Skip to content
Snippets Groups Projects
  1. Dec 05, 2018
    • Sean Klein's avatar
      [fs] Convert FIDL handler to C++, increase ZXFIDL_MAX_MSG_HANDLES · 33fddee9
      Sean Klein authored
      Test: CQ
      ZX-3091 #comment In Progress
      
      Change-Id: Ib04c43b355cb99db9910434b33162c6118c611e6
      33fddee9
    • Ruchira Ravoori's avatar
      [usb-audio]Remove more usage of req->node · bec3b089
      Ruchira Ravoori authored
      With the recent changes to usb stack, each driver in the usb stack has its own list_node_t*
      so as to avoid stamping on other driver's lists. This changeset moves
      audio drivers away from using the public node. Once all
      references are removed the public list_node_t pointer will be removed.
      
      Test: fx full-build. fx serve on pixelbook.
      Plugged in a usb audio device.
      audio -d 1 info.
      audio -d 1 tone.
      
      Change-Id: Ic46f82acb58074a54dd85018f99a1c357f619bc9
      bec3b089
    • Braden Kell's avatar
      [mt8167][thermal] First version of thermal driver · 00102c35
      Braden Kell authored
      ZX-2789
      Test: wrote program to call temperature ioctl,
            temperature seemed reasonable and changed
            as expected under different conditions
      
      Change-Id: Ic1adf892b13184fea364d41a44cfb90677b86949
      00102c35
    • Scott Graham's avatar
      [syscalls][rights][doc] abigen comments to reflect syscall handle rights · 862a398c
      Scott Graham authored
      This is a first pass at documenting the rights required for handles
      passed to syscalls. There's lots more to do:
      
      1) Something more like parseable syntax;
      2) More complete specification;
      3) Use this to autoupdate .md docs;
      4) Verify that each handle actually has some associated documentation;
      5) Test the markup vs. implementation using something like syzkaller;
      6) Consider generating helpers to reduce the source code length in
         kernel/syscalls/;
      ... and probably lots more.
      
      But writing down what the rights appear-to-be, as a first step.
      
      ZX-968 #comment [syscalls][rights][doc] abigen comments to reflect syscall handle rights
      ZX-2399 #comment [syscalls][rights][doc] abigen comments to reflect syscall handle rights
      ZX-2967 #comment [syscalls][rights][doc] abigen comments to reflect syscall handle rights
      
      Test: CQ, no behaviour change intended
      Change-Id: I51f53d0d5b3f28238c0a9fa158477720f8021e34
      862a398c
    • Mike Voydanoff's avatar
      [banjo][usb] Convert usb-bus protocol to banjo · 20c84177
      Mike Voydanoff authored
      TEST: USB works on hikey960
      
      Change-Id: If6d078ce6cc988d277a059a52fb27d7655d7557d
      20c84177
    • Mike Voydanoff's avatar
      [banjo][usb] Convert USB hub protocol to banjo · f8ae81bb
      Mike Voydanoff authored
      TEST: USB hub works on hikey960
      Change-Id: Iba92722a12b1c0a6e0605aa5425fe557e3c28ffb
      f8ae81bb
    • Mike Voydanoff's avatar
      [banjo][usb] Convert USB mode switch protocol to banjo · e28f34d7
      Mike Voydanoff authored
      TEST: switching USB modes on hikey960 with usbctl
      Change-Id: I7aee91336932bcf527cb41c0826db6e955f4cd80
      e28f34d7
    • Mike Voydanoff's avatar
      [fidl][usb] Use array instead of vector for device descriptor · 4144ee17
      Mike Voydanoff authored
      TEST: lsusb on hikey960
      Change-Id: Ib8104389c6e4c209f7543735bdc4834712fd2154
      4144ee17
    • Ed Coyne's avatar
      [zircon][docs] Mention k ut on the testing page. · e9fd364a
      Ed Coyne authored
      Mention how unit tests compiled into the kernel are run.
      
      Test: Previewed in gittles.
      Change-Id: Ia78db6e07c83ac063028d2b94a12492b95e51b13
      e9fd364a
    • Fadi Meawad's avatar
      [tracing] Stop splitting bounded duration events into start and end events · cf52b262
      Fadi Meawad authored
      This is step 3, where is switch the calls that already uses bounded
      trace event to use the duration complete events.
      
      This will transform the kernel syscalls and the vcpu events to
      the new duration events, almost halving the size needed for those
      in the trace buffer and in the output file.
      
      The next step is to make more calls use the bounded event records
      mainly the TRACE_DURATION macro.
      
