- Apr 02, 2019
-
-
Joshua Seaton authored
Change-Id: I9c9d8f2a043c0aad1272ecc6ca31f07e1ae40e73
-
Joshua Conner authored
Reference the cipd.gni file in build, not the one in garnet. WLAN-1031 Test: Builds Change-Id: I92cc37b3a76242be83a0be4417dd6eb1c70448e9
-
Vickie Cheng authored
/system/data is being deprecated so we should use /config/data for modular configs Change-Id: I85c35f413eeea00bb5d46d47fa93c5ba479e49e1
-
Mark Schott authored
Previously, ktl was using templated using statements or simple function wrappers to create aliases for std types, which is fine in some cases but sometimes leaves out members of an overload set that could help with optimization or allow different arguments. In fact, in building with this I already caught a few misuses of ktl::move since clang's std::move plugs into special compiler warnings that the previous wrapper wasn't able to use. Now we just use simple using declarations to import the std types we want exactly as they're defined in the STL. TEST: no functional change Change-Id: Ib890527daec7b4847fc912df393e129d57425b71
-
Christopher R. Johnson authored
Change-Id: I7b4b67ebfed9eba3d75d4d1d0ee2937d66332bd2
-
Ryan Heacock authored
NET-2150 #done Testing: net_cli if add /dev/class/ethernet/000 net_cli if list (shows interface 1) net_cli if del 1 net_cli if list (no longer shows 1) net_cli if add /dev/class/ethernet/000 -> Added interface 2 Change-Id: I5d89763ca514ced7bb20b240e1cb458f617b387c
-
Fadi Meawad authored
TraceEvent Ids can be of the form: 7, "42" or "0x123", but we were only handling the simple numeric version. This CL handles all type of Ids a typical json trace have. DNO-448 #done. Change-Id: Ib67c4a03fd33bd8e6568a490b0119fad9ff403a8
-
John Grossman authored
Temporarily disable priority inheritance in the blocking reader/writer lock. Currently, there is a small race which can prevent writers from properly receiving priority pressure, and readers don't really participate in any sort of priority inversion avoidance behavior. The POR for now is to shut off BRW PI for the time being, to split the lock into two different flavors, one which implements PI behavior and one which does not bother. The PI enabled flavor will make readers holding the lock non-preemptable (and probably non-blockable as well), while a writer inside of the lock will become the owner of the lock's WaitQueue (and therefor automatically receive the priority pressure of the waiters) For now, disabling this will allow the new OwnedWaitQueue class to take exclusive responsibility for the inherited priority of threads, and for propagating priority pressure through the PI chain when needed. Tests: build and runtests in QEMU Change-Id: Ifbed6494f24a377318988f5b73b9b567cfef2e08
-
James Tucker authored
Bug: PKG-630 #comment Cobalt has API to report the channel Change-Id: I81c635306b9804ec8b80a50ef9c36e5b7ca5cf67
-
Tamir Duberstein authored
This was added in https://github.com/google/netstack/commit/0754aa9 and allows us to avoid repeating the local address handling code in our ethernet endpoint. Change-Id: Ifcbdf969dead30dfb731371833b6f8dfdcce16e0
-
P.Y. Laligand authored
Bug: DX-1200 Change-Id: I6ce6c2b43746d3734fd2ded0fbf78bb3d4e2cf0d
-
Ghanan Gowripalan authored
BUG: NET-2184 #done - Create hermetic Log and LogSink services with configurable LogListener - Improve logging by ManagedLogger Change-Id: I8a86b25e4eba8c08ea9f61f3248d6ed14930ef11
-
Arjun authored
realm_integration_test depends on echo_server_cpp Change-Id: I77e9f5637212bf0a33efa43f409c4e20e319d960
-
Francois Rousseau authored
