- Jan 02, 2019
-
-
George Kulakowski authored
Test: no functional change; booted under arm qemu with asan Change-Id: I5372d6ac791b678bf8536fb9120e803b73195e40
-
Tamir Duberstein authored
Test: CQ Change-Id: Id3247b39b457a88e239a18a1b131511e66d48c67
-
Adam Barth authored
Test: No behavior change Change-Id: I63aa1b9a6d6e8d82c4821875de3e9ab9af425d0d
-
Adam Barth authored
These functions are the same as the default versions. Test: No behavior change Change-Id: Ia035137f8d2756118acf4585a5bac692f0d26a53
-
Tricia Landers authored
During a JournalEntry sync callback, it is possible for the entry to be deleted immediately after the status is updated, but before the Journal::SendSignal call. To prevent this, update the entry status as part of the SendSignal call (now renamed to ProcessEntryResult). ZX-3064 #done ZX-3182 #done Test: Manually tested by forcing the thread to sleep between setting entry status and sending signal to the journal. Change-Id: I97bb10b3cadb70221fd13bb21caa2912145599ce
-
Tricia Landers authored
Rename WritebackWork::Reset() to MarkCompleted(). Also update the MockJournal to store entries enqueued from the JournalProcessor, which can later be processed at our discretion (or not at all). Test: JournalProcessorResetWorkTest Change-Id: I72496feed9a1cfbee70d62db8a2d915cd973eea2
-
Christopher Tam authored
Fixing minor nits Test: no behavioral changes Change-Id: I70673deefa67eb3cb4f8e4b3065bad5f4ab30925
-
Todd Eisenberger authored
There are no out-of-zircon dependencies, so this is a full rename. Test: CI Change-Id: I8fc7884e35f7c0c1137122e19219c7f7e672f108
-
Todd Eisenberger authored
There are no out-of-zircon dependencies, so this is a full rename. Test: CI Change-Id: I7f62c02f96babe10417b0b03917d57aff8525c13
-
Braden Kell authored
Add implementations to the MmioBuffer weak specialization methods so that they work in the drivers as well as the tests. ZX-3222 Test: lsblk works on Cleo Test: tests pass Change-Id: I75ba9a0888acbc071158e876a18dd4a686498b4f
-
- Dec 29, 2018
-
-
Abdulla Kamar authored
We should be setting the VFIQEn bit, to present group 0 virtual interrupts as virtual FIQs, as this is required by a virtual machine that enables GIC system register access. Test: Ran "hypervisor-test". Change-Id: Ie7d1016c087cb4a9cf3eee8914f915a78459b857
-
Abdulla Kamar authored
When storing arguments for HVC invocations, ensure we use caller-saved registers. Since we control the ABI between EL1 and EL2, this change gives us some additional flexibility with the number of arguments that can be passed to these functions without trampling on the additional arguments that need to be set for EL2 to decode the HVC invocation. Test: Ran "linux_guest" on a VIM2 and QEMU. Change-Id: Ie9ca97674f90f4aefb661d90a6ad4c4d6469cd7a
-
- Dec 28, 2018
-
-
Roland McGrath authored
Test: CQ & manual Change-Id: Ie32208d7e9af6fe7407446ba066b77aae3089bd8
-
Roland McGrath authored
The old names zircon.bin and bootdata.bin are no longer used. The run-zircon script no longer supports the deprecated -x switch that required direct Multiboot kernel booting support or a shimmed kernel. The only way to specify the ZBI (and thus the kernel) is with the -z switch. Bug: ZX-2614 #done Test: CQ Change-Id: Id68af0b0e21c5019c6fb40d2087c2c30a4a15d8a
-
Venkatesh Srinivas authored
make clean attempted to remove every output object file at once. This overflows ARG_MAX on Mac OS with a standard zircon tree. Remove objects, dependency files, and generated files one-at-a-time to avoid ARG_MAX limits. This may negatively affect build times. Tested: Mac OS 10.13.6: (ARG_MAX = 256K) - make clean works Change-Id: I88e1e3046518904d2df9e13e2077c35fde5dfba9