      BUG: PT-72
      
      TEST: manually verifying that those events appear correctly in traces +
      CQ
      
      Change-Id: I6f8a71d281845b431077b914541565cd32065e0f
      cf52b262
    • George Kulakowski's avatar
      [devmgr] Class-ify Devhost in anticipation of RefPtr-izing it · 55ec689d
      George Kulakowski authored
      Test: booted under asan, no functional change
      Change-Id: I39d5b3392ec785b434cc71929401cabc4624308a
      55ec689d
    • Doug Evans's avatar
      [trace] Rename trace.static to trace.with-static-engine · 9a4ba05a
      Doug Evans authored
      And similarly, trace-provider.static to trace-provider.with-static-engine
      Much clearer.
      
      This is being done as a soft transition, this step adds the new version.
      The old version will be deleted when garnet is updated.
      
      Tested: CQ
      
      Change-Id: Idb9ab3b74d9b3cd23eccb462964fa17d65a10a6a
      9a4ba05a
    • Jocelyn Dang's avatar
      [usb-test-fwloader] Initial FX3 setup part 2. · fa96c3fc
      Jocelyn Dang authored
      Adds flash programmer driver, which will implement the same
      usb-test-fwloader fidl API as the testing firmware loader.
      
      Implementation left to future CL to keep this CL small.
      
      usb-test-fwloader app updated to wait for the flash programmer
      device.
      
      ZX-3045 #comment
      
      TEST= copy flash programmer to device
      usb-test-fwloader -b -p flash.img -f foo
      
      Change-Id: Ibf8eb84d722b951544bbb5cec5faacbb44697d8d
      fa96c3fc
    • Roland McGrath's avatar
      [kernel][ktl] fbl::unique_ptr -> ktl::unique_ptr · ab582f2a
      Roland McGrath authored
      <fbl/unique_ptr.h> -> <ktl/unique_ptr.h>
      fbl::make_unique_checked -> ktl::make_unique
      
      The ktl::unique_ptr implementation uses std::unique_ptr with a custom
      deleter to handle the fbl::Recyclable protocol.  The old fbl::unique_ptr
      implementation is not used at all in the kernel.
      
      Bug: ZX-1751 #comment kernel fbl::unique_ptr -> ktl::unique_ptr
      Test: CQ
      Change-Id: Iad2a6b3a37dc5ac9671b718f9b30081e27c864b0
      ab582f2a
    • Christopher Anderson's avatar
      [µpci] Fix ecam size used in vmo creation · 4200f942
      Christopher Anderson authored
      The bus range defined by a segment group's start_bus_num and
      end_bus_num is an inclusive one, but the vmo size was being
      calculated as inclusive only on the start_bus_num end. This
      fixes that to include end_bus_num.
      
      Test: Verified no behavior change in upci branch.
      
      Change-Id: Ie76c66a3a3222df152f626988f01b8bf0c29f27d
      4200f942
    • Felipe Archondo's avatar
      [tracing] add explicit timestamps to VTHREAD macros · 221a6bdc
      Felipe Archondo authored
      Test: trace-vthread-test
      
      Change-Id: I3352ef38a7ad8d294e1281d5691fbed511aafd0c
      221a6bdc
  2. Dec 04, 2018
    • George Kulakowski's avatar
      [devmgr] Delete dead code · 02f11fcb
      George Kulakowski authored
      Test: no functional change
      Change-Id: Ieb18428f06c7317d73af9c5093929fe9f989f4ee
      02f11fcb
    • David Stevens's avatar
      [display] Remove fidl padding workaround · 48e8ce26
      David Stevens authored
      Test: boot terminal product
      Change-Id: Iead9fc0fc67c1de116c300c3fda0f3894d0151cf
      48e8ce26
    • Braden Kell's avatar
      [sysinfo] Fix get_board_name string handling · a27c459a
      Braden Kell authored
      get_board_name should return the board name string
      without the null terminator, not the whole array.
      
      ZX-3079
      Test: tests pass
      
      Change-Id: If30b138779f3471cf1eca9c045e230345adca01a
      a27c459a
    • James Tucker's avatar
      [bin] export a working /bin in serial shells · 2e47e9f8
      James Tucker authored
      Originally we had planned to launch a fuchsia sys realm shell on serial,
      however this causes issues with tests and requires a more complex setup.
      Instead this solution should last until componentmgr takes charge.
      
