=<optimized out>, i=<optimized out>, data=<optimized out>) at src/wayland-util.c:247 #1 0x00007fffed7f8ad3 in proxy_destroy (proxy=0x5555557a2d60) at src/wayland-client.c:506 #2 0x00007fffed7f8ad3 in wl_proxy_destroy (proxy=0x5555557a2d60) at src/wayland-client.c:537 #3 0x00007ffff3c0c536 in wl_compositor_destroy (wl_compositor=<optimized out>) at /usr/include/wayland-client-protocol.h:1194 #4 0x00007ffff3c0c536 in WebCore::WlPtrDeleter<wl_compositor>::operator()(wl_compositor*) const (this=<optimized out>, ptr=<optimized out>) at /buildstream/build/Source/WebCore/platform/graphics/wayland/WlUniquePtr.h:55 #5 0x00007ffff3c0c536 in std::unique_ptr<wl_compositor, WebCore::WlPtrDeleter<wl_compositor> >::~unique_ptr() (this=0x7fffdf9f5040, __in_chrg=<optimized out>) at /usr/include/c++/7/bits/unique_ptr.h:268 #6 0x00007ffff3c0c536 in WebCore::PlatformDisplayWayland::~PlatformDisplayWayland() (this=0x7fffdf9f5000, __in_chrg=<optimized out>) at /buildstream/build/Source/WebCore/platform/graphics/wayland/PlatformDisplayWayland.cpp:68 #7 0x00007ffff3c0c569 in WebCore::PlatformDisplayWayland::~PlatformDisplayWayland() (this=0x7fffdf9f5000, __in_chrg=<optimized out>) at /buildstream/build/Source/WebCore/platform/graphics/wayland/PlatformDisplayWayland.cpp:72 #8 0x00007ffff6056248 in __run_exit_handlers (status=1, listp=0x7ffff63cf6f8 <__exit_funcs>, run_list_atexit=run_list_atexit@entry=true, run_dtors=run_dtors@entry=true) at exit.c:83 #9 0x00007ffff605629a in __GI_exit (status=<optimized out>) at exit.c:105 #10 0x00007ffff603ff31 in __libc_start_main (main= 0x55555555d974 <main>, argc=1, argv=0x7fffffffea58, init=<optimized out>, fini=<optimized out>, rtld_fini=<optimized out>, stack_end=0x7fffffffea48) at ../csu/libc-start.c:344 #11 0x00005555555599fa in _start () 1405515 14 mcatanzaro 2018-03-10 17:37:19 -0800 (In reply to Michael Catanzaro from comment #13) > we're trying to use the > same wl_display as both a Wayland client and Wayland server, and I guess > that isn't right. That's not true, I got confused by wl_compositor. That's client API too. Perhaps it should be destroyed before its wl_display. 