-
Roland McGrath authored
The kernel is always started in 64-bit mode with the ZBI protocol. The way to boot via Multiboot is using the multiboot.bin trampoline. Bug: ZX-2225 #comment Multiboot support removed from kernel Test: CQ Change-Id: Iff577042285a578c4629ccb4a838e8950a5b3ea5
-
Abdulla Kamar authored
More than one APR register may be in use, therefore we should make sure that we correctly save and restore all of them, based on the number of preemption bits that the hardware supports. Furthermore, for GICv3 we should be saving and restoring active priorities group 1 registers. Test: Ran "hypervisor-test". Change-Id: I6a7ea92ae7a79da3f3f2126b07a9929467d7b2d6
-
Todd Eisenberger authored
In this batch: display, gpu-clock. The temporary ordinals added to the display interface are removed, too. Test: CI Change-Id: I7488c361855dbbd27dd2ab7ee053a3f1b3403340
-
Abdulla Kamar authored
Ensure that all LRs are within range for the specific GIC version. Test: Ran "hypervisor-test". Change-Id: If96a0b81bef70ea74b230ac19691200235429e1c
-
- Dec 27, 2018
-
-
Todd Eisenberger authored
This will be a breaking change that requires a manual roll. The fuchsia-net-stack dependency makes this difficult to do without that. Test: CI Change-Id: I639d115f11ca8609c2537f01daae1fc21dec93cd
-
Todd Eisenberger authored
There are no out-of-zircon dependencies, so this is a full rename. Test: CI Change-Id: I9deace12efa32bcac8a3023b07ec3da8eb0e59cc
-
Todd Eisenberger authored
There are no out-of-zircon dependencies, so this is a full rename. Test: CI Change-Id: I706441378927e0cb98ea45e31914a036b4b5e9e0
-
Todd Eisenberger authored
There are no out-of-zircon dependencies, so this is a full rename. Test: CI Change-Id: I73de1079a8cee80069abe0c31124afa9d53fc301
-
Marty Faltesek authored
Interface was previously transitioned. Test: None Change-Id: I7b554d862411bcd2792123f7c70cf4f76efb66bb
-
Todd Eisenberger authored
In this batch: block, nand, skipblock Test: CI Change-Id: Id6abb878e607bdbac44ab225200ccbe6dad892c2
-
Todd Eisenberger authored
In this batch: backlight, input, rtc. Test: CI Change-Id: I7278814ec4efa381dab5087668706388bb4d08b1
-
Todd Eisenberger authored
In this batch: block, nand, skipblock. The interfaces are copied rather than moved because of external dependencies. Since these interfaces all use manual ordinals, it should Just Work. Test: CI Change-Id: If7d4f35ce91a256df9ed6b8dbaef924bfc7b74d4
-
Todd Eisenberger authored
It was previously in the same directory as zircon-nand? Test: CI Change-Id: I26673aed33c0c195c004e5cd5d85f57093ce2a13
-
Todd Eisenberger authored
In this batch: display, gpu-clock. The interfaces are copied rather than moved because of Garnet dependencies. As part of this, we temporarily assign ordinals to the interface to make the transition easier. Test: CI Change-Id: Ib1cd87e3311265c84e0ac307f1379e99b0e0033c
-
- Dec 26, 2018
-
-
Yifei Teng authored
Emit additional code to check the size of vectors and strings when an explicit maximum is declared in the FIDL definition. This prevents stack overflows introduced by invalid messages with a large size/count field. FIDL-385 #done Test: fidl-simple-test Change-Id: Ifc68681c2ecb97d222339a572caef1744eedead0
-
Todd Eisenberger authored
There are no out-of-zircon dependencies, so this is a full rename. Test: CI Change-Id: Ia3f8aa6a8812ae2fc304a9fa8b9bef107649d02e
