summaryrefslogtreecommitdiff
path: root/chromium
Commit message (Collapse)AuthorAgeFilesLines
* Allow the embedder to suppress an error page for http errors83-basedKirill Burtsev2021-01-051-0/+5
| | | | | Change-Id: Ifc0bd4a88a477de6bb786f5fec6f2c2e3278d44a Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* Fix embedded builds with printing enabledAllan Sandfeld Jensen2021-01-041-1/+1
| | | | | | | | Correct check from is_desktop_linux to is_linux, if the code isn't needed it isn't included in if the first place. Change-Id: I97523bbf476e95057abfee99914bb9118be7fd2a Reviewed-by: Michael Brüning <michael.bruning@qt.io>
* Add WebEnginePageRenderFrame mojo interfaceMichal Klocek2020-12-102-0/+9
| | | | | Change-Id: I227e5119d943938c07b386a85a24494a55acb5b9 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* Add user script data mojo interface and traitsMichal Klocek2020-12-105-0/+295
| | | | | Change-Id: I13e2d365fd7c13e7aa18c2a5882fb71267fdcd60 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Revert] Don't use harfbuzz-subsetPeter Varga2020-12-085-21/+4
| | | | | | | | | | | | | | Also reverts "Disable font subsetting to support system harfbuzz < 2.4.0" sfntly has been removed from Chromium and now it uses HarfBuzz as the font subsetter. See https://crbug.com/931719 Change-Id: I55255a82fdb7c10ca69678e2cd04672b520fc6f8 Fixes: QTBUG-88976 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* Add qt static dependencies: qtzlibMichal Klocek2020-12-021-0/+20
| | | | | | | | | QtZlib is a part of QtCore, and we need also shim headers since v8 will refer to zlib in 3rdaprty. Task-number: QTBUG-87154 Change-Id: I85c063b77eacd8dafae59cc1ef2b391e3f7c1ce0 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* Add qt static dependencies: qtlibjpegMichal Klocek2020-12-022-0/+16
| | | | | | | | | Most subprojects need jpeg shim headers, however pdfium does not require those. Task-number: QTBUG-87154 Change-Id: I9c41aa357f96ae28cc22031ddd3acf93a7f1dee6 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* Add qt static dependencies: qtlibpngMichal Klocek2020-12-022-0/+12
| | | | | | | | | | | | Pdfium extensions can be compiled with system libpng or internal pdfium bundled libpng, which is different than chromium's third_party libpng (where shim headers are used). To keep changes to minimum update only pdfium third party build. Task-number: QTBUG-87154 Change-Id: I87668d2a8e45089193bc369e3c202440e398d68d Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* Add qt static dependencies: qtfreetype, qtharfbuzzMichal Klocek2020-12-027-17/+50
| | | | | | | | | | | | | | | | | | | For static build of pdfium we need to take deps form qtbase, pdfium uses freetype, which needs harfbuzz, Update build for freetype and harfbuzz in third party, so code can be compiled with that headers. It seems if qtfreetype is used we no longer need harfbuzz dependency, but harfbuzz build changes are introduced for completeness. Note we are only interested in include path, since qmake is doing final link, otherwise lib_dirs can be passed in same manner. Task-number: QTBUG-87154 Change-Id: I46ade3cfe134d5102a941270fbb90a1fa455e46a Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* Fix build with msvc2019 16.8.0Peter Varga2020-11-233-4/+18
| | | | | | Fixes: QTBUG-88708 Change-Id: I3554ceec0437801b4861f68edd504d01fc01cf93 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* Fix build with system ICU 68Allan Sandfeld Jensen2020-11-1812-27/+27
| | | | | | Fixes: QTBUG-88116 Change-Id: I935babf51c2670fad7cc7950a2fe07eb2829c4cb Reviewed-by: Jüri Valdmann <juri.valdmann@qt.io>
* [Backport] mac: make find_sdk.py work when the sdk goes to 11Peter Varga2020-11-161-5/+6
