Commit Graph

132 Commits

Author SHA1 Message Date
Jim Mathies
0ccee0c61c Bug 539020 - OOPP: mouse events get duped. r=bsmedberg. 2010-01-11 12:53:08 -06:00
Chris Jones
61403b3c24 Fix windows bustage caused by 04d6f128da25 2010-01-04 16:41:34 -06:00
Chris Jones
ad47f303ae Bug 535687: Fix regression in forwarding NPN_InvalidateRect() for OOPP on linux. r=bent 2010-01-04 15:46:34 -06:00
Walter Meinl
c20eb4938d Bug 536264 - "[OS/2] build break in modules\plugins" [r=josh] 2009-12-23 15:12:00 -06:00
Chris Jones
526c1eb4fb bug 535073: prevent plugins from corrupting the stack by making word-size stores to pointers to NPBool (unsigned char). r=bsmedberg 2009-12-17 00:32:44 -06:00
Benjamin Smedberg
19e77143fb Bug 518924 - Implement NPN_AsyncCallback r=cjones 2009-12-16 17:27:18 -05:00
Benjamin Smedberg
1f863b7fb6 Bug 532751 - Notify the nsNPAPIPlugin and the related nsNPAPIPluginInstances when a plugin crashes so that reloading will create a new plugin process, r=josh 2009-12-16 15:08:45 -05:00
Chris Jones
63b57abda9 bug 534863: the plugin DLL path needs to be quoted on windows. irc r=bent 2009-12-15 17:10:41 -06:00
Benjamin Smedberg
f2b62bb114 Merge mozilla-central into electrolysis. 2009-12-14 21:27:17 -05:00
Jim Mathies
3c68e53543 Bug 534027 - input event coords incorrect for oop windowless plugins. r=jmuizelaar. 2009-12-14 18:28:51 -06:00
Jim Mathies
8334063adb Bug 531860 - plugin-alpha-zindex ref test fails with oopp enabled. r=jmuizelaar. 2009-12-14 18:27:25 -06:00
Benjamin Smedberg
4d837b5abc Followup to e10s landing (and back to bug 516509): on Mac, NP_GetMIMEDescription and NP_GetValue are optional entry points, because the metadata can be retrieved from the .plist file, r=josh
--HG--
extra : rebase_source : 0fc5db971736612606c66c20cc2f0a3cc62ca444
2009-12-14 16:19:25 -05:00
Ben Turner
f9ead8a5d6 Backing out changesets ed8d3323a50d, 3930bbdffdd5, 36e32abd2cc8, f4bcfc703ff9, 21051d997076, and 49f489315dd9 at the request of bsmedberg. 2009-12-11 12:48:37 -08:00
Benjamin Smedberg
6b8e2464c1 Defer the PluginCrashed notification so that we don't delete the PluginModuleParent while we're in the middle of a crash notification. 2009-12-04 21:20:06 -05:00
Benjamin Smedberg
badf2095ff Partly fix test bustage from bug 532751: set mShutdown before notifying the nsNPAPIPlugin, so that it doesn't re-enter IPC methods. 2009-12-04 18:41:56 -05:00
Benjamin Smedberg
061a369048 Bustage fix 2009-12-03 15:44:33 -05:00
Benjamin Smedberg
87941a941e Bug 532751 - Stop cacheing the nsNPAPIPlugin when it crashes, so that refreshing the page immediately reloads the plugin, r=bent pending r?joshmoz 2009-12-04 13:24:57 -05:00
Benjamin Smedberg
42a284d56b Bug 532671 - After a plugin crashes, the module entry points (most importantly NPP_New) should fail, r=bent
--HG--
extra : rebase_source : 389d54e70fea2801d28bdf2e48740dcaaec31a07
2009-12-03 15:31:52 -05:00
Benjamin Smedberg
4fdb3bfb23 When the plugin crashes, set NPP->pdata to NULL, and null-check NPP->pdata in the callbacks that matter. r=bent
--HG--
extra : rebase_source : 32489b985b05b9da68546d37a2f0302bac8fd467
2009-12-03 15:31:01 -05:00
Benjamin Smedberg
b695aba156 Bug 532659 - PluginScriptableObjectParent gets double-deleted when the plugin crashes, and fix the IPDL bug where we delete the actor if construction fails, instead of calling Deallocate on it, r=bent
--HG--
extra : rebase_source : dbe9a100735b1e0d0a8b07f4421e1ccb6055a78a
2009-12-03 15:30:05 -05:00
Ben Turner
f5b425c04f Fix bad invalidation bug in child NPObjects 2009-12-03 11:53:09 -08:00
Chris Jones
c47450d9a7 bug 529005: detect child process shutdowns vs. crashes, and expose this information to IPDL actors in a new |ActorDestroy(why)| interface. also ensure that subprotocol actors are notified of shutdown and cleaned up properly. r=bsmedberg r=bent 2009-12-03 02:16:28 -06:00
Chris Jones
270962252a bug 525342: make IPDL dtors part of the destructed actor's protocol instead of the manager's. r=bsmedberg 2009-12-03 02:16:14 -06:00
Chris Jones
16291938f9 bug 524220: add a |nullable| type qualifier denoting whether C++ actor params are allowed to be NULL. r=bent 2009-12-03 00:35:22 -06:00