* this will get rid of the "file_io_posix.cc:148] open /tmp/temp_dir_PICHHi/pending/$uuid.lock: File exists (17)" on failures * this is what Crashpad does: https://chromium.googlesource.com/crashpad/crashpad/+/master/handler/crash_report_upload_thread.cc#230 TESTED=`fx run-test crashpad_agent_tests` Change-Id: I3c2d1cb905f2be8d518239124dc6a5a0252bec5a
-
Eric Wang authored
Change-Id: Ic0d57e27edd6ae12d3fb0fa9ba620db9d5288c80
-
Scott Graham authored
Useful to see what has the oom-killer bit set. ZX-3637 #comment [ps] Add STATE column for jobs to display KILL_ON_OOM bit Test: Manual runs of `ps` with https://fuchsia-review.googlesource.com/c/fuchsia/+/264517/ running over `fx shell` Change-Id: I42267e52375df5de42c257e863ad4a87d45f0a5b
-
James Tucker authored
This should be wrapped in scripts that use fx-config-read, and so should instead consume $FUCHSIA_BUILD_DIR. Note that `out_dir` is a misnomer here, and should be renamed to build_dir. I avoided changing the public flag surface as I have no way to assert if that breaks other things. Change-Id: Ie38fb874b9360e597deb6d5c28d14ba329e1cf8d
-
Carlos Pizano authored
ZX-2264 #done ZX-2967 #comment progress TEST: exsiting utest sufice Change-Id: Ic4eb98881f99ec4193953ecba6a4e006b2ff783c
-
Leo Lung authored
When a component uses a capability, bind to the component that the capability comes from, potentially causing it to start. Added two more tests to test sibling propagation without going through the root. CF-647 #done TEST= fx set core.x64 --with //bundles:buildbot fx run-test component_manager_tests -- --nocapture Change-Id: I1c9f0a95d338486308d4a2f16b957465f4e49a05
-
Dustin Green authored
We should be checking that the output frame format says uncompressed, not compressed. Tested: fx shell "present_view fuchsia-pkg://fuchsia.com/use_media_decoder#meta/use_media_decoder.cmx --h264 --imagepipe --fps=24 /tmp/bear.h264" Change-Id: I05fd3c9ef0cb069e47f6e4924a14ef9b70050518
-
Doug Horn authored
This allows proper integration with code that relied on the historical usage of __unused as an alias for __attribute__((__unused__)). Bug: 129725611 Change-Id: I6d77d4ba6c4d823bc9a3d3bab32e7a6c9c5fa2d4
-
Konstantin Pozin authored
- unic-char-property - unic-char-range - unic-common - unic-ucd-block - unic-ucd-category - unic-ucd-version Approved in OSRB-154 OSRB-154 #comment TEXT-60 #comment Change-Id: Ic033adbb72acaaad642bc64b8e6ffa9cac5208db
-
Joshua Seaton authored
Test: no ninja warnings on a build Change-Id: Icdf228da90200ba8f772271f2626286959a74b1b
-
Gary Bressler authored
`to` is more symmetric with `from` and less to type. `offer.targets.to` becomes `offer.to.dest`. Change-Id: Ia3e456a564b88495c3d30ed74f880ca4ac718537
-
Mike Voydanoff authored
Platform bus product and device IDs (PIDs and DIDs) are relative to a specific vendor ID (VID), so it is unsafe to mix PIDs and DIDs with a different VID. TEST: drivers still bind on VIM2 Change-Id: Ifb662926397c72100790ac9dcd7b484367973b91
-
David Gilhooley authored
Re-enable hardcoded touchscreens because the generic driver is having issues on astro. DNO-455 #done Change-Id: I769dc7182aaf58c550dc44621facb41c4150d395
-
Tamir Duberstein authored
TC-417 #comment Change-Id: Ifb54bd91ffea2833b921c3c3328dcc1a408d637c
-
Brett Wilson authored
How to trigger an attached debugger from within code. Change-Id: I4ebc1a0cd7f0804a139d062b079cc8c6074bee31
-
Mark Schott authored
`find -perm +<octal bits>` has been deprecated in GNU find since 2005 and apparently no longer works on gLinux. This causes the command list not to be printed when calling fx with no args. TEST: run fx with no args (or `fx help`) Change-Id: I13cbb7bab4b8d4e7dd49e5b8805c15aecd812eb8
-
Ariana Mott authored