      Test: cowsay works from serial shells.
      Bug: ZX-3043 #comment serial shells can run /bin trampolines.
      Change-Id: I219e1b585c23d9d15e7c4784755a4ee768301fc5
      2e47e9f8
    • Christopher Anderson's avatar
      [µpci] Fix build breakage caused by 7ebc5af3 · cc2dcfff
      Christopher Anderson authored
      The change made including libzx necessary for the driver to build.
      
      Test: build using ENABLE_USER_PCI=1
      
      Change-Id: I5d9b131ad4f1c724c115d4ebc9fd5cd950f46e99
      cc2dcfff
    • Mike Voydanoff's avatar
      [usb] Remove unnecessary bind.c files · c34fe533
      Mike Voydanoff authored
      TEST: builds, drivers load on Hikey960
      Change-Id: Id5b759665e395a2e67d328fd8e1c218c85914625
      c34fe533
    • John Bauman's avatar
      [mt8167][clk] Implement clock measurement · 9fedea66
      John Bauman authored
      Test: "clkctl measure" on mt8167
      
      Change-Id: Icdb3c1545efdd2d22f538eef8b09abc66a79e423
      9fedea66
    • Roland McGrath's avatar
      [kernel][ktl] fbl::move -> ktl::move · febb816c
      Roland McGrath authored
      <ktl/move.h> declares ktl::move for the kernel.
      fbl::move is entirely gone.
      
      Test: CQ
      
      Change-Id: I294f51a396d3196a7614ed311c56ba731065e607
      febb816c
    • David Gilhooley's avatar
      [perftest] Create README.md · a61211bc
      David Gilhooley authored
      Create a README with pointers to the style guide and how to
      run the perf tests
      
      Test: No tests, Documentation change only
      Change-Id: Ic6aa03fbe12811de6144f6642fc226b43b870550
      a61211bc
    • André Pang's avatar
      [fidl] Empty struct support for zircon (fidlc). · 30741deb
      André Pang authored
      For this to be a non-breaking change, this CL should land after the
      garnet <https://fuchsia-review.googlesource.com/c/garnet/+/228456> and
      topaz <https://fuchsia-review.googlesource.com/c/topaz/+/229308> CLs.
      
      Test: make -j 12 tools HOST_USE_ASAN=true && ./build-x64/host_tests/fidl-compiler-test
      
      FIDL-283 #progress
      
      Change-Id: I7cce14a107d99f909c7b1601940a58017e221d76
      30741deb
    • Robert Hahn's avatar
      [netreflector] Delete netreflector · 6100349f
      Robert Hahn authored
      We don't maintain and use this tool anymore.
      
      NET-2009 #comment
      
      Test=not required
      
      Change-Id: Ifa1eb2815e18db07712d06d6a2f825a07b204ee8
      6100349f
    • Scott Graham's avatar
      [syscalls][rights] Use AutoCall to RemoveHandles in channel code · ac3ff664
      Scott Graham authored
      Use RAII to remove handles on failure cases in channel, rather than
      explicitly calling up->RemoveHandles(...) inline for each failure case.
      
      ZX-2967 #comment [syscalls][rights] Use AutoCall to RemoveHandles in channel code
      
      Test: CQ, added channel_write_takes_all_handles
      Change-Id: I2b332b2a7a85cf7718fc2ed0734dd61bae7d44f0
      ac3ff664
    • Ruchira Ravoori's avatar
      [usb peripheral] Remove usage of req->node · f8171dd3
      Ruchira Ravoori authored
      With the recent changes to usb stack, each driver in the usb stack has its own list_node_t*
      so as to avoid stamping on other driver's lists. This changeset moves
      usb peripheral and test drivers away from using the public node. Once all
      references are removed the public list_node_t pointer will be removed.
      
      Test: USB CDC ethernet function and usb-peripheral-test on Hikey960.
      
      Change-Id: I4032e06304465ebd382b2205072038445965a685
      f8171dd3
    • Todd Eisenberger's avatar
      [dmctl] Delete dead code for ioctl_dmctl_watch_devmgr · c9c1e0cf
      Todd Eisenberger authored
      There are no callers of this in Fuchsia.  AFAICT it was never used?
      
      Test: Compiled and booted
      
      Change-Id: I8f404747aca58875daae606940587d03da4930dc
      c9c1e0cf
    • Ruchira Ravoori's avatar
      [usb-request]Remove more usage of req->node · 4405e23e
      Ruchira Ravoori authored
      With the recent changes to usb stack, each driver in the usb stack has its own list_node_t*
      so as to avoid stamping on other driver's lists. This changeset moves
      network drivers away from using the public node. Once all
      references are removed the public list_node_t pointer will be removed.
      