Chris Jones
c059302c3a Build fix 2009-12-01 17:33:51 -06:00
Ben Turner
d98c67138d Bug 526401 - Fixup handling of null NPP. r=bsmedberg. 2009-12-01 13:27:13 -08:00
Jim Mathies
705471067a Bug 522299 - Electrolysis: Get windowless plugins drawing on win32. r=jmuizelaar. 2009-12-01 15:05:26 -06:00
Benjamin Smedberg
822c6cd02f XXX removal: don't destroy the stream on failure (a future ActorDestroy callback will take care of that), and other shutdown- and crash-related XXX removal which are handled by bug 529005 2009-11-25 15:34:49 -05:00
Benjamin Smedberg
9b8b8bbfe2 TODO-removal: make NullableStringGet an inline function instead of a macro, and make NullableString an inline function in mozilla::plugins instead of in an anonymous namespace: no need to have a separate copy of it for each translation unit. 2009-11-25 15:07:45 -05:00
Benjamin Smedberg
53bf5266c6 More TODOs in the the plugin stream code, use the NullableString mechanisms which were invented after this landed. 2009-11-25 10:34:59 -05:00
Benjamin Smedberg
3e0ae20e17 Bug 530894 - Clean up TODOs in the plugin stream remoting: if NPN_{Get,Post}URLNotify fail, clean up the actor properly. Also deal with null postdata, which currently causes significant problems for the in-process case (NPN_PostURLNotify succeeds, but NPP_URLNotify is never called). r=bent 2009-11-25 09:26:01 -05:00
Benjamin Smedberg
6acc2d4d3b Backed out changeset 97c456993dbd -- red herring, no need for this debugging spew. 2009-11-20 13:48:42 -05:00
Chris Jones
b70e1dc5c2 temporary debugging spew to check bug 529234 2009-11-20 11:54:46 -06:00
Benjamin Smedberg
188d096d5c Don't crash if NP_Initialize failed remotely: PluginLibrary::NP_Initialize returns an nsresult, not a NPError; nsNPAPIPlugin::CreatePlugin fails but still sets the outparam; nsPluginHost doesn't properly rv-check nsNPAPIPlugin::CreatePlugin, which causes us to think the bad plugin is actually good. I hate stupid failure chains! 2009-11-20 10:34:20 -05:00
Chris Jones
d6c803e859 bug 530007: fix crashing edge cases in the hope of getting better diagnostics about failed tests 2009-11-20 02:23:07 -06:00
Benjamin Smedberg
7101a519cc Don't use a static nsCString (causing leaks in the plugin process). Also, use NS_LogInit/NS_LogTerm for the child process so that leaks are reported at the right time. 2009-11-19 17:57:56 -05:00
Ben Turner
9e4dea1fd1 Fix bad allocation pattern in NPP_New 2009-11-19 14:47:09 -08:00
Ben Turner
5d92ec3ff2 Followup to bug 527876 2009-11-13 13:19:33 -08:00
Ben Turner
8e9c6fea1d Bug 527876 - 'test_windowed_invalidate.html hangs during mochitest-ipcplugins'. 2009-11-13 12:20:03 -08:00
Chris Jones
825bd599dd no need to include in C++ protocol-common headers |include protocol|'d from IPDL. this is an accident of history 2009-11-12 16:20:20 -06:00
Ben Turner
82191b418a Bug 525792 - 'Fix leaks running mochitests'. 2009-11-12 14:16:54 -08:00
Benjamin Smedberg
661a77e3d4 Don't rely on the plugin methods to return a non-null actor when the say NPERR_NO_ERROR... crashing here is a no-no even if the plugin is stupid. 2009-11-11 17:01:07 -05:00
Ben Turner
0a324cd616 Fix invalidation algorithm so that we never leave a deleted actor in the live hash map. 2009-11-11 17:12:01 -08:00
Chris Jones
5c7a5ab290 mac builds and runs with --enable-ipc, but the subprocess nsAppShell dies for an apparently Cocoa-related reason. IPC is fine though 2009-11-10 20:25:10 -06:00
Ben Turner
d515b14aeb Fix silly bugs 2009-11-09 17:30:02 -08:00
Phineas T. Farnsworth
3da7352b78 bug 526626: band-aids for shutdown assertions 2009-11-09 16:56:55 -06:00
Ben Turner
d10ccd2c5a Bug 522414 - 'RPC and Sync channels will deadlock easily on Windows'. r=jimm+rs. 2009-11-06 14:33:12 -08:00
Chris Jones
2eb8767355 bug 523894: wstrings are bad news in OPT builds due to chromium/Mozilla -fshort-wchar mismatch 2009-11-05 18:24:27 -06:00
Chris Jones
c916fbb575 bug 525792: fix leaking PluginModules, call NP_Shutdown() on plugins. minor, no r= 2009-11-03 15:37:07 -06:00
Benjamin Smedberg
4ae77766a7 Remove stale TODO, this will be properly covered by parent-side error handling events. 2009-11-02 11:48:52 -05:00