-
Bruno Dal Bo authored
* Ethertap was leaking a thread handle for "ethertap-thread" * Core ethernet driver was leaking tx fifo buffer memory TEST: ran tests in garnet layer that use ethertap Change-Id: Ic0d87cfec71f932cf183faa4d4b83873140e23fa
-
Todd Eisenberger authored
In this batch: backlight, input, rtc. The interfaces are copied rather than moved because of Garnet dependencies. Since these interfaces all use manual ordinals, it should Just Work. Test: CI Change-Id: I2dbf2da6bc3321444fb85abe0214216e5b8bb49f
-
Todd Eisenberger authored
This reverts commit 9b2ea3dc. Reason for revert: Broke GI, rename scripts accidentally modified original fidl files Original change's description: > [fidl][hardware] Begin renaming hardware interfaces > > In this batch: backlight, input, rtc. The interfaces are copied rather > than moved because of Garnet dependencies. Since these interfaces all > use manual ordinals, it should Just Work. > > Test: CI > > Change-Id: If8d976ed4cb6dfcf35b5af24fdceed5d3d8c6052 TBR=voydanoff@google.com,kulakowski@google.com,teisenbe@google.com Change-Id: I581132718870d5b2d507127a1abff4f4641aa7df No-Presubmit: true No-Tree-Checks: true No-Try: true
-
Vikram Auradkar authored
This is a first patch among few to convert gpt library to C++. In addition to fixing compilation errors, patch also - adds namespace - changes "typedef struct" to "struct" - rearranges few functions to make them fall under one anon namespace - adds const qualifier - retains few changes proposed by clang-fmt Test: - Paved a device - Compiled, booted qemu and dumped gpt table with $ gpt dump /dev/sys/pci/00:1e.4/pci-sdhci/sdhci/sdmmc/sdmmc-block/block $ lsblk - On device and on qemu ran system/test/sys/chromeos-disk-setup-test Change-Id: I3e998bea20d1632ea6a19e1a0ff0b360c6b3676e
-
Todd Eisenberger authored
In this batch: backlight, input, rtc. The interfaces are copied rather than moved because of Garnet dependencies. Since these interfaces all use manual ordinals, it should Just Work. Test: CI Change-Id: If8d976ed4cb6dfcf35b5af24fdceed5d3d8c6052
-
- Dec 23, 2018
-
-
Brijen Raval authored
ZX-2579 #done Test: boot up on vim2, and validated temperature Change-Id: I50eea19a015f1160c9c51bcf39586f7c1c143872
-
- Dec 22, 2018
-
-
Suraj Malhotra authored
All drivers now specify ddk::base_protocol as necessary so we can remove DefaultProtocol from banjo. Tested: booted on qemu, vim2, pixelbook, and astro. Change-Id: I90163291b619b18bc64d04be63fd5631e8f74637
-
Carlos Pizano authored
Since we are willing to give a restricted root job to svchost, crash handling can be greatly simplified. The JIRA bug below has a diagram, but the TLDR is that crashsvc and devmgr were doing very little other than redirecting. - Remove crashsvc process and all that code except for the core that moves to svchost - Remove code in devmgr that acts as middleman for crashes - Remove the crashanalyzer FIDL svc code use the inspector library directly. The logic of selecting which analyzer to use is kept as before. ZX-3199 #comment progress TEST=manual: use $crasher and inspect that dumps are seen in the log output, otherwise unittests suffice Change-Id: I84b24c85437ea0414e25e3fdabd53b55bb730257
-
Sean Klein authored
This patch removes cache policy and management from the VnodeBlob class, instead relocating it into two new isolated classes. ZX-3180 #comment Done ZX-3112 #comment In Progress Test: Cache Node tests added Change-Id: Ia562861ba891003dad317fff117101a1236de0bd
-