1405525 15 335523 mcatanzaro 2018-03-10 18:03:31 -0800 Created attachment 335523 Patch 1405526 16 335523 mcatanzaro 2018-03-10 18:05:33 -0800 Comment on attachment 335523 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=335523&action=review > Source/WebCore/platform/graphics/wayland/PlatformDisplayWayland.cpp:73 > + wl_display_disconnect(m_display); Note: this does, in fact, destroy the wl_display. WlUniquePtr<wl_display> is a bit of a footgun, as it's not safe to use on any wl_display created with wl_display_connect(). 1405598 17 335523 mcatanzaro 2018-03-11 16:35:41 -0700 Comment on attachment 335523 Patch (cq? for cgarcia) 1405679 18 cgarcia 2018-03-12 01:49:04 -0700 (In reply to Michael Catanzaro from comment #16) > Comment on attachment 335523 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=335523&action=review > > > Source/WebCore/platform/graphics/wayland/PlatformDisplayWayland.cpp:73 > > + wl_display_disconnect(m_display); > > Note: this does, in fact, destroy the wl_display. > > WlUniquePtr<wl_display> is a bit of a footgun, as it's not safe to use on > any wl_display created with wl_display_connect(). Good catch! this is exactly what I wanted, once we understand the problem we can easily fix the actual bug. WlUniquePtr<wl_display> is only used once in Waylandcompositor, which is a singleton, so I think we can simply remove it to avoid confusions and use destroy/disconnect directly. 1405680 19 335523 cgarcia 2018-03-12 01:49:34 -0700 Comment on attachment 335523 Patch Thank you! 1405683 20 335523 commit-queue 2018-03-12 02:14:23 -0700 Comment on attachment 335523 Patch Clearing flags on attachment: 335523 Committed r229525: <https://trac.webkit.org/changeset/229525> 1405684 21 commit-queue 2018-03-12 02:14:25 -0700 All reviewed patches have been landed. Closing bug. 1407274 22 mcatanzaro 2018-03-17 16:25:09 -0700 One week later, we're now up to 3958 reports. I think this might be our biggest WebKit crasher ever.... 324987 2017-10-26 04:32:54 -0700 2018-03-10 18:03:29 -0800 Patch bug-176490-20171026063253.patch text/plain 4720 mcatanzaro U3VidmVyc2lvbiBSZXZpc2lvbjogMjIzOTk5CmRpZmYgLS1naXQgYS9Tb3VyY2UvV2ViQ29yZS9D aGFuZ2VMb2cgYi9Tb3VyY2UvV2ViQ29yZS9DaGFuZ2VMb2cKaW5kZXggMDU5Y2MxZGMwYTRlNTE2 OWVkMDMyMzBmNzI3N2VlODNmOGM3NDExMS4uOTNmNzZkNDkwNGJkYTdmZmFkYjg5ODA1MmQ0MjI3 