      Test: fx full-build. fx serve on pixelbook using a network adapter using
      usb-cdc-ecm works. fx netboot on Astro using a network adpater using
      ax88179 driver.
      
      Change-Id: Ic2ec43d7c78f75f7a7a2d0a4d8a44f05e52d0630
      4405e23e
    • George Kulakowski's avatar
      [fit] Remove empty empty.c · d12350f9
      George Kulakowski authored
      Test: no functional change
      Change-Id: I2504b6d7f22af755c2a708ec76bb2378492ee047
      d12350f9
    • John Bauman's avatar
      [mt8167][gpu] Expose MFG clocks to GPU driver · 97a82df2
      John Bauman authored
      Test: build
      
      Change-Id: I96f5ab61a28585192cfdc6beada5880abd67add1
      97a82df2
    • Gianfranco Valentino's avatar
      [cobalt-client] Provide host dummy impl. · e508992f
      Gianfranco Valentino authored
      Most of the users don't need host implementation,
      but those who do, would end up wrapping every statement
      in ifdef declarations.
      
      To avoid that provide a host compatible version, which essentially
      does nothing, uses a dummy implementation of Logger.
      
      User still need to wrap zircon specific code, but that is limited to
      the options for initialization. (zx::vmo, zx::duration).
      
      TEST=cobalt-client-tests
      
      Change-Id: Ic7defb847e721687ca4a2142af007ed0483f9c76
      e508992f
    • Abdulla Kamar's avatar
      [arm64][hypervisor] Prioritise SGIs and PPIs · 3110a809
      Abdulla Kamar authored
      Prioritise SGIs and PPIs over SPIs. This ensures that IPIs and timer
      interrupts are given preference over device interrupts.
      
      MAC-180
      
      Test: Ran "linux_guest" and "zircon_guest" on a VIM2.
      Change-Id: I13ac83db7dd27243100008e8b8cfd82a89477102
      3110a809
    • Jeff Belgum's avatar
      [dev][bcm-hci] Read BD Address from board in Broadcom Bluetooth driver · 2a1a8967
      Jeff Belgum authored
      Set address in controller to address set in the bootloader. Fallback to
      the default address stored in the controller.
      
      NET-1766 #done
      
      TEST=manual test: fx shell bt-hci-tool read-bdaddr. Confirm Google OUI
      
      Change-Id: I1e21c1dfeeede97ff0fd3d3749cfb830c78bd0df
      2a1a8967
    • George Kulakowski's avatar
      [devmgr][coordinator] Mostly clean up global state access · 8138d3cc
      George Kulakowski authored
      This puts all the functions using Coordinator state as member
      functions within that struct.
      
      The use of the g_coordinator global in the fidl bindings is not
      essential, but is a somewhat more invasive change than the rest of
      this patch, so it will be done later.
      
      Test: booted on hardware, suspend/shutdown work, ran with asan
      
      Change-Id: Ia5d0c65593cec0e7847641cb9a63a3a816c178f1
      8138d3cc
  3. Dec 03, 2018
    • Stephen Demos's avatar
      [blobfs] use FS_TRACE macros instead of (f|x)printf · 40a19639
      Stephen Demos authored
      this also redefines FS_TRACE_INFO as printf, so that there are no
      observable runtime changes.
      
      ZX-3058 #comment In Progress
      Test: CQ (no visible runtime changes)
      
      Change-Id: Icff324cc59fc47ee043331f6f2b56b802eb39f32
      40a19639
    • Ruchira Ravoori's avatar
      [bluetooth] Remove unnecessary logs · 784493ce
      Ruchira Ravoori authored
      Remove the debug lines that I added accidentally.
      
      Test:build in progress
      
      Change-Id: Ifad7453663f0e00174b33e5aa92187d5b4cf575d
      784493ce
    • Jocelyn Dang's avatar
      [usb-test-fwloader] FX3 initial setup part 1. · 5917d9cc
      Jocelyn Dang authored
      The default FX3 bootloader has a bug, so we want to load a second
      stage bootloader to the I2C EEPROM when setting up a new test device.
      
      The first step is to load the FX3 Flash Programmer image to RAM,
      which provides a vendor command to write to the EEPROM.
      
      ZX-3045 #comment
      
      TEST= copy flash programmer image
      usb-test-fwloader -e -i firmware.img -p flash.img
      lsusb and confirm PID 0x4720 shows up.
      
      Change-Id: Ide2f3e1a1c00411e237840ad9219a694a740b2ac
      5917d9cc
Loading