| | | | | | | | | | | Bug: 1098738 Change-Id: I25b84537a445ecb8f80241c98d4753932f5f7c90 Commit-Queue: Nico Weber <thakis@chromium.org> Commit-Queue: Mark Mentovai <mark@chromium.org> Auto-Submit: Nico Weber <thakis@chromium.org> Reviewed-by: Mark Mentovai <mark@chromium.org> Cr-Commit-Position: refs/heads/master@{#781835} Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] Security bug 1137608v5.15.2Thibaud Michaud2020-11-031-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/v8/v8/+/2474784: [codegen] Skip invalid optimization in tail calls Preparing for tail call is usually done by emitting the gap moves and then moving the stack pointer to its new position. An optimization consists in moving the stack pointer first and transforming some of the moves into pushes. In the attached case it looks like this (arm): 138 add sp, sp, #40 13c str r6, [sp, #-4]! 140 str r6, [sp, #-4]! 144 str r6, [sp, #-4]! 148 str r6, [sp, #-4]! 14c str r6, [sp, #-4]! ... 160 vldr d1, [sp - 4*3] The last line is a gap reload, but because the stack pointer was already moved, the slot is now below the stack pointer. This is invalid and triggers this DCHECK: Fatal error in ../../v8/src/codegen/arm/assembler-arm.cc, line 402 Debug check failed: 0 <= offset (0 vs. -12). A comment already explains that we skip the optimization if the gap contains stack moves to prevent this, but the code only checks for non-FP slots. This is fixed by replacing "source.IsStackSlot()" with "source.IsAnyStackSlot()": 108 vldr d1, [sp + 4*2] ... 118 str r0, [sp, #+36] 11c str r0, [sp, #+32] 120 str r0, [sp, #+28] 124 str r0, [sp, #+24] 128 str r0, [sp, #+20] ... 134 add sp, sp, #20 R=jgruber@chromium.org Bug: chromium:1137608 Change-Id: If2b85dde49bf31a6bd3f5e0255407f9390727f9d Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020-16011: Heap buffer overflow in UI on Windows.Adrian Taylor2020-11-031-1/+1
| | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2513324: Prevent overflow of drag image on Windows. Bug: 1144489 Change-Id: I130adffc1c69073295537aaff3ce7054260064fc Reviewed-by: Darwin Huang <huangdarwin@chromium.org> Commit-Queue: Adrian Taylor <adetaylor@chromium.org> Cr-Commit-Position: refs/heads/master@{#823018} Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020-16009: Inappropriate implementation in V8Leszek Swirski2020-11-032-1/+12
| | | | | | | | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/v8/v8/+/2507715: [map] Try to in-place transition during map update When searching for a target map during map update, attempt to update field representations in-place to the more general representation, where possible. Bug: chromium:1143772 Change-Id: I6a43c94910a1d2d8f8b0ad89048f94b51461f76c Commit-Queue: Leszek Swirski <leszeks@chromium.org> Commit-Queue: Igor Sheludko <ishell@chromium.org> Reviewed-by: Igor Sheludko <ishell@chromium.org> Auto-Submit: Leszek Swirski <leszeks@chromium.org> Cr-Commit-Position: refs/heads/master@{#70887} Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020-16008: Stack buffer overflow in WebRTCBjorn Terelius2020-11-032-10/+8
| | | | | | | | | | | | | Cherry-pick of patch originally reviewed on https://webrtc-review.googlesource.com/c/src/+/186720: Allow RTCP packets longer than 1500 bytes in RTC event log. Bug: chromium:1134107 Change-Id: I05da32c57537c3c2fddae96918ff4e4685d62043 Reviewed-by: Elad Alon <eladalon@webrtc.org> Commit-Queue: Björn Terelius <terelius@webrtc.org> Cr-Commit-Position: refs/heads/master@{#32315} Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020-16005: Insufficient policy enforcement in ANGLEJamie Madill2020-11-032-0/+2
| | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/angle/angle/+/2485581: Fix missing validation cache update on VAO binding. Bug: chromium:1139398 Change-Id: I85a0d7a72bc2c97b07ebc5f86effd8e36aefd544 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* Fix assert when running tests build with -no-webengine-webrtcAllan Sandfeld Jensen2020-10-301-1/+1
| | | | | Change-Id: I81fdf3689074b7b1e5d6a548f7b9d7f41ebbcbf7 Reviewed-by: Jüri Valdmann <juri.valdmann@qt.io>
* [Backport] CVE-2020-15985: Inappropriate implementation in Blink.Charlie Harrison2020-10-271-5/+18
| | | | | | | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2278597: Large cursor fallback: ensure correct coordinate space for OOPIFs A previous change ensured that large custom cursors > 32x32 would be dropped if they are not fully contained with the visual viewport. However, the computation did not account properly for OOPIFs, where cursor coordinates were not adjusted to the viewport offset. This CL further adjusts the cursor rect by translating it to the root view's coordinate space via LocalToAncestorPoint, before checking for containment within the visual viewport. Bug: 1099276 Change-Id: I0a03e7cc249cd785f9e76f931cfc7931b127d56b Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020-6557: Inappropriate implementation in networkingEric Roman2020-10-279-0/+210
| | | | | | | | | | | | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2350395: Add a per-process limit on open UDP sockets. This adds a default limit of 6000 on the open UDP sockets throughout the entire process, configurable with the "LimitOpenUDPSockets" feature. An "open UDP socket" specifically means a net::UDPSocket which successfully called Open(), and has not yet called Close(). Once the limit has been reached, opening UDP socket will fail with ERR_INSUFFICIENT_RESOURCES. In Chrome Browser, UDP sockets are brokered through a single process (that hosting the Network Service), so this is functionally a browser-wide limit too. Bug: 1083278 Change-Id: Ib95ab14b7ccf5e15410b9df9537c66c858de2d7d Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] Security bug 1106091Caleb Raitto2020-10-274-80/+280
| | | | | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2309004: Remove SimpleSerialize() and SimpleDeserialize(). By handling serialization of each type explicitly, we avoid serializing padding bytes. Bug: 973801,1106091 Change-Id: I9f4e6b82fb484b9256b25b531cc2d51c800425bb Commit-Queue: Khushal <khushalsagar@chromium.org> Reviewed-by: Khushal <khushalsagar@chromium.org> Cr-Commit-Position: refs/heads/master@{#791110} Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020-15987: Use after free in WebRTC (2/2)Taylor Brandstetter2020-10-271-2/+12
| | | | | | | | | | | | | | | | | | Partial backport of patch originally reviewed on https://webrtc-review.googlesource.com/c/src/+/184240: Fix for OnSctpInboundPacket being called after transport destruction. OnSctpInboundPacket is called not only for incoming packets, but for notifications, which can be delivered on the usrsctp timer thread. I suspect that these notifications can be delivered after we attempt to close the socket, because if we attempt to close it while the timer thread holds a reference, it isn't actually destroyed until the timer thread finishes its operation. Bug: chromium:1127774 Change-Id: Id6a883b14796e8f5bf1c2990f3d9d389d72c8a46 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020-15987: Use after free in WebRTC (1/2)Taylor Brandstetter2020-10-271-1/+30
| | | | | | | | | | Manual backport of patch originally reviewed on https://webrtc-review.googlesource.com/c/src/+/184283: Check length before dereferencing SCTP notifications. Bug: chromium:1127774 Change-Id: I6ccf1f5246dfacb26f480bac899f295f89b53d08 Reviewed-by: Michal Klocek <michal.klocek@qt.io>
* Fixup: [Backport] CVE-2020-16001: Use after free in mediaMichael Brüning2020-10-261-2/+2
| | | | | | | The constants were moved into the blink::mojom namespace. Change-Id: Id302cdc95048f3dd6aceb2fb183a31555e11181b Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020-16001: Use after free in media.Michael Brüning2020-10-223-1/+9
| | | | | | | | | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2472397: Validate input of MediaStreamDispatcherHost::OpenDevice() This method forwards to MediaStreamManager::OpenDevice(), which DCHECKs for the stream type to be device video or audio capture (i.e., webcam or mic). However, MSDH admits other stream types, which cause MSM::OpenDevice to hit this DCHECK. This CL ensures that a message containing an incorrect stream type, which could be sent by a malicious renderer, results in killing the renderer process. Bug: 1135018 Change-Id: I3884dde95d92c41f44966a8ab1dd7bdfd4b23b9b Reviewed-by: Michal Klocek <michal.klocek@qt.io> (cherry picked from commit 6475589b7edc2f70f1b9fd4c1cf49b39d82b267b)
* [Backport] CVE-2020-16002: Use after free in PDFiumTom Sepez2020-10-221-1/+1
| | | | | | | | | | | | | | | | Cherry pick of patch originally reviewed on https://pdfium-review.googlesource.com/c/pdfium/+/75090: Reverse order of CPWL_ListCtrl and CPWL_List_Notify cleanup (Speculative) fix for the crash in 1137630, since it only reproduces sporadically on my system, but hasn't re-occured since applying the patch. Bug: chromium:1137630 Change-Id: I4f52c7109eca00dfa8faee9bc6341cd94c25b60c Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Tom Sepez <tsepez@chromium.org> Reviewed-by: Michal Klocek <michal.klocek@qt.io>
* [Backport] CVE-2020-16003: Use after free in printingLei Zhang2020-10-222-1/+4
| | | | | | | | | | | | | Cherry pick of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2454732: Prevent double deletion in PrintRenderFrameHelper. Bug: 1134960 Change-Id: I765cc3f1463fce4b8d7c2ca99f429031566a4645 Reviewed-by: Jesse Schettler <jschettler@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org> Cr-Commit-Position: refs/heads/master@{#814756} Reviewed-by: Michal Klocek <michal.klocek@qt.io>
* [Backport] CVE-2020-15989: Uninitialized Use in PDFiumLei Zhang2020-10-221-2/+7
| | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2321339: Check FPDFText_GetCharBox() return value in pdfium_page.cc. Make sure the call succeeds before continuing, to avoid potentially using uninitialized values. Bug: 1108351 Change-Id: Ife6f8b861a53cad0bbaec8130eef0dd1341ab71c Reviewed-by: Michal Klocek <michal.klocek@qt.io>
* [Backport] CVE-2020—15973: Insufficient policy enforcement in extensions (2/2)Marijn Kruisselbrink2020-10-2117-174/+218
| | | | | | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2330314: [FileAPI] Split BlobUrlRegistry of from BlobStorageRegistry. In preparation for making BlobUrlRegistry per storage partition, this cleans up the code a bit by moving all the blob URL logic to a separate class. This CL itself should not have any behavior changes, it is purely moving code around. Also fixes some naming inconsistencies (URL vs Url among others). Bug: 1106890 Change-Id: I0e7f9f8597d280ee2f69286c5a35cbb3997c9d9f Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020—15973: Insufficient policy enforcement in extensions (1/2)Marijn Kruisselbrink2020-10-219-49/+78
| | | | | | | | | | | | | | | | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2330311: [FileAPI] Refactor how BlobURLLoaderFactory works. Rather than resolving a mojo Blob to a BlobDataHandle, just forward the request to the Blob and let the blob create the URL Loader. This way we decouple BlobURLLoaderFactory from any blob internals, paving the way for further separating out the blob URL registry from the rest of the blob system, making it easier to change where parts of the blob system live (for example moving Blob URLs to be per storage partition, or in the future perhaps per agent cluster). This does increase binary size significantly because the added mojom method results in java bindings being generated for a lot of interfaces and structs that were previously not generated. In the future this increase can be eliminated by making it possible to tag methods in mojom files with what languages they should create bindings for, or by rewriting BlobURLLoader itself to operate on a mojo Blob, rather than forward the entire URLRequest to the mojo blob (https://crbug.com/1111835). Bug: 1106890 Binary-Size: Size increase is unavoidable (see above). Change-Id: I4fa3c6a5ddf6f8be5ce299e9d1fd95eaef75ec5e Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020-15977: Insufficient data validation in dialogs.Peter Boström2020-10-212-0/+9
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2357520: Do not accept dialogs on key-repeat This prevents dialogs from being instantly accepted on Mac if holding VKEY_RETURN as the dialog pops up. Specifically it addresses a security issue where people can be tricked into holding enter while the site pops up a folder-selector dialog. The attack worked as following. The attacker tricks the victim into holding VKEY_RETURN. While the key is being held, the attacker triggers folder upload, which the OS auto-accepts by holding VKEY_RETURN. Before this change, FolderUploadConfirmationView would trigger on Mac with this VKEY_RETURN. With this change, the victim at least has to re-press VKEY_RETURN, which they are presumably at least more likely to notice. This also adds initialization of the EF_IS_REPEAT flag on Mac from [NSEvent isARepeat]. Hopefully that doesn't have wider negative consequences where ui::Event and NSEvent disagrees on the definition of a repeat event. Bug: 1097724 Change-Id: I3bb0589867cbfa059a2406e73ef43e66b5326fbc Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] Security bug 1107824Antonio Sartori2020-10-211-1/+2
| | | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2316105: Make CSP default-src without 'unsafe-eval' block eval in iframes This CL fixes the fallback behaviour of the Content Security Policy script-src to default-src with regards to blocking eval in iframes and, under certain conditions, when navigating to a new page. Bug: 1107824 Change-Id: Ia5cbe82188fde25cec8ccb5a09322e598a419434 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] Security bug 1125199Bruce Dawson2020-10-211-2/+4
| | | | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2407065: Avoid use-after-free SetNotWaitingForResponse can trigger a message pump which can then free the object which |this| points to. This use-after-free can be avoided by not dereferencing |this| after the call, by ensuring that calling SetNotWaitingForResponse is the last thing done. Bug: 1125199 Change-Id: Ie1289c93112151978e6daaa1d24326770028c529 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020-15974: Integer overflow in BlinkAdam Rice2020-10-211-0/+3
| | | | | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2296549: TransformStream deserialize: perform additional validation The TransformStream deserialization code did not check for a message port index of 0xffffffff. Add a check for it. Also add a unit test for this condition. BUG=1104103 Change-Id: Ic04af9d7a27171c471c0125662ee68dccb88abb4 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020-15992 Insufficient policy enforcement in networkingTakashi Toyoshima2020-10-211-0/+8
| | | | | | | | | | | | | | | | | | Manual Backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2355534: Network Service: Add method validation as the secondary check Today, the method valid of the HTTP request is validated in Blink for user exposed APIs such as Fetch and XHR to conform the RFC 7230. But it's still possible that compromised renderers insert arbitrary ASCII strings to the method value. This patch adds the same RFC 7230 token check in the network service as the secondary check. Bug: 1110195 Change-Id: Ia99a986f82034875f7e8c0b2224f2260a99eeffa Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020-15976: Use after free in WebXRAlexander Cooper2020-10-211-1/+5
| | | | | | | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2385885: Update ReferenceSpace reset to operate on a copy When a reference space's origin is reset, an event is dispatched to the page. This allows additional javascript to run, which could request a new reference space. Depending on the timing of this new reference space request returning, it would cause the reference_spaces_ list to be modified, and thus invalidate the iterators. Fix this by iterating over a copy of the list. Fixed: 1123522 Change-Id: I121a5fa3dde1cfc18abbef579148a2ace86f73f5 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020-15979: Inappropriate implementation in V8.Georg Neis2020-10-211-4/+3
| | | | | | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/v8/v8/+/2416383: Merged: [turbofan] Fix bug in inlining Revision: 219b28bfe2ea76de63f034eb75b67e8ded339d94 BUG=chromium:1127319 NOTRY=true NOPRESUBMIT=true NOTREECHECKS=true R=tebbi@chromium.org Change-Id: I98e77bac81e2cf822a4a4987115e0cf01b1dbc52 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020-15972: Use after free in audio.Raymond Toy2020-10-215-56/+105
| | | | | | | | | | | | | | | | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2410743: Add mutex for allowing the graph to be pulled Basically add a mutex to protect access to allow_pulling_audio_graph_. The main thread waits until it has the lock before setting this. This prevents the main thread from deleting things while the audio thread is pulling on the graph. A try lock is used so as not to block the audio thread if it can't get lock. This is applied to both real time and offline contexts which required moving the original real-time-only implementation to audio_destination_node so we can use the same methods for the offline context. Tested the repro case from 1125635, and the issue does not reproduce. We're assuming this will fix 1115901, but I've not been able to reproduce that locally. Bug: 1125635, 1115901 Change-Id: I1037d8c44225c6dcc8fe906c29a5a86740a15e1d Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020-15968: Use after free in BlinkRobert Phillips2020-10-211-0/+3
| | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2405644: Disallow creation of CanvasResourceProviders for zero sized images Bug: 1126424 Change-Id: I17ddbdce78d89a997a73c37f18cd945b83936f7f Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020-15969: Use after free in WebRTC.Michael Brüning2020-10-211-2/+5
| | | | | | | | | | | | | | | | | | | | | | Manual backport of patch originally committed on https://github.com/sctplab/usrsctp/commit/ffed0925f27d404173c1e3e750d818f432d2c019: Fix a use-after-free bug for the userland stack. Thanks to Mark Wodrich from Google for finding and reporting the bug. Cherry pick fix for usrsctp UAF bug to M86. The commit being cherry picked is: https://github.com/sctplab/usrsctp/commit/ffed0925f27d404173c1e3e750d818f432d2c019 TBR=hta@chromium.org Bug: chromium:1124659 Change-Id: I229c8a01fd051b1dfe499a609f72484ce7611af3 Reviewed-by: Michal Klocek <michal.klocek@qt.io> (cherry picked from commit 0f55630c2f4e811da4e674b217add4d8022efcac) Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020-15978 Insufficient data validation in navigationLukasz Anforowicz2020-10-211-0/+5
| | | | | | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2385921: Avoid fixing/rewriting/mutating invalid URLs in RewriteURLIfNecessary. This CL changes BrowserURLHandlerImpl::RewriteURLIfNecessary so that it returns early(and doesn't mutate the |url| in the in-out argument) if |url| is invalid. This helps avoid scenarios where RewriteUrlForNavigation (in navigation_controller_impl.cc) ends up generating a NavigationEntry with an invalid virtual URL that (accidentally/incorrectly) gets rewritten into a valid URL. Bug: 1116280 Change-Id: I114cf8c8d9459b6931ae659f62a100679b994d5e Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* CVE-2020-15999: Heap buffer overflow in freetypeWerner Lemberg2020-10-201-7/+7
| | | | | | | | | | | | Manual cherry-pick of fix in freetype: Fix heap buffer overflow (#59308). This is CVE-2020-15999. * src/sfnt/pngshim.c (Load_SBit_Png): Test bitmap size earlier. Change-Id: I4f5fd188f268de8929d2759f64cec7cb55644c42 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020-6571: Incorrect security UI in OmniboxJoe DeBlasio2020-10-121-1/+1
| | | | | | | | | | | | | Manual backport of patch originally reviewed on: https://chromium-review.googlesource.com/c/chromium/src/+/2212807 [IDN Spoof Checks] Add U+0A5C to digit lookalikes. This CL adds U+0A5C as a lookalike character to the digit 3. This change appears to impact no domains seen in UKM. Fixed: 1085315 Change-Id: Iea9930363d853f154e2d781646a1b0b5da7fbbfd Reviewed-by: Michal Klocek <michal.klocek@qt.io>
* Fix assert when checking if RendererAppContainer feature is enabledPeter Varga2020-10-051-2/+5
| | | | | | | | | | | | | | | It asserts with MSVC 32bit but it might fail with other configurations too because the RendererAppContainer feature will have more than one instance in the memory and it can break pointer-based identity lookup (see documentation of base::Feature in chromium/base/feature_list.h). Make the Feature instance static to ensure it has the same address for each of the IsAppContainerEnabled() calls. Change-Id: I937cbd7c4dd46df2d4b48c4af239db1e6e8feb6f Fixes: QTBUG-86945 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* [Backport] CVE-2020-15964: Insufficient data validation in mediaJohn Rummell2020-09-285-7/+53
| | | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2378889: Check for context destroyed in MediaKeys Don't allow calls to proceed once the associated content has been destroyed. Bug: 1121414 Test: example in the bug no longer crashes Change-Id: I3bdeb86f2020f684958b624fcc30438babfb5004 Reviewed-by: Michal Klocek <michal.klocek@qt.io>
* [Backport] Security bug 1111149Guido Urdaneta2020-09-281-1/+3
| | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2377811: [ElementCapture] Do not produce frames if media player is tainted This prevents potential cross-origin mid-stream redirects from braking the cross-origin restrictions. Bug: 1111149 Change-Id: I18d05a5836b9a390dec50e10c43d3d2b9ec5915a Reviewed-by: Michal Klocek <michal.klocek@qt.io>
* [Backport] CVE-2020-15960: Out of bounds read in storageRamin Halavati2020-09-2810-50/+161
| | | | | | | | | | | | | | | | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2398701: Reland Run ObfuscatedFileUtilMemoryDelegate entirely on TaskRunner. MemoryFileStreamWriter called some ObfuscatedFileUtilMemoryDelegate functions through IO thread while other functions in OFUMD are called on a threadpool sequence. This could result in races in updating directory structure. To fix the issue, MemoryFileStreamWriter and MemoryFileStreamReader are updated to call all OFUMD on the default task runner of the file system context. This CL was landed in crrev.com/c/2308721 and reverted due to flakiness. The flaky crashes are believed to be because the buffer passed to MemoryFileStreamReader::Read and MemoryFileStreamWrite::Write are not thread safe. Patchset1 is a copy of the previous CL and the issue is fixed in the next patchsets. Bug: 1100136 Change-Id: I619b82c2f4d23a020e9ce7e5e6c16980907b501b Reviewed-by: Michal Klocek <michal.klocek@qt.io>
* [Backport] CVE-2020-15962: Insufficient policy enforcement in serialReilly Grant2020-09-282-4/+20
| | | | | | | | | | | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2393001: serial: Check that port is open before reading or writing This change adds checks to the platform-specific implementations of Read() and Write() to make sure that the file descriptor is valid before. This makes the assumptions validated by later DCHECK correct. This cannot be done in the platform-independent layer because test code depends on being able to call some SerialIoHandler methods without an actual file descriptor. Bug: 1121836 Change-Id: If182404cf10a2f3b445b9c80b75fed5df6b5ab4b Reviewed-by: Michal Klocek <michal.klocek@qt.io>
* Only upgrade to WebSocket when WebSocket was requestedAllan Sandfeld Jensen2020-09-221-1/+1
| | | | | | | | Otherwise we can end up rejecting pre-upgraded HTTP/2 requests. Task-number: QTBUG-86824 Change-Id: I0d7e4a716729c4bba8fff8351d08c54285319a60 Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
* [Backport] CVE-2020-15961: Insufficient policy enforcement in extensionsAndrey Kosyakov2020-09-221-1/+21
| | | | | | | | | | Manual backport of patch originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/2380855: Delegate TargetHandler::Session permission checks to the root client Bug: 1114636 Change-Id: Iba3865206d7e80b363ec69180ac05e20b56aade2 Reviewed-by: Michal Klocek <michal.klocek@qt.io>
View Lorry Controller Status