ZjM5YTUyOTM1OCAxMDA2NDQKLS0tIGEvU291cmNlL1dlYkNvcmUvQ2hhbmdlTG9nCisrKyBiL1Nv dXJjZS9XZWJDb3JlL0NoYW5nZUxvZwpAQCAtMSwzICsxLDI5IEBACisyMDE3LTEwLTI2ICBNaWNo YWVsIENhdGFuemFybyAgPG1jYXRhbnphcm9AaWdhbGlhLmNvbT4KKworICAgICAgICBbV1BFXVtH VEtdIENyYXNoIGluIFdlYkNvcmU6OlBsYXRmb3JtRGlzcGxheVdheWxhbmQ6On5QbGF0Zm9ybURp c3BsYXlXYXlsYW5kCisgICAgICAgIGh0dHBzOi8vYnVncy53ZWJraXQub3JnL3Nob3dfYnVnLmNn aT9pZD0xNzY0OTAKKworICAgICAgICBSZXZpZXdlZCBieSBOT0JPRFkgKE9PUFMhKS4KKworICAg ICAgICBXZSBoYXZlIGEgcnVsZSBpbiBXZWJLaXQgdGhhdCBzdGF0aWMgb2JqZWN0cyBtdXN0IGJl IE5ldmVyRGVzdHJveWVkIHRvIGVuc3VyZSBkZXN0cnVjdG9ycyBkbworICAgICAgICBub3QgcnVu IGF0IGV4aXQgdGltZSwgaW4gb3JkZXIgdG8gYXZvaWQgZGlmZmljdWx0IGNyYXNoZXMuIFdlIGhh dmUgZnJlcXVlbnRseSBiZWVuIGlnbm9yaW5nCisgICAgICAgIHRoaXMgcnVsZSBmb3Igb3VyIExp bnV4IHBvcnRzLCBhbmQgdGhlIHByZWRpY3RhYmxlIGNvbnNlcXVlbmNlIGhhcyBiZWVuIG1hbnkg Y3Jhc2hlcy4KKyAgICAgICAgUGxhdGZvcm1EaXNwbGF5LCBpbiBwYXJ0aWN1bGFyLCBoYXMgYmVl biBhIGZyZXF1ZW50IHNvdXJjZSBvZiBwcm9ibGVtcyAoc2VlIGUuZy4gcjIwMTU5NSBhbmQKKyAg ICAgICAgcjIxOTg2NikuIFdlIG11c3Qgc3RvcCBkb2luZyB3b3JrIGF0IGV4aXQgdGltZS4KKwor ICAgICAgICBXaGlsZSBpdCBpcyBuaWNlIHRvIHJ1biB0aGUgUGxhdGZvcm1EaXNwbGF5J3MgZGVz dHJ1Y3RvciBpbiBvcmRlciB0byByZWxlYXNlIG9wZXJhdGluZworICAgICAgICBzeXN0ZW0gcmVz b3VyY2VzIGJlZm9yZSBhcHBsaWNhdGlvbiB0ZXJtaW5hdGlvbiwgYnV0IHRoaXMgaXMgbm90IHJl cXVpcmVkIGZvciBzdGF0aWMgb2JqZWN0cworICAgICAgICBhcyB0aGUgYXBwbGljYXRpb24gaXMg YWJvdXQgdG8gdGVybWluYXRlIGFuZCB0aGUgb3BlcmF0aW5nIHN5c3RlbSBpcyByZXNwb25zaWJs ZSBmb3IKKyAgICAgICAgcmVsZWFzaW5nIHJlc291cmNlcyB3aGVuIHRoZSBhcHBsaWNhdGlvbiB0 ZXJtaW5hdGVzLiBXZSBkbyBoYXZlIHRvIGJlIGNhcmVmdWwgdG8gZW5zdXJlIHRoYXQKKyAgICAg ICAgbm90aGluZyAiaW1wb3J0YW50IiBvY2N1cnMgaW4gdGhlIGRlc3RydWN0b3IsIGJ1dCB0aGF0 J3MgYWxyZWFkeSBub3QgdGhlIGNhc2UgZm9yCisgICAgICAgIFBsYXRmb3JtRGlzcGxheS4KKwor ICAgICAgICAqIHBsYXRmb3JtL2dyYXBoaWNzL1BsYXRmb3JtRGlzcGxheS5jcHA6CisgICAgICAg IChXZWJDb3JlOjpQbGF0Zm9ybURpc3BsYXk6OnNoYXJlZERpc3BsYXkpOgorICAgICAgICAoV2Vi Q29yZTo6UGxhdGZvcm1EaXNwbGF5Ojppbml0aWFsaXplRUdMRGlzcGxheSk6CisgICAgICAgICgp OiBEZWxldGVkLgorICAgICAgICAoV2ViQ29yZTo6UGxhdGZvcm1EaXNwbGF5OjpzaHV0RG93bkVn bERpc3BsYXlzKTogRGVsZXRlZC4KKwogMjAxNy0xMC0yNSAgUnlvc3VrZSBOaXdhICA8cm5pd2FA d2Via2l0Lm9yZz4KIAogICAgICAgICBTdHlsZTo6U2NvcGU6OmZsdXNoUGVuZGluZ1VwZGF0ZSgp IGNhbiByZXBsYWNlIHRoZSBlbnRpcmUgZG9jdW1lbnQgaW4gWFNMVFByb2Nlc3Nvcjo6Y3JlYXRl RG9jdW1lbnRGcm9tU291cmNlCmRpZmYgLS1naXQgYS9Tb3VyY2UvV2ViQ29yZS9wbGF0Zm9ybS9n cmFwaGljcy9QbGF0Zm9ybURpc3BsYXkuY3BwIGIvU291cmNlL1dlYkNvcmUvcGxhdGZvcm0vZ3Jh cGhpY3MvUGxhdGZvcm1EaXNwbGF5LmNwcAppbmRleCA1OTg5OGQ2YzljZWI3YTE1ZjBlNDc1YmYx MTg2MTI1ZjBkODMwOTQ2Li42ODRhNjViNmFmNTk1ZjU0ZTM3MDQ3ZDgyMTZkMjc0MTBjZjVhZmEx IDEwMDY0NAotLS0gYS9Tb3VyY2UvV2ViQ29yZS9wbGF0Zm9ybS9ncmFwaGljcy9QbGF0Zm9ybURp c3BsYXkuY3BwCisrKyBiL1NvdXJjZS9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL1BsYXRmb3Jt RGlzcGxheS5jcHAKQEAgLTExOSwxOCArMTE5LDExIEBAIHN0ZDo6dW5pcXVlX3B0cjxQbGF0Zm9y bURpc3BsYXk+IFBsYXRmb3JtRGlzcGxheTo6Y3JlYXRlUGxhdGZvcm1EaXNwbGF5KCkKIFBsYXRm b3JtRGlzcGxheSYgUGxhdGZvcm1EaXNwbGF5OjpzaGFyZWREaXNwbGF5KCkKIHsKICAgICBzdGF0 aWMgc3RkOjpvbmNlX2ZsYWcgb25jZUZsYWc7Ci0jaWYgQ09NUElMRVIoQ0xBTkcpCi0jcHJhZ21h IGNsYW5nIGRpYWdub3N0aWMgcHVzaAotI3ByYWdtYSBjbGFuZyBkaWFnbm9zdGljIGlnbm9yZWQg Ii1XZXhpdC10aW1lLWRlc3RydWN0b3JzIgotI2VuZGlmCi0gICAgc3RhdGljIHN0ZDo6dW5pcXVl X3B0cjxQbGF0Zm9ybURpc3BsYXk+IGRpc3BsYXk7Ci0jaWYgQ09NUElMRVIoQ0xBTkcpCi0jcHJh Z21hIGNsYW5nIGRpYWdub3N0aWMgcG9wCi0jZW5kaWYKKyAgICBzdGF0aWMgTmV2ZXJEZXN0cm95 ZWQ8c3RkOjp1bmlxdWVfcHRyPFBsYXRmb3JtRGlzcGxheT4+IGRpc3BsYXk7CiAgICAgc3RkOjpj YWxsX29uY2Uob25jZUZsYWcsIFtdewotICAgICAgICBkaXNwbGF5ID0gY3JlYXRlUGxhdGZvcm1E aXNwbGF5KCk7CisgICAgICAgIGRpc3BsYXkuZ2V0KCkgPSBjcmVhdGVQbGF0Zm9ybURpc3BsYXko KTsKICAgICB9KTsKLSAgICByZXR1cm4gKmRpc3BsYXk7CisgICAgcmV0dXJuICpkaXNwbGF5Lmdl dCgpOwogfQogCiBzdGF0aWMgUGxhdGZvcm1EaXNwbGF5KiBzX3NoYXJlZERpc3BsYXlGb3JDb21w b3NpdGluZzsKQEAgLTE3MCwxMiArMTYzLDYgQEAgR0xDb250ZXh0KiBQbGF0Zm9ybURpc3BsYXk6 OnNoYXJpbmdHTENvbnRleHQoKQogI2VuZGlmCiAKICNpZiBVU0UoRUdMKQotc3RhdGljIEhhc2hT ZXQ8UGxhdGZvcm1EaXNwbGF5Kj4mIGVnbERpc3BsYXlzKCkKLXsKLSAgICBzdGF0aWMgTmV2ZXJE ZXN0cm95ZWQ8SGFzaFNldDxQbGF0Zm9ybURpc3BsYXkqPj4gZGlzcGxheXM7Ci0gICAgcmV0dXJu IGRpc3BsYXlzOwotfQotCiBFR0xEaXNwbGF5IFBsYXRmb3JtRGlzcGxheTo6ZWdsRGlzcGxheSgp IGNvbnN0CiB7CiAgICAgaWYgKCFtX2VnbERpc3BsYXlJbml0aWFsaXplZCkKQEAgLTIxMCwyNCAr MTk3LDYgQEAgdm9pZCBQbGF0Zm9ybURpc3BsYXk6OmluaXRpYWxpemVFR0xEaXNwbGF5KCkKIAog ICAgIG1fZWdsTWFqb3JWZXJzaW9uID0gbWFqb3JWZXJzaW9uOwogICAgIG1fZWdsTWlub3JWZXJz aW9uID0gbWlub3JWZXJzaW9uOwotCi0gICAgZWdsRGlzcGxheXMoKS5hZGQodGhpcyk7Ci0KLSNp ZiAhUExBVEZPUk0oV0lOKQotICAgIHN0YXRpYyBib29sIGVnbEF0ZXhpdEhhbmRsZXJJbml0aWFs aXplZCA9IGZhbHNlOwotICAgIGlmICghZWdsQXRleGl0SGFuZGxlckluaXRpYWxpemVkKSB7Ci0g ICAgICAgIC8vIEVHTCByZWdpc3RlcnMgYXRleGl0IGhhbmRsZXJzIHRvIGNsZWFudXAgaXRzIGds b2JhbCBkaXNwbGF5IGxpc3QuCi0gICAgICAgIC8vIFNpbmNlIHRoZSBnbG9iYWwgUGxhdGZvcm1E aXNwbGF5IGluc3RhbmNlIGlzIGNyZWF0ZWQgYmVmb3JlLAotICAgICAgICAvLyB3aGVuIHRoZSBQ bGF0Zm9ybURpc3BsYXkgZGVzdHJ1Y3RvciBpcyBjYWxsZWQsIEVHTCBoYXMgYWxyZWFkeSByZW1v dmVkIHRoZQotICAgICAgICAvLyBkaXNwbGF5IGZyb20gdGhlIGxpc3QsIGNhdXNpbmcgZWdsVGVy bWluYXRlKCkgdG8gY3Jhc2guIFNvLCBoZXJlIHdlIHJlZ2lzdGVyCi0gICAgICAgIC8vIG91ciBv d24gYXRleGl0IGhhbmRsZXIsIGFmdGVyIEVHTCBoYXMgYmVlbiBpbml0aWFsaXplZCBhbmQgYWZ0 ZXIgdGhlIGdsb2JhbAotICAgICAgICAvLyBpbnN0YW5jZSBoYXMgYmVlbiBjcmVhdGVkIHRvIGVu c3VyZSB3ZSBjYWxsIGVnbFRlcm1pbmF0ZSgpIGJlZm9yZSB0aGUgb3RoZXIKLSAgICAgICAgLy8g RUdMIGF0ZXhpdCBoYW5kbGVycyBhbmQgdGhlIFBsYXRmb3JtRGlzcGxheSBkZXN0cnVjdG9yLgot ICAgICAgICAvLyBTZWUgaHR0cHM6Ly9idWdzLndlYmtpdC5vcmcvc2hvd19idWcuY2dpP2lkPTE1 Nzk3My4KLSAgICAgICAgZWdsQXRleGl0SGFuZGxlckluaXRpYWxpemVkID0gdHJ1ZTsKLSAgICAg ICAgc3RkOjphdGV4aXQoc2h1dERvd25FZ2xEaXNwbGF5cyk7Ci0gICAgfQotI2VuZGlmCiB9CiAK IHZvaWQgUGxhdGZvcm1EaXNwbGF5Ojp0ZXJtaW5hdGVFR0xEaXNwbGF5KCkKQEAgLTI0MCwxNCAr MjA5LDYgQEAgdm9pZCBQbGF0Zm9ybURpc3BsYXk6OnRlcm1pbmF0ZUVHTERpc3BsYXkoKQogICAg IG1fZWdsRGlzcGxheSA9IEVHTF9OT19ESVNQTEFZOwogfQogCi12b2lkIFBsYXRmb3JtRGlzcGxh eTo6c2h1dERvd25FZ2xEaXNwbGF5cygpCi17Ci0gICAgd2hpbGUgKCFlZ2xEaXNwbGF5cygpLmlz RW1wdHkoKSkgewotICAgICAgICBhdXRvKiBkaXNwbGF5ID0gZWdsRGlzcGxheXMoKS50YWtlQW55 KCk7Ci0gICAgICAgIGRpc3BsYXktPnRlcm1pbmF0ZUVHTERpc3BsYXkoKTsKLSAgICB9Ci19Ci0K ICNlbmRpZiAvLyBVU0UoRUdMKQogCiB9IC8vIG5hbWVzcGFjZSBXZWJDb3JlCg== 335523 2018-03-10 18:03:31 -0800 2018-03-12 02:14:23 -0700 Patch bug-176490-20180310200330.patch text/plain 1890 mcatanzaro U3VidmVyc2lvbiBSZXZpc2lvbjogMjI5NDQyCmRpZmYgLS1naXQgYS9Tb3VyY2UvV2ViQ29yZS9D aGFuZ2VMb2cgYi9Tb3VyY2UvV2ViQ29yZS9DaGFuZ2VMb2cKaW5kZXggNTI2MWNjNjg4NDVlOTU2 YTg5ZWE1OTNkNDFhYmZkYTJhNmRhOTJlMy4uOWUyNzEyMzg0MzI5MGU2NDk0ZmQyZTYwZmFjNmQ2 MzViNTVmMmJjNiAxMDA2NDQKLS0tIGEvU291cmNlL1dlYkNvcmUvQ2hhbmdlTG9nCisrKyBiL1Nv dXJjZS9XZWJDb3JlL0NoYW5nZUxvZwpAQCAtMSwzICsxLDE3IEBACisyMDE4LTAzLTEwICBNaWNo YWVsIENhdGFuemFybyAgPG1jYXRhbnphcm9AaWdhbGlhLmNvbT4KKworICAgICAgICBbR1RLXSBD cmFzaCBpbiBXZWJDb3JlOjpQbGF0Zm9ybURpc3BsYXlXYXlsYW5kOjp+UGxhdGZvcm1EaXNwbGF5 V2F5bGFuZAorICAgICAgICBodHRwczovL2J1Z3Mud2Via2l0Lm9yZy9zaG93X2J1Zy5jZ2k/aWQ9 MTc2NDkwCisKKyAgICAgICAgUmV2aWV3ZWQgYnkgTk9CT0RZIChPT1BTISkuCisKKyAgICAgICAg RGVzdHJveSB0aGUgd2xfZGlzcGxheSB3aXRoIHdsX2Rpc3BsYXlfZGlzY29ubmVjdCgpIChjbGll bnQgcHJvY2VzcyBBUEkpLCBub3QKKyAgICAgICAgd2xfZGlzcGxheV9kZXN0cm95KCkgKHNlcnZl ciBwcm9jZXNzIEFQSSkuIEl0IGhhcyB0byBiZSBkZXN0cm95ZWQgbGFzdCwgc28gZXhwbGljaXRs eQorICAgICAgICBkZXN0cm95IHRoZSB3bF9yZWdpc3RyeSBhbmQgd2xfY29tcG9zaXRvciBmaXJz dC4KKworICAgICAgICAqIHBsYXRmb3JtL2dyYXBoaWNzL3dheWxhbmQvUGxhdGZvcm1EaXNwbGF5 V2F5bGFuZC5jcHA6CisgICAgICAgIChXZWJDb3JlOjpQbGF0Zm9ybURpc3BsYXlXYXlsYW5kOjp+ UGxhdGZvcm1EaXNwbGF5V2F5bGFuZCk6CisKIDIwMTgtMDMtMDggIE1lZ2FuIEdhcmRuZXIgIDxt ZWdhbl9nYXJkbmVyQGFwcGxlLmNvbT4KIAogICAgICAgICBFbnN1cmUgZGVmYXVsdCBhcHBlYXJh bmNlIGZvciBzZWxlY3Rpb24gZm9ybSBjb250cm9sCmRpZmYgLS1naXQgYS9Tb3VyY2UvV2ViQ29y ZS9wbGF0Zm9ybS9ncmFwaGljcy93YXlsYW5kL1BsYXRmb3JtRGlzcGxheVdheWxhbmQuY3BwIGIv U291cmNlL1dlYkNvcmUvcGxhdGZvcm0vZ3JhcGhpY3Mvd2F5bGFuZC9QbGF0Zm9ybURpc3BsYXlX YXlsYW5kLmNwcAppbmRleCBkYWMwMWQ4NDZhYTM2YTMxYjZiNWQ5ZTUyZmZmYTRmNmY5ZjAxOTk5 Li4zYjliMmNlOTNjMzA5NDQ1MmRhYjY0ZTViNzBjYzMyZmRlZWE4MmEyIDEwMDY0NAotLS0gYS9T b3VyY2UvV2ViQ29yZS9wbGF0Zm9ybS9ncmFwaGljcy93YXlsYW5kL1BsYXRmb3JtRGlzcGxheVdh eWxhbmQuY3BwCisrKyBiL1NvdXJjZS9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL3dheWxhbmQv UGxhdGZvcm1EaXNwbGF5V2F5bGFuZC5jcHAKQEAgLTY3LDggKzY3LDExIEBAIFBsYXRmb3JtRGlz cGxheVdheWxhbmQ6OlBsYXRmb3JtRGlzcGxheVdheWxhbmQoc3RydWN0IHdsX2Rpc3BsYXkqIGRp c3BsYXksIE5hdGl2CiAKIFBsYXRmb3JtRGlzcGxheVdheWxhbmQ6On5QbGF0Zm9ybURpc3BsYXlX YXlsYW5kKCkKIHsKLSAgICBpZiAobV9uYXRpdmVEaXNwbGF5T3duZWQgPT0gTmF0aXZlRGlzcGxh eU93bmVkOjpZZXMpCi0gICAgICAgIHdsX2Rpc3BsYXlfZGVzdHJveShtX2Rpc3BsYXkpOworICAg IGlmIChtX25hdGl2ZURpc3BsYXlPd25lZCA9PSBOYXRpdmVEaXNwbGF5T3duZWQ6Olllcykgewor ICAgICAgICBtX2NvbXBvc2l0b3IgPSBudWxscHRyOworICAgICAgICBtX3JlZ2lzdHJ5ID0gbnVs bHB0cjsKKyAgICAgICAgd2xfZGlzcGxheV9kaXNjb25uZWN0KG1fZGlzcGxheSk7CisgICAgfQog fQogCiB2b2lkIFBsYXRmb3JtRGlzcGxheVdheWxhbmQ6OmluaXRpYWxpemUod2xfZGlzcGxheSog ZGlzcGxheSkK