Logs an error for clients that try to create more than one root view per session. SCN-1249#comment Tested: - ran default astro - ran unittests and triggered CI - removed unused unittest which created multiple (currently unused) Views. Change-Id: I550382dbe2c19904222c78f2bb2380203480427d
-
mesch authored
This reverts commit f9187b11. Reason for revert: Allegedly broke Global Integration, because it reached GI before prerequisite CL in other repository. Original change's description: > [e2e] Factor out trace dump logic. > > So we can reuse it in the audio facade for dumping captured audio. > > Change-Id: I964d37e1fcf428f36d66e7dda542d0413e526a21 TBR=mesch@google.com,rosswang@google.com,isma@google.com Change-Id: I8300fbe0be2460132050e27d9c97ce7179321732 No-Presubmit: true No-Tree-Checks: true No-Try: true
-
Jeremy Manson authored
Because we were previously keeping track of position in the message with a uint8_t*, and tracking handles requires both a uint8_t* and a position in the handle table, we change places where we used uint8_t* to track our position to use a new Marker interface that keeps track of both at once. Change-Id: I7abd84f7728ddea163874c8f33bb9ff2e88d775f
-
James Tucker authored
Change-Id: Id0b9a54fd9ce6d0742b6522bc89ac167128c005d
-
Konstantin Pozin authored
OSRB-152 #comment Change-Id: Iff4ac7b2891f1f78d0f4677be1264b384f541a82
-
Kiet Tran authored
Test: 1. fx run-test wlanstack2-tests 2. Sanity test connecting to network and replugging device Change-Id: I6a5e7366e988a9449022bdedbcf9913e16f009c9
-
John Grossman authored
Introduce the concept of an OwnedWaitQueue as an extension to WaitQueue. Owned wait queues implement a concept of ownership for the purposes of priority inheritance. In addition, move the concept of ownership out of the FutexContex::FutexState code and push it down into the OwnedWaitQueue level, making sure that all of our futex ownership operations still pass their tests. Actual priority inheritance still needs to be implemented, but at this point, this should be as simple as implementing OwnedWaitQueue::ApplyPriorityPressure and OwnedWaitQueue::RelaxPriorityPressure (the easy part), and also writing a set of tests (the harder part). Once this has been completed, we can switch kernel mutexes over from using simple WaitQueues to OwnedWaitQueues instead, and then finally switch the user mode sync objects which need to implement PI over to a form where they signal ownership properly. Once we reach this point, we should finally have PI working for both kernel and user mode sync objects, as well as working together (eg; user mode futex owners will properly apply PI pressure to kernel mode mutex owners when blocked by them) ZX-1798 #comment Futex ownership has now been generalized to ownership of wait queues. Change-Id: Iee248d8bc348fd81f704fd027f6a133fc3814936
-
Mike Voydanoff authored
Removing some remnants of support for proxying custom protocols TEST: builds Change-Id: I29e0c25eb3746194398f5785b73b5491338852ed
-
Steffen Meschkat authored
So we can reuse it in the audio facade for dumping captured audio. Change-Id: I964d37e1fcf428f36d66e7dda542d0413e526a21
-
Brett Wilson authored
Hooks up template matching to name lookup. This was mostly implementing a TODO in global name lookup and factoring the code so it can be shared by both implicit and exact name lookup. Some prefix matching code has moved from the root ModuleSymbolIndex to the ModuleSymbolIndexNode so that a prefix search can be done from anywhere in the index tree. Change-Id: I5b202a2633ef8ba3177817f7c5bdb13d21727305
-
Emircan Uysaler authored
This CL lets us schedule multiple input taps for: - convenience when testing locally - higher frequency taps for automated tests SCN-1345 #comment Change-Id: Icba030454c21e770a07c61d1dc9261ad